SSDF didn't stop cheating in mags CSS and ER

Rolf W. Tueschen

unread,

Dec 5, 1996, 3:00:00 AM12/5/96

to

Title: SSDF Rating Listing ---- is cheating

Preface

When I put some questions on SSDF after the Oct 23 update I got massive
attacks. Because I was German, because I didn’t know statistics and because
SSDF was amateur, without any money involved, and most important NOT an
academical institution.

I closed my thread in the hope that for the Dec update SSDF might find some
corrections. First of all in their presentation in magazines as CSS and
EUROPA ROCHADE.

But I was dissapointed.

First topic: Presentation of the results in magazines (CSS and ER)

A) Commentary by President or chairman Thoralf Karlsson

Unbelievable sloppy writing.
Look at this: *... Reb8 P90 leads say it and write it with 61 points
advantage before the second. As we’ve played yet 568 games the score of
2482 will not (!!) change very stongly in future.*

This is so called stupid garbage and cheating factor two. Why?
It’s very simple.
Even the most stupid user knows very well that the concurrent prgs of the
class of Reb8 are not yet on the market or couldn’t been tested. So if I
want to cheat a little bit I start with hundreds of statistical important
but from the view of the user UNIMPORTANT matches between one single NEW
prg against oldtimers. I receive a tremendosly high Elo on the base of
almost 600 games. (BTW *I* don’t critizise the enormous strength of Reb8 at
all!).
What does this mean if other prgs of the new generation could be tested in
future? And first of all prgs with difficult testing presuppositions (you
know what I mean ).
It’s almost to expect that THEN the score of Reb8 couldn’t be changed ...

For me a neutral tester should write such emphatical nonsense.

But Karlsson goes even further:
*To ameliorate the strength of computer chess (!!) so much is a big
progress.*
Interesting but sloppy remark. Let’s forget *computer chess* by Deep Blue
and so on for a while, folks.
*... and the men one (!) honours for are programmer Ed Schroder and Jeroen
Noomen ... We congratulate.*

But SSDF has a really difficult task. Look at this:
*The fact that always more prgs are written for WIN3.1 or 95 is a problem
for us in the SSDF.*
Well, SSDF always is proudly telling that they started their testings long
time ago. Was it 15 years ago?
What I saw at that time? No PC prgs at all. I saw Mephistos and Saiteks and
Fidelitys and Novags wooden pieces or silicon pieces. HOW did SSDF at those
days? Did they complain? Did the ask all programmers to follow this single
prg with the lifting feature? Never read of that one.

Look, there’s reality outside. And there are a multitude of different chess
playing *subjects sorry objects*.

And a testing crew has to deal with this reality. Period.
Little remark for those who don’t like *academic* questionings.
IF suddenly there’s a tool like the AUTOPLAYER, *I* would be very careful
in thinking about the consequences. Is it really progress to use like a
brain amputated this feature only to get hundreds of games over night ---
for some certain prgs??
What’s with those prgs which couldn’t be tested that way??
Once I tried to give SSDF the chance to understand this fundamental base of
all statistics, understood correctly.
It’s a simple rule: All objects tested must have equal chances to be
tested/ to have some results!
It’s simply cheating to procedure in other ways. The results are no more
balanced and statistically right.
B) A ranking of PORSCHEs and horses

SSDF didn’t change their 1., 2., 3. And so on ranking list. Therefore time
has come to show all ignorants the consequences of such nonsense.

If we look at the list from down under. What could we see?
We see on *place* 40 *Montreux* with a winner performance of
74%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
What Elo did they give him as partners? Elo 2008. Ok. I’m not sure, but
this fabulous prg deserved a special possibility to fight against the
highest PC prgs, no?
74% is the record of the list. But if you don’t give him high Elo partners
he couldn’t come in higher.

And Montreux is an old 14 Mhz! MMIV was 5 Mhz. WHAT does this tell us about
*computer chess strength*, Thoralf?

Let’s take a look at the place 15. There we find a fabulous R30 v. 2.5 with
a winner performance of 68%! And again SSDF gave him ONLY Elo 2226
adversairys. 68% is the second best score overall! Why not testing him
*out* with 500 games against the top?? Ahh, I forgot the Autofunction. And
so on. And perhaps only 2 testers had the R30. I understand ...

Look, I gave a speed test for the famous mag *Mobiles and horses* till the
late sixties. And I did the same hidden tricks. I always went in the
mountains first of all. Then I searched for real stony pists. And then I
got a PORSCHE and my favorite horse. And you must not forget I informed the
sheperd to come over with his cows. So there was a bit more traffic on the
*road*. Just like in real formule one racing.
Now I shot my pistol for the start. The hoarse first took his time to study
the country side whereas PORSCHE slipped through the slim curves like a
real champ. BUT then after the third obstacle (you know what I mean) the
hoarse always passed easily my PORSCHE. And so that I could write in my
testing results that PORSCHE was only slighty faster than horses. That
were the days all Germans bought Volkswagen like fools. The knew they could
save a lot of money -- completely believing my test results.

Unfortunately I was kept from any further testings by the somewhat old
fashioned german courts. They thought I did cheat on my results. But I
still think they completely overlooked my deep roots which lay certainly in
the higher mountains. *I* couldn’t change these conditions.

Therefore today each German has a at least one PORSCHE in his garage ...

Seriously it’s a big cheat to place Montreux on 40 and R30 on 15. Those
machines had earned a much better placing in a different ranking. They are
NOT in the same class of the P90 hybrides. But their chess wisdom might be
the same!

Therefore commentary of the chairman (see above) is so much wrong.

C) Errorsignificance small talk

It’s another ridiculous sideeffect of this strange testings that afterwards
the singular results are taken serious for the debate of DOS vs. WIN. 5%
means 4-5 points. You got it?
That’s the problem with maths. When I once started my calculations at
school none ever could have stopped my behaviour. If I once made the
translation of the task in mathematical proceduring – Arnold Schwarzenegger
wouldn’t have stopped me.
But nevertheless it happened that I had some red marks from the teacher on
my paper. Not in my calculs but right in my first steps! Consequently I
tried to work especially on the very beginning phase of those tasks. And I
continued study in university later on. Just for the record.

Consequences

I wouldn’t critizise the swedish testers at the basis. They do their job as
best they can. But what I oppose here is
the incredible confusion in the SSDF president suite.

The last sentence of Thoralf’s commentary tells a language for itself.
*We follow discussion (on the net) with great attention, but we need some
more time to find decisions, (NOW IT COMES) if there are justified reasons
to change our test work.*

These lines try to give the impression that SSDF staff always thought with
a lot of care about procedurings. Always. And because of that they simple
couldn’t change s.th. in their work quasi a tempo. This would be really
unsober.

But reality is another one.

SSDF started with good ideas long time ago. At that time all testing had to
be done by hand. The games were not in the 500 hundred region so quickly.
And the existing prgs could be matched against each other without effort.
But then after almost 200 prgs on the market (historically) the whole
testing got out of control.
Which prg what machine. How many games between certain prgs. And so on.

Another strange side. A programmer throws his new version on the market 4
months before his concurrents.
And he is in close contact with the SSDF staff. He waits for instant
results. Because 200 games on his own page could be regarded as not
*heutral*. And after some weeks the deliver him 500 games. No 1 place! Best
prg of all time ... Real progress in the history of computer chess ...

No.

A real serious non cheating staff would have said, look Eddie, that’s nice,
but we’ll wait till chrismas. Then we could match Reb8 against Genius5,
Mchess6, HIARCS5 and so on. Because we are really independent testers.
What would it help you if you could tell the world you’ve beaten all those
oldtimers? Wer zuletzt lacht, lacht am besten. No?

Rolf Tueschen 5.12.96

Herbert Groot Jebbink

unread,

Dec 10, 1996, 3:00:00 AM12/10/96

to

On 5 Dec 1996 17:18:11 GMT, TUESCHEN.MEDIZ...@t-online.de
(Rolf W. Tueschen) wrote:

>Title: SSDF Rating Listing ---- is cheating

I did never think that I would use the Agent Kill-list function for
this newsgroup, but YOU are in it now!

Greetings, hgj

mclane

unread,

Dec 10, 1996, 3:00:00 AM12/10/96

to

TUESCHEN.MEDIZ...@t-online.de (Rolf W. Tueschen) wrote:

>Preface

>But I was dissapointed.

>C) Errorsignificance small talk

>No.

>Rolf Tueschen 5.12.96

Rebel8 will win over Genius and Co because it is stronger. Or ?
Although it has no special XYZ-book it is better.
But we will see.

Rolf W. Tueschen

unread,

Dec 10, 1996, 3:00:00 AM12/10/96

to

h...@xs4all.nl (Herbert Groot Jebbink) wrote:

>On 5 Dec 1996 17:18:11 GMT, TUESCHEN.MEDIZ...@t-online.de
>(Rolf W. Tueschen) wrote:

>>Title: SSDF Rating Listing ---- is cheating

>I did never think that I would use the Agent Kill-list function for

>this newsgroup, but YOU are in it now!

>Greetings, hgj

----------------------------------

Hello there in the Netherlands.

It's not easy to please everyone if you attack sort of historical monument.

But if my arguments were that stupid/wrong/horrible someone already had all
the time to correct me. If it were the case I would surely apologize.

Just for the record because you are the second dutch user besides Marcel v.
Kervinck who wrote private mail long time ago with almost the same words, I
want to explain that my use of *cheating* has the notion *mogeln* (perhaps
in your language mogelen, I don't know). If I was right with my points of
critics would you still conclude that the notion was wrong?

Greetings

Rolf Tueschen

-----*Man is unable not to know what he knows.* Leibowitz----

Peter Fendrich

unread,

Dec 11, 1996, 3:00:00 AM12/11/96

to

????
Get cured.

--
J-P Fendrich

a bean

unread,

Dec 11, 1996, 3:00:00 AM12/11/96

to

Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in
article <58704j$o...@news00.btx.dtag.de>...

| Title: SSDF Rating Listing ---- is cheating
|

This is a most serious accusation,

If SSDF does not post a SERIOUS response then what are we to conclude?

Its time to ask the following question -

"Standardized Computer Chess Testing - Is it possible?"
--
Catchulater Co.
{:o) /|\|/|\|/\ (o:}
{:o) ( @ @) (o:}
{:o) ( _> ) (o:}
-------o00o---------o00o-------
*@*=*@*=*@*=*@*=*@*
[keep seeking the Truth]
spr...@discover-net.net

Robert Hyatt

unread,

Dec 11, 1996, 3:00:00 AM12/11/96

to

a bean (spr...@discover-net.net) wrote:
: Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in
: article <58704j$o...@news00.btx.dtag.de>...
: | Title: SSDF Rating Listing ---- is cheating
: |
:
: This is a most serious accusation,

:
: If SSDF does not post a SERIOUS response then what are we to conclude?
:
: Its time to ask the following question -
:
: "Standardized Computer Chess Testing - Is it possible?"

It's totally impossible, because anyone can claim a bias. For example,
both programs get 64mb. This helps the fast searcher more than a slow
searcher. Unfair.

both programs get 8mb. This hurts the fast searcher more than the slow
searcher. Unfair.

both programs get equal fast machines. This helps the program that gets
the most benefit from another ply more than the one that doesn't. Unfair
to one. Ditto for both on a slow machine.

In short, it's impossible to be totally fair. The SSDF guys simply do
what they can, and post results with the program/machine combination clearly
specified, so at least the results are understandable...

Bob

BTW I don't consider any of the above to be a problem. Just that a couple
have pointed the problems out...

Hansjoerg Zech

unread,

Dec 11, 1996, 3:00:00 AM12/11/96

to

mclane wrote:
>
> Rebel8 will win over Genius and Co because it is stronger. Or ?
> Although it has no special XYZ-book it is better.
> But we will see.

Rebel 8 has a negative score (9.5-10.5) against Genius 4 in the SSDF
list ?!

Hansjoerg
--
Hansjoerg Zech
Internet: zech...@sni.de
X.400: c=DE;a=DBP;p=SCN;o=SNI;ou1=PDB1;ou2=S1;s=ZECH;g=HANSJOERG
SNI ASW C8, Heinz Nixdorf Ring 1, 33106 Paderborn
Phone: +49 5251 8 14926 Fax: +49 5251 8 14902

Rolf W. Tueschen

unread,

Dec 11, 1996, 3:00:00 AM12/11/96

to

mcl...@prima.ruhr.de (mclane) wrote:

>>Rolf Tueschen 5.12.96

------------------------------------------------------------------------------------------------------------------------
For me Genius, MChess, Hiarcs and Rebel are the top prgs. Nimzo seems to
habe entered this class. I don't know enough of all that. It's simply
belief derived from the past developements.

Ahh, I forgot Fritz. But in today's SSDF he seems to have difficulties.

Now let's make a deal. :)

You send me a lot of games you played with computers if Rebel will be
topped by one of those. I think you had a lot of important games, Please
send me some for chrismas, yes?

Do you know why I'm so sure about things that would come?

Well, I simply follow Ed's feelings. Last days he developped some negative
thoughts about all what would come. And Ed surely knows why he's a little
bit afraid of his concurrent partners. :)

Either Mchess or Genius will be on the toplist in February 97. I'm almost
sure. :)

Rolf Tueschen

Dave Gomboc

unread,

Dec 12, 1996, 3:00:00 AM12/12/96

to

In article <58704j$o...@news00.btx.dtag.de>,
Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:
#Title: SSDF Rating Listing ---- is cheating
#
#
#Preface
#
#When I put some questions on SSDF after the Oct 23 update I got massive
#attacks. Because I was German, because I didn’t know statistics and because
#SSDF was amateur, without any money involved, and most important NOT an
#academical institution.

As if it had something to do with you being German. More like it was
because the contents of posts of yours which are intellgible span the
range from vapid speculation to sinister disingenuity, crossing abject
stupidity and sheer bollocks en route. How many months do you have to
be a regular poster on the net before you're NOT a "newbie"? (Hint:
You've posted 188 messages to the rec.games.chess.* hierarchy since
September 1st, and probably quite a few more before then.)

#
#I closed my thread in the hope that for the Dec update SSDF might find some
#corrections. First of all in their presentation in magazines as CSS and
#EUROPA ROCHADE.
#
#But I was dissapointed.
#
#
#First topic: Presentation of the results in magazines (CSS and ER)
#
#A) Commentary by President or chairman Thoralf Karlsson
#
#Unbelievable sloppy writing.
#Look at this: *... Reb8 P90 leads say it and write it with 61 points
#advantage before the second. As we’ve played yet 568 games the score of
#2482 will not (!!) change very stongly in future.*
#
#This is so called stupid garbage and cheating factor two. Why?
#It’s very simple.
#Even the most stupid user knows very well that the concurrent prgs of the
#class of Reb8 are not yet on the market or couldn’t been tested. So if I
#want to cheat a little bit I start with hundreds of statistical important
#but from the view of the user UNIMPORTANT matches between one single NEW
#prg against oldtimers. I receive a tremendosly high Elo on the base of
#almost 600 games. (BTW *I* don’t critizise the enormous strength of Reb8 at
#all!).
#What does this mean if other prgs of the new generation could be tested in
#future? And first of all prgs with difficult testing presuppositions (you
#know what I mean ).
#It’s almost to expect that THEN the score of Reb8 couldn’t be changed ...
#
#For me a neutral tester should write such emphatical nonsense.
#

There's no cheating factor. He's simply stating the obvious (well,
apparently not obvious, since it needs to be explained to you): the
more games that are played in the SSDF system, the less that future
games will affect measurements are taken. Rebel 8.0's rating will
continue to shift with all the rest, as games continue to be played.
(Exercise: demonstrate to yourself that the rating of a program on the
SSDF list can move between lists despite not having any games played
between a list and the list previous to it.)

#But Karlsson goes even further:
#*To ameliorate the strength of computer chess (!!) so much is a big
#progress.*
#Interesting but sloppy remark. Let’s forget *computer chess* by Deep Blue
#and so on for a while, folks.
#*... and the men one (!) honours for are programmer Ed Schroder and Jeroen
#Noomen ... We congratulate.*

60 points ahead of the nearest competitor *on the same hardware* was
big progress at a time when all the top program were within 10 points
of each other. (It's only 20 points ahead now, with MChess 6.0 on the
new list.) There's always "congratulations" from the chairman when
there's a new number one. You're trying to insinuate bias then there
isn't any.

#
#But SSDF has a really difficult task. Look at this:
#*The fact that always more prgs are written for WIN3.1 or 95 is a problem
#for us in the SSDF.*
#Well, SSDF always is proudly telling that they started their testings long
#time ago. Was it 15 years ago?
#What I saw at that time? No PC prgs at all. I saw Mephistos and Saiteks and
#Fidelitys and Novags wooden pieces or silicon pieces. HOW did SSDF at those
#days? Did they complain? Did the ask all programmers to follow this single
#prg with the lifting feature? Never read of that one.

Once technology eases people's lives, they seldom want to "go back to
the old way". I don't have the fortune to own a snow blower, but you
can bet that once I had it the snow shovel would stay in the garage.
Keeping in mind (the obvious idea) that it is easier to autotest than
to test manually, do you actually have a point to make here? All I
see is some moron taking shots at people who provide what amounts to a
public service because he doesn't have anything better to do.

#
#Look, there’s reality outside. And there are a multitude of different chess
#playing *subjects sorry objects*.
#
#And a testing crew has to deal with this reality. Period.
#Little remark for those who don’t like *academic* questionings.
#IF suddenly there’s a tool like the AUTOPLAYER, *I* would be very careful
#in thinking about the consequences. Is it really progress to use like a
#brain amputated this feature only to get hundreds of games over night ---
#for some certain prgs??
#What’s with those prgs which couldn’t be tested that way??
#Once I tried to give SSDF the chance to understand this fundamental base of
#all statistics, understood correctly.
#It’s a simple rule: All objects tested must have equal chances to be
#tested/ to have some results!
#It’s simply cheating to procedure in other ways. The results are no more
#balanced and statistically right.

They post margins of error which reflect the number of games played by
each hardware and software combination. Why that is not good enough
for you is a mystery to me. Your "fundamental base of statistics" as
written is some of that bollocks I mentioned earlier.

#B) A ranking of PORSCHEs and horses
#
#SSDF didn’t change their 1., 2., 3. And so on ranking list. Therefore time
#has come to show all ignorants the consequences of such nonsense.
#
#If we look at the list from down under. What could we see?
#We see on *place* 40 *Montreux* with a winner performance of
#74%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
#What Elo did they give him as partners? Elo 2008. Ok. I’m not sure, but
#this fabulous prg deserved a special possibility to fight against the
#highest PC prgs, no?
#74% is the record of the list. But if you don’t give him high Elo partners
#he couldn’t come in higher.
#
#And Montreux is an old 14 Mhz! MMIV was 5 Mhz. WHAT does this tell us about
#*computer chess strength*, Thoralf?
#
#Let’s take a look at the place 15. There we find a fabulous R30 v. 2.5 with
#a winner performance of 68%! And again SSDF gave him ONLY Elo 2226
#adversairys. 68% is the second best score overall! Why not testing him
#*out* with 500 games against the top?? Ahh, I forgot the Autofunction. And
#so on. And perhaps only 2 testers had the R30. I understand ...

Why don't you buy a couple of these machines and test some games.
You'll find that it may or may not matter as much as 10 points. In
fact, why don't you just go and add the 10 points right now, just to
make yourself happy? Do you really know there's such a big difference
between 68% against average rating A and 64% against average rating B
which is close to A? Back your argument with arithmetic. Skip the
excess verbiage, it's tedious to read through.

#
#Look, I gave a speed test for the famous mag *Mobiles and horses* till the
#late sixties. And I did the same hidden tricks. I always went in the
#mountains first of all. Then I searched for real stony pists. And then I
#got a PORSCHE and my favorite horse. And you must not forget I informed the
#sheperd to come over with his cows. So there was a bit more traffic on the
#*road*. Just like in real formule one racing.
#Now I shot my pistol for the start. The hoarse first took his time to study
#the country side whereas PORSCHE slipped through the slim curves like a
#real champ. BUT then after the third obstacle (you know what I mean) the
#hoarse always passed easily my PORSCHE. And so that I could write in my
#testing results that PORSCHE was only slighty faster than horses. That
#were the days all Germans bought Volkswagen like fools. The knew they could
#save a lot of money -- completely believing my test results.
#
#Unfortunately I was kept from any further testings by the somewhat old
#fashioned german courts. They thought I did cheat on my results. But I
#still think they completely overlooked my deep roots which lay certainly in
#the higher mountains. *I* couldn’t change these conditions.
#
#Therefore today each German has a at least one PORSCHE in his garage ...
#

Does this have something to do with chess? Or the SSDF?

#Seriously it’s a big cheat to place Montreux on 40 and R30 on 15. Those
#machines had earned a much better placing in a different ranking. They are
#NOT in the same class of the P90 hybrides. But their chess wisdom might be
#the same!
#
#Therefore commentary of the chairman (see above) is so much wrong.

Oh, it was a stupid analogy, with an attempt to be comic near the end.
I see. Ha ha. Real funny, that. It would be nice if the win
percentage was closer to 50%, but statistical aberrations such as
these are bound to happen once in a while. Indeed, I would be
suspicous if they did not.

#C) Errorsignificance small talk
#
#It’s another ridiculous sideeffect of this strange testings that afterwards
#the singular results are taken serious for the debate of DOS vs. WIN. 5%
#means 4-5 points. You got it?
#That’s the problem with maths. When I once started my calculations at
#school none ever could have stopped my behaviour. If I once made the
#translation of the task in mathematical proceduring – Arnold Schwarzenegger
#wouldn’t have stopped me.
#But nevertheless it happened that I had some red marks from the teacher on
#my paper. Not in my calculs but right in my first steps! Consequently I
#tried to work especially on the very beginning phase of those tasks. And I
#continued study in university later on. Just for the record.

You may or may not be trying to say something of relevance in the
above paragraph. Don't bother clarifying if you don't feel like it.
In fact, you don't even have to bother posting, I won't miss you.

#
#Consequences
#
#I wouldn’t critizise the swedish testers at the basis. They do their job as
#best they can. But what I oppose here is
#the incredible confusion in the SSDF president suite.
#
#The last sentence of Thoralf’s commentary tells a language for itself.
#*We follow discussion (on the net) with great attention, but we need some
#more time to find decisions, (NOW IT COMES) if there are justified reasons
#to change our test work.*
#
#These lines try to give the impression that SSDF staff always thought with
#a lot of care about procedurings. Always. And because of that they simple
#couldn’t change s.th. in their work quasi a tempo. This would be really
#unsober.
#
#But reality is another one.
#
#SSDF started with good ideas long time ago. At that time all testing had to
#be done by hand. The games were not in the 500 hundred region so quickly.
#And the existing prgs could be matched against each other without effort.
#But then after almost 200 prgs on the market (historically) the whole
#testing got out of control.
#Which prg what machine. How many games between certain prgs. And so on.
#
#Another strange side. A programmer throws his new version on the market 4
#months before his concurrents.
#And he is in close contact with the SSDF staff. He waits for instant
#results. Because 200 games on his own page could be regarded as not
#*heutral*. And after some weeks the deliver him 500 games. No 1 place! Best
#prg of all time ... Real progress in the history of computer chess ...
#
#No.
#
#A real serious non cheating staff would have said, look Eddie, that’s nice,
#but we’ll wait till chrismas. Then we could match Reb8 against Genius5,
#Mchess6, HIARCS5 and so on. Because we are really independent testers.
#What would it help you if you could tell the world you’ve beaten all those
#oldtimers? Wer zuletzt lacht, lacht am besten. No?
#
#Rolf Tueschen 5.12.96
#

SSDF tests new versions when they get them. By March's list, you'll
be able to see how Rebel 8.0, M-Chess 6.0, Chess Genius 5.0, Hiarcs
5.0, and any other new programs do, with a lot of interplay between
them in particular. Why would you want to delay publication of
results of games? If a programmer beats competitors to market by a
couple months, why shouldn't his product appear sooner? Perhaps
another programmer decides to delay for a bit and continue to improve
his software, eventually yielding a higher-rated product than it would
have been had he published it sooner. Why should you not begin
testing with the released product? Consumers are especially
interested in the latest software available. If you want to see how
the top programs do against only each other it is very easy to pull
this information out of the SSDF list, which includes a result
breakdown by opponent. Really independent testers test what they've
got when they get it.

Dave Gomboc
drgo...@a.stu.athabascau.ca

brucemo

unread,

Dec 12, 1996, 3:00:00 AM12/12/96

to

Dave Gomboc wrote:

> There's no cheating factor. He's simply stating the obvious (well,
> apparently not obvious, since it needs to be explained to you): the
> more games that are played in the SSDF system, the less that future
> games will affect measurements are taken. Rebel 8.0's rating will
> continue to shift with all the rest, as games continue to be played.
> (Exercise: demonstrate to yourself that the rating of a program on the
> SSDF list can move between lists despite not having any games played
> between a list and the list previous to it.)

This depends a lot on how they do the rating system.

In the rating system used on ICC, and I believe that in the rating system
used by FIDE and by the USCF, if you have a rating of 2000 and have played
a thousand rated games, and you lose five games in a row to some weaker
players, your rating will drop, a lot. If you have a rating of 2000 and
have played fifty rated games, and you lose five games in a row to these
same players, your rating will drop by the same amount.

Once your rating is established, it doesn't matter how many games you have
played, you stand to gain or lose the same amount from any given game.

In the rating system used on FICS, well, I'm not sure how it works
exactly, but it does have something to do with how you've done recently.
If you've played hundreds of games in the last few days your "RD" will go
down, which means that the amount you gain or lose from a single game will
decrease.

So if the SSDF is using the first system I mentioned, this statement may
well be false.

Here's something else to think about, assuming we're talking about a
rating list created the same way ICC ratings work.

Three programs, A, B, and C.

A beats B 55% of the time.
A beats C 30% of the time.

In this situation, C appears to be best, because it crushes A. B appears
to be the worst, because A scores well against it. But if I add this
line:

B beats C 55% of the time.

How do you rate this pool? The situation is unstable, so if your strategy
is to play a series of long matches, and my program is program C, I'd very
much like the last match played before the ratings are published to be a
match with program A, and if possible (assuming there are other programs
than A, B, and C), I would like program A to have gotten nice and fat from
playing a big match with program B right before it encounters me.

I very much would like to avoid playing a match with program B right after
this match (and before the rating list is published), because I will have
gotten fat off of A, who has gotten fat off of program B, and program B
will be very skinny. So I will lose points in this match, and as a
consequence I'd like to play it right after the list is published, so I
will have time to regain my points before the next list comes out. Using
the ICC formula, games you played last month, assuming you play a lot,
have almost no affect upon your current rating.

Perhaps in this case if you have a whole bunch of games played in matches,
it would make sense to rate the games in random order, as if they weren't
played in large blocks like this. Maybe the problem with this idea is
that someone could manipulate the random order, possibly even to the point
where they could recalc the spreadsheet several times, until they
generated the ranking list they most preferred :-)

Is the SSDF list vulnerable to any of this, or do you guys use a
completely different rating system or something like that?

bruce

a bean

unread,

Dec 12, 1996, 3:00:00 AM12/12/96

to

Robert Hyatt <hy...@crafty.cis.uab.edu> wrote in article
<58mba6$l...@juniper.cis.uab.edu>...

It is true that Chess Aptitude cannot be reliably tested given the current
methodologies for the tests.

Change the method and the impossible will become possible!

Howard Exner <hex...@dlcwest.com> wrote in article
<01bbd8b4$e5d3da80$3725...@dlcwest.dlcwest.com>...
|
|
| a bean <spr...@discover-net.net> wrote in article
| <01bbd810$35403fa0$66cb...@discover-net.net.discover-net.net>...
| > Can this be the start of a standardized CAT (Computer Aptitude
| Test)!
|
| There are a number of test suites available that attempt
| to correlate results to a rating(Aptitude).Louguet-2 and
| the Kaufman one hour tests are two examples. Others exist
| also. I am hesitant in tagging a rating to such suites.Some of my
| reasons are:
|
| 1. The samples are too small. Chess with its vast array of
| possibilities and variants to themes would make it difficult
| to create a comprehensive test.
|

It may be difficult, but certainly not impossible.
ELO or USCF ratings start becoming fairly accurate in as few as 20 games.
If a CAT contained between 100 and 1000 positions it would easily produce
statistically reliable results.
This test by the way does NOT have to produce a 100% correlation to a OTB
rating.

An important point that needs to be clarified is that:
The primary purpose of a CAT test would be to determine the computers (or
humans) RELATIVE chess playing ability in different types of positions and
time controls - INDEPENDENT of the opening book or strength/weaknesses of
the various OTB opponents.

Another plus is that all the testing would be done with the same positions.

| 2. Solving certain tactical positions quickly may be a good
| indicator of speed chess ability but may not translate
| well into 40/2 games.

Exactly why we need this - the purpose of a CAT test is to determine the
relative playing strengths for various time controls and positional types.

|
| 3. Some positions solved correctly by computers, usually
| positional ones, may not result in the correct follow through.
|

I agree with this 100% - Part of the solution would be to insist that the
computer give an evaluation within a correct range in addition to the
correct 1st move in order to get credit for the answer. I will concede
though that some positions are so tactically (or positionally) complex that
it would require a whole tree with many (!) nodes in order to convincingly
demonstrate that the computer "understands" the solution. These positions
though are in effect -testing for many solutions simultaneously. It is
important to keep the testing as simple as possible in order to better
quantify the results.

Insisting on test positions having clear solutions will take care of the
other part of the problem.

| 4. Positional tests are not as objective as tactical ones. A
| right move may be debatable.

If the right move is debatable then this position has no place in a test.

|
| 5. The different software programs all seem to have
| particular strengths and weaknesses. It would be easy
| to compose a set of tests which could tap into various
| strengths thus having program A emerging on top for
| one suite while program B would rate higher on a different
| suite.

The purpose of the test would not be to unfairly prop up one program over
another.
The test would in fact show where each of the programs (or humans) is
truly strong or truly weak.

|
| 6. Opinions of other rgcc readers?

When did opinion become more important than the truth?
The testing will be developed whether some want it or not.
Do the majority of high-school students support the SAT?

|
| That been said testing computers on different positions has
| advantages. It may lead to programmers making changes to improve
| there engines or in my case trying to improve my own abilities with
| the
| computer software as the sparring partner. Do programmers rely on
| various test positions to improve there engines?

Chess programs today are relying more and more on Opening book and Endgame
tablebase in order to improve their ratings. If the rating is improved in
this way is it really a true measure of the programs ability to "play
chess" or is it a measure of something else?

|
| If you are interested
| in testing software there are also numerous books out there. One I'm
| having fun with is called "The Big Book Of Combinations" by Eric
| Schiller. It
| ranks problems by degree of difficulty as solved by a version
| of Zarkov on a P-90. Problems in this book with a 9 or 10
| rating seem quite challenging for computers. They are tactical
| problems by the way.

Some of those positions could very well be used for a CAT.

It would be important to record the results of testing as such:

Name: SuperChessX
Test Name: CAT 1A
Pos. Time
1 165
2 xxxx
3 32
.
.
.
1000 617

Summary of results:
sec per move - 5 12 30 60 180 900 3600
%correct 7 16 28 42 57 65 72

This type of test result would allow for a very effective way to compare
and evaluate different chess programs for different time controls and
positions.

Rolf W. Tueschen

unread,

Dec 12, 1996, 3:00:00 AM12/12/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>In article <58704j$o...@news00.btx.dtag.de>,
>Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:
>#Title: SSDF Rating Listing ---- is cheating

I snipped my post because of ease of reading.
---------------------

>As if it had something to do with you being German. More like it was
>because the contents of posts of yours which are intellgible span the
>range from vapid speculation to sinister disingenuity, crossing abject
>stupidity and sheer bollocks en route. How many months do you have to
>be a regular poster on the net before you're NOT a "newbie"? (Hint:
>You've posted 188 messages to the rec.games.chess.* hierarchy since
>September 1st, and probably quite a few more before then.)

Yes, I'm in that business from 1955 on. And wrote zillions of posts. :)

Dave, I stopped claiming newbie status some weeks ago. But I'm still very
new here compared to others, no?

I'd like to answer your commentary just because you write below some
interesting stuff. Interesting because I normally write speculation,
stupidity, bollocks. If I understood you right?

Let me tell you one single argument. Normally you shouldn't begin a debate
with such heavy verdicts simply because you made yourself sort of fool to
discuss thereafter, no? So I'll give my answer but afterwards you should
decide to search for better correspondence. If I rattled your equilibrance
that much with my statment about SSDF, that you lost some needed abilities
to read carefully, I apologize of having done this. It was not my
intention.

Right at the beginning you jump in and your nose tipped the dust. Why that?
The *German* thing WAS mentioned. I always read about german anti SSDF
gang. Didn't you realize this?
---------------------

>There's no cheating factor. He's simply stating the obvious (well,
>apparently not obvious, since it needs to be explained to you): the
>more games that are played in the SSDF system, the less that future
>games will affect measurements are taken. Rebel 8.0's rating will
>continue to shift with all the rest, as games continue to be played.
>(Exercise: demonstrate to yourself that the rating of a program on the
>SSDF list can move between lists despite not having any games played
>between a list and the list previous to it.)

First of all it would be very important if you would be pleased to define
your notion *games*. That's important for my statement. But as you are so
well educated I could freely follow that you understood the games as
equally comparable from which *match* (important second notion) they might
ever had come. Am I right? Well then my commentary is again that that is
the first serious error in SSDF listing/statistics. And the second
therefore is that the matches are in reality of SSDF testings NOT defined.
You see 1-0 matches and surely a lot of 20 games matches.

Second, I want to point out that it is cheating if you test rather
unimportent matches in hundreds and the very interesting ones are almost
hidden in this mass. Look e.g. at the result of Genius4 vs Rebel8. These
two are concurrents. Not Comet or whatelse. Sorry I play with Comet, I like
it, but it is not in the same class.
--------------------

>60 points ahead of the nearest competitor *on the same hardware* was
>big progress at a time when all the top program were within 10 points
>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
>new list.) There's always "congratulations" from the chairman when
>there's a new number one. You're trying to insinuate bias then there
>isn't any.

Nope, friendly Dave. Don't fuzzle my words to the invisible. You're not
Spock, no?
It's not the 60 points. I agree with you if each winner gets
congretulations. But I will prove this. I assume that you're wrong. They
didn't applaude each time with such emphazis.
But my point was another one.
Karlsson wrote about *chess computer wisdom*. And that is at this very
moment without tests between the real concurrents simple cheating. Period.
----------------------

>Once technology eases people's lives, they seldom want to "go back to
>the old way". I don't have the fortune to own a snow blower, but you
>can bet that once I had it the snow shovel would stay in the garage.
>Keeping in mind (the obvious idea) that it is easier to autotest than
>to test manually, do you actually have a point to make here? All I
>see is some moron taking shots at people who provide what amounts to a
>public service because he doesn't have anything better to do.

Hey, what's moron, my dicionary iseems so empty ...
I thought we had passed this first claim of the amateur tester in swedish
woods who was insulted by talking about his master instrument - the
autoplayer. You seem completely overlook today's questioning of details of
that autoplayer. Read Ed. Read Bert. Period.
-----------------------

>They post margins of error which reflect the number of games played by
>each hardware and software combination. Why that is not good enough
>for you is a mystery to me. Your "fundamental base of statistics" as
>written is some of that bollocks I mentioned earlier.

Yes, as concerning your *games*, read above.
------------------------

>Why don't you buy a couple of these machines and test some games.
>You'll find that it may or may not matter as much as 10 points. In
>fact, why don't you just go and add the 10 points right now, just to
>make yourself happy? Do you really know there's such a big difference
>between 68% against average rating A and 64% against average rating B
>which is close to A? Back your argument with arithmetic. Skip the
>excess verbiage, it's tedious to read through.

What do you like most, me buying some machines and test or going back to
school or ameliorate autoplayer or anwering you at the moment?
-----------------------

>Does this have something to do with chess? Or the SSDF?

You refered my PORSCHE dingdong? Yes, it has something to do with testing
procedures. Didn't you read that as such? Don't you like horses?
No serious, you're trying to astonish me.
-----------------------

>Oh, it was a stupid analogy, with an attempt to be comic near the end.
>I see. Ha ha. Real funny, that. It would be nice if the win
>percentage was closer to 50%, but statistical aberrations such as
>these are bound to happen once in a while. Indeed, I would be
>suspicous if they did not.

If I didn't know I write to Dave I would suppose MB is on the phone. :)
Or did you translate some passages?
But serious. You didn't get the point again.
If I would detect such results I would continue testing those special
machines. Didn't you?
----------------------

>You may or may not be trying to say something of relevance in the
>above paragraph. Don't bother clarifying if you don't feel like it.
>In fact, you don't even have to bother posting, I won't miss you.

Dave, but it's so simple, just kill my posts. *I* as new -- sorry Dave,
really -- I never do this because I like all writings of people from
everywhere. :)
----------------------

>SSDF tests new versions when they get them. By March's list, you'll
>be able to see how Rebel 8.0, M-Chess 6.0, Chess Genius 5.0, Hiarcs
>5.0, and any other new programs do, with a lot of interplay between
>them in particular. Why would you want to delay publication of
>results of games? If a programmer beats competitors to market by a
>couple months, why shouldn't his product appear sooner? Perhaps
>another programmer decides to delay for a bit and continue to improve
>his software, eventually yielding a higher-rated product than it would
>have been had he published it sooner. Why should you not begin
>testing with the released product? Consumers are especially
>interested in the latest software available. If you want to see how
>the top programs do against only each other it is very easy to pull
>this information out of the SSDF list, which includes a result
>breakdown by opponent. Really independent testers test what they've
>got when they get it.

Dave, why, when, would, if. You're really on the slope by now.
Surely they are alowed to do this. They could do whatever they want.
Problem remains, if statistically that would be sound. And my point was
that it was NOT. But we're at the end of our debate. And you still didn't
mention notion games, notion match, how many games, what pairings and so
on.
But that wasn't your favorite intention I see. But *bollock* notion as
verdict is written much more easily. I understand. You did your job. But
you've failed. Perhaps we'd meet in another thread. As I announced I
wouldn't answer you again if you continue in that style, ok? Couldn't we
change that? I'm still not convinced that I couldn't learn something of
you.
-----------------

>Dave Gomboc
>drgo...@a.stu.athabascau.ca
-------------------------------------------------------------
-------------------------------------------------------------

mclane

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

hy...@crafty.cis.uab.edu (Robert Hyatt) wrote:

>a bean (spr...@discover-net.net) wrote:
>: Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in
>: article <58704j$o...@news00.btx.dtag.de>...
>: | Title: SSDF Rating Listing ---- is cheating
>: |
>:
>: This is a most serious accusation,
>:
>: If SSDF does not post a SERIOUS response then what are we to conclude?
>:
>: Its time to ask the following question -
>:
>: "Standardized Computer Chess Testing - Is it possible?"

>It's totally impossible, because anyone can claim a bias. For example,
>both programs get 64mb. This helps the fast searcher more than a slow
>searcher. Unfair.

>both programs get 8mb. This hurts the fast searcher more than the slow
>searcher. Unfair.

>both programs get equal fast machines. This helps the program that gets
>the most benefit from another ply more than the one that doesn't. Unfair
>to one. Ditto for both on a slow machine.

>In short, it's impossible to be totally fair. The SSDF guys simply do
>what they can, and post results with the program/machine combination clearly
>specified, so at least the results are understandable...

Oh - your words could be mine! All above points are absolutely right.
You got it!

>Bob

>BTW I don't consider any of the above to be a problem. Just that a couple
>have pointed the problems out...

I also do not see any problems in the above points.
But many people do not understand that
EQUAL IS NOT EQUAL and therefore statistic is a science that is
senseless.
Even if chances are equal, you have to play 5000 or more events.
Therefore we don't need statistics when we have better instruments.
And we have.
Nose. Ear. Stomach. Heart. Emotions.
Everything is much more precise than statistics.

mclane

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

h...@xs4all.nl (Herbert Groot Jebbink) wrote:

>On 5 Dec 1996 17:18:11 GMT, TUESCHEN.MEDIZ...@t-online.de
>(Rolf W. Tueschen) wrote:

>>Title: SSDF Rating Listing ---- is cheating

>I did never think that I would use the Agent Kill-list function for

>this newsgroup, but YOU are in it now!

>Greetings, hgj
I guess a kill-list is a kind of self-lie .
Horses have this nice , I don't know the english word, SCHEUKLAPPEN.
With these nice things they don't have to see what happens next to
them.
Some people don't need SCHEUKLAPPEN because they don't see much at
all.
Others are active in censorship.
Others are too stupid to follow at all.

mclane

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

Hansjoerg Zech <zech...@sni.de> wrote:

>mclane wrote:
>>
>> Rebel8 will win over Genius and Co because it is stronger. Or ?
>> Although it has no special XYZ-book it is better.
>> But we will see.

>Rebel 8 has a negative score (9.5-10.5) against Genius 4 in the SSDF
>list ?!

Can be. I have not played many games with Genius4. Genius4 is weaker
than Genius3. So I played autoplayer with Genius3.
Also I have not played manually many games vs. Genius4 because Genius4
plays (for me) too boring.

mclane

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>As if it had something to do with you being German. More like it was
>because the contents of posts of yours which are intellgible span the
>range from vapid speculation to sinister disingenuity, crossing abject
>stupidity and sheer bollocks en route. How many months do you have to
>be a regular poster on the net before you're NOT a "newbie"? (Hint:
>You've posted 188 messages to the rec.games.chess.* hierarchy since
>September 1st, and probably quite a few more before then.)

What is the paradigm of the fast-searchers:
sometimes quantity can lead into quality ?!

...

>60 points ahead of the nearest competitor *on the same hardware* was
>big progress at a time when all the top program were within 10 points
>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
>new list.) There's always "congratulations" from the chairman when
>there's a new number one. You're trying to insinuate bias then there
>isn't any.

Rebel8 IS a great progress. Your are right and Rolf is wrong.

>Once technology eases people's lives, they seldom want to "go back to
>the old way". I don't have the fortune to own a snow blower, but you
>can bet that once I had it the snow shovel would stay in the garage.
>Keeping in mind (the obvious idea) that it is easier to autotest than
>to test manually, do you actually have a point to make here? All I
>see is some moron taking shots at people who provide what amounts to a
>public service because he doesn't have anything better to do.

Come on. Thats unfair.

>They post margins of error which reflect the number of games played by
>each hardware and software combination. Why that is not good enough
>for you is a mystery to me. Your "fundamental base of statistics" as
>written is some of that bollocks I mentioned earlier.

Statistic is not a science. It is a religion.
There are no FUNDAMENTAL BASE OF STATISTICS.
They are as wrong as the idea that forces influence material parts so
that the apple falls down the earth.
The apple is nbot falling because there is a force called gravitation.
And there is no fundamental base of .... that RT has not understand.
Your statement is a prejudice against RT. You hide it because you
don't say: Rolf - you are not well-educated enough to understand.
But instead you say: you don't have the fundamental Base of
statistics.

Thats prejudicel. Not more. Not less.
Without any attack: you should read a book about
latest partical-physics and mathzematical-theorems instead of praying
ideas of closed statistical-rule-bases. Because these fundamental
stuff is out of an old paradigm called classisistic science.
This war was lost a couple of years before Einstein found out his
theories of relativity. This is a couple of years ago. 90 or 100 years
ago. So YOU are living in the wrong century, not RT.
All these mechanical ideas are wrong since the time Einstein made hs
first attack on them.

I know that this has not much to do with chess, but your attacks that
RT has no fundamental base of statistics has also nothing to do with
chess. Again: thats a prejudice.

>Does this have something to do with chess? Or the SSDF?

Do you want to force him what he should say and what not ?
"I will fight for the right that you can say what you want" or how was
the sentence ?!

>Oh, it was a stupid analogy, with an attempt to be comic near the end.
>I see. Ha ha. Real funny, that. It would be nice if the win
>percentage was closer to 50%, but statistical aberrations such as
>these are bound to happen once in a while. Indeed, I would be
>suspicous if they did not.

>You may or may not be trying to say something of relevance in the
>above paragraph. Don't bother clarifying if you don't feel like it.
>In fact, you don't even have to bother posting, I won't miss you.

Nice language. Not colloquial but also not very friendly.
But now I undersand what politeness is.
Say somebody is shit without saying it!!
Very polite.

>SSDF tests new versions when they get them. By March's list, you'll
>be able to see how Rebel 8.0, M-Chess 6.0, Chess Genius 5.0, Hiarcs
>5.0, and any other new programs do, with a lot of interplay between
>them in particular. Why would you want to delay publication of
>results of games? If a programmer beats competitors to market by a
>couple months, why shouldn't his product appear sooner? Perhaps
>another programmer decides to delay for a bit and continue to improve
>his software, eventually yielding a higher-rated product than it would
>have been had he published it sooner. Why should you not begin
>testing with the released product? Consumers are especially
>interested in the latest software available. If you want to see how
>the top programs do against only each other it is very easy to pull
>this information out of the SSDF list, which includes a result
>breakdown by opponent. Really independent testers test what they've
>got when they get it.

>Dave Gomboc
>drgo...@a.stu.athabascau.ca

You are right- we are 1/2 or even more advance in time. When I wrote
my Rebel8 article in AUGUST, I had Rebel8 before and tested it.
So now in November they discuss what I forgot in august after have
written down the article.

Tord Kallqvist Romstad

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

mclane (mcl...@prima.ruhr.de) wrote:
: Can be. I have not played many games with Genius4. Genius4 is weaker

: than Genius3. So I played autoplayer with Genius3.
: Also I have not played manually many games vs. Genius4 because Genius4
: plays (for me) too boring.

Interesting...
I have excactly the oposite opinions. Genius 4 seems slightly stronger
(especially in tactics), and Genius 3 definitely plays more boring than
Genius 4. I look forward to testing Genius 5, which I ordered from
Gambit-soft a few days ago.

Tord

Peter Fendrich

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

brucemo wrote:

>
> Dave Gomboc wrote:
>
> > There's no cheating factor. He's simply stating the obvious (well,
> > apparently not obvious, since it needs to be explained to you): the
> > more games that are played in the SSDF system, the less that future
> > games will affect measurements are taken. Rebel 8.0's rating will
> > continue to shift with all the rest, as games continue to be played.
> > (Exercise: demonstrate to yourself that the rating of a program on the
> > SSDF list can move between lists despite not having any games played
> > between a list and the list previous to it.)
>

I think some clarifications are needed.
When ratings are computed *all* the games for all times are used
every time.
Suppose a program has played 500 games and achieved some rating and will
play another 20 games until the next list, then on this list the
rating will be based on 520 games and its rating is completely
recomputed.
The new rating will probably not be changed much.

The ELO-formulas are the same but there is no need to let the last games
be of greater importance than the first ones.

This is not suitable for us humans because our strength is varying over
time. Ratings for humans are a lot more unstable than the SSDF ratings
are in this perspective.

So far so good but soon we are facing a new, really bad, problem:
Programs will learn and change their strength during time and this
will make the SSDF approach inappropriate.
Even worse:
Different individuals of the same program will have different history
and
therefore have different strength and all is changing over time :-<

Every copy of a program will be an 'individual' by itself.
Another con: this affect the way testing can be made.
The balancing and the order the of the opponents will be more important
than before. If some humans are playing the program for fun now and then
this will have impact on the strength as well.

I don't think this is of any importance right now and not in the nearest
future, but learning alg's will be better. When the same program with
and without learning shows completely different strengths this is a
real
problem for the SSDF testing.

I have no good ideas how to deal with this and solutions are welcome!
Hopefully the leading team of SSDF still will listen to my suggestions
when I have any... :)

To hell with learning ... ;-)

--
J-P Fendrich

Peter Fendrich

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

brucemo wrote:

>
> Is the SSDF list vulnerable to any of this, or do you guys use a
> completely different rating system or something like that?
>
> bruce

I gave an answear to this with the new subject 'The SSDF rating
program'.
Except for the new topics covered, guess why!

--
J-P Fendrich

Dave Gomboc

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

In article <E2C48...@news.prima.ruhr.de>,
mclane <mcl...@prima.ruhr.de> wrote:
#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
#
#>As if it had something to do with you being German. More like it was
#>because the contents of posts of yours which are intellgible span the
#>range from vapid speculation to sinister disingenuity, crossing abject
#>stupidity and sheer bollocks en route. How many months do you have to
#>be a regular poster on the net before you're NOT a "newbie"? (Hint:
#>You've posted 188 messages to the rec.games.chess.* hierarchy since
#>September 1st, and probably quite a few more before then.)
#
#What is the paradigm of the fast-searchers:
#sometimes quantity can lead into quality ?!

:-)

#...
#
#>60 points ahead of the nearest competitor *on the same hardware* was
#>big progress at a time when all the top program were within 10 points
#>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
#>new list.) There's always "congratulations" from the chairman when
#>there's a new number one. You're trying to insinuate bias then there
#>isn't any.
#
#Rebel8 IS a great progress. Your are right and Rolf is wrong.

Ok.

#>Once technology eases people's lives, they seldom want to "go back to
#>the old way". I don't have the fortune to own a snow blower, but you
#>can bet that once I had it the snow shovel would stay in the garage.
#>Keeping in mind (the obvious idea) that it is easier to autotest than
#>to test manually, do you actually have a point to make here? All I
#>see is some moron taking shots at people who provide what amounts to a
#>public service because he doesn't have anything better to do.
#
#Come on. Thats unfair.

Actually, I thought that calling people racist because they disagree
with you was unfair. In fact, it was what pissed me off and caused me
to write my earlier response in the first place.

#
#>They post margins of error which reflect the number of games played by
#>each hardware and software combination. Why that is not good enough
#>for you is a mystery to me. Your "fundamental base of statistics" as
#>written is some of that bollocks I mentioned earlier.
#
#Statistic is not a science. It is a religion.
#There are no FUNDAMENTAL BASE OF STATISTICS.
#They are as wrong as the idea that forces influence material parts so
#that the apple falls down the earth.
#The apple is nbot falling because there is a force called gravitation.
#And there is no fundamental base of .... that RT has not understand.
#Your statement is a prejudice against RT. You hide it because you
#don't say: Rolf - you are not well-educated enough to understand.
#But instead you say: you don't have the fundamental Base of
#statistics.

You really feel that statistics is a religion, and not a science?
I guess you're entitled to your opinion. In my view, statistics CAN be
done rigoursly, and often is.

#
#Thats prejudicel. Not more. Not less.
#Without any attack: you should read a book about
#latest partical-physics and mathzematical-theorems instead of praying
#ideas of closed statistical-rule-bases. Because these fundamental
#stuff is out of an old paradigm called classisistic science.
#This war was lost a couple of years before Einstein found out his
#theories of relativity. This is a couple of years ago. 90 or 100 years
#ago. So YOU are living in the wrong century, not RT.
#All these mechanical ideas are wrong since the time Einstein made hs
#first attack on them.

I think you've wandered off somewhere above...

#I know that this has not much to do with chess, but your attacks that
#RT has no fundamental base of statistics has also nothing to do with
#chess. Again: thats a prejudice.
#
#>Does this have something to do with chess? Or the SSDF?
#
#Do you want to force him what he should say and what not ?
#"I will fight for the right that you can say what you want" or how was
#the sentence ?!
#
#
#>Oh, it was a stupid analogy, with an attempt to be comic near the end.
#>I see. Ha ha. Real funny, that. It would be nice if the win
#>percentage was closer to 50%, but statistical aberrations such as
#>these are bound to happen once in a while. Indeed, I would be
#>suspicous if they did not.
#
#
#>You may or may not be trying to say something of relevance in the
#>above paragraph. Don't bother clarifying if you don't feel like it.
#>In fact, you don't even have to bother posting, I won't miss you.

..,here you are returning to 'earth' :-)

#Nice language. Not colloquial but also not very friendly.
#But now I undersand what politeness is.
#Say somebody is shit without saying it!!
#Very polite.

Actually I thought it was rather impolite. I am not saying that
someone is 'shit', as you put it. I am genuinely unclear as to
whether he was trying to say something of importance in that paragraph
or not! As for the "don't even ... bother posting", that is more a
reflection of the high esteem I hold of the series of posts spanning a
two-month period (so far!) by Rolf Tueschen which attempted to drag
Ed Schroder through the mud than it pertains to the SSDF discussion,
though the threads do cross from time to time.

#>SSDF tests new versions when they get them. By March's list, you'll
#>be able to see how Rebel 8.0, M-Chess 6.0, Chess Genius 5.0, Hiarcs
#>5.0, and any other new programs do, with a lot of interplay between
#>them in particular. Why would you want to delay publication of
#>results of games? If a programmer beats competitors to market by a
#>couple months, why shouldn't his product appear sooner? Perhaps
#>another programmer decides to delay for a bit and continue to improve
#>his software, eventually yielding a higher-rated product than it would
#>have been had he published it sooner. Why should you not begin
#>testing with the released product? Consumers are especially
#>interested in the latest software available. If you want to see how
#>the top programs do against only each other it is very easy to pull
#>this information out of the SSDF list, which includes a result
#>breakdown by opponent. Really independent testers test what they've
#>got when they get it.
#
#>Dave Gomboc
#>drgo...@a.stu.athabascau.ca
#
#You are right- we are 1/2 or even more advance in time. When I wrote
#my Rebel8 article in AUGUST, I had Rebel8 before and tested it.
#So now in November they discuss what I forgot in august after have
#written down the article.

Dave Gomboc
drgo...@a.stu.athabascau.ca

Dave Gomboc

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

In article <32B049...@nwlink.com>, brucemo <bru...@nwlink.com> wrote:

>Dave Gomboc wrote:
>
>> There's no cheating factor. He's simply stating the obvious (well,
>> apparently not obvious, since it needs to be explained to you): the
>> more games that are played in the SSDF system, the less that future
>> games will affect measurements are taken. Rebel 8.0's rating will
>> continue to shift with all the rest, as games continue to be played.
>> (Exercise: demonstrate to yourself that the rating of a program on the
>> SSDF list can move between lists despite not having any games played
>> between a list and the list previous to it.)
>

>This depends a lot on how they do the rating system.

Yes, I agree.

[Bruce's remarks regarding standard ELO setups deleted]

>Is the SSDF list vulnerable to any of this, or do you guys use a
>completely different rating system or something like that?

I was confident when I posted my earlier message that the SSDF treat
all games as one large dataset, i.e. no "update" calculations are done
from one list to the next. Instead ratings for the entire table are
solved for based on all results in the table in roughly the manner I
will detail below:

1) Collect all results of players against each other, and record them
in a large 2-d array.

2) Assign every player an identical initial rating (RI) rating (we'll
pick 2000 for now).

3) Assign an 'updated rating' (RU) based upon the performance rating formula,
which is RO + 400 * (W-L)/G, where RO is average opponent rating
W is number of wins
L is number of losses
G is number of games played

To calculate average opponent rating, you need to 'scale', e.g.

If I play a 2200 once, a 2000 twice, and a 1900 once, my RO is 2025.

2200*1 + 2000*2 + 1900*1
------------------------ = 2025 = RO
1+2+1

4) Make the RU value for each player their new RI, and iterate until
the ratings converge.

5) If you have done a previous list, it is probably a good idea to
find the best fit between the previous list and this one, so that the
ratings are assured to be relatively level from list to list. (There
is an assumption being made here, and indeed with the whole
time-invariant procedure, see below for a discussion regarding that.)
So you add (or subtract) a constant to ALL the ratings on the new
list. The right constant to use is the one which minimizes the least
sum of squares of differences.

Maybe I haven't been too clear here, I'll give an example:

Old list: A 1800 B 2200 C 2000

New list before step 5: A 1600 B 2150 C 1900 D 2100

We wish to minimize the function

(1800-1600+x)^2 + (2200-2150+x)^2 + (2000-1900+x)^2

= 40000 + 400x + x^2 + 2500 + 50x + x^2 + 10000 + 100x + x^2
= 3x^2 + 550x + 52500

So we equate zero and the derivative:

0 = 6x + 550, which works out to x=-91.667

Now we subtract this constant to the new values to get the new list:

A = 1600-(-91.667) = 1691.667
B = 2150-(-91.667) = 2241.667
C = 1900-(-91.667) = 1991.667
D = 2100-(-91.667) = 2191.667

A couple of simplifications have been made in step 5.. really it is
assuming that all ratings were equally reliable, as in the case

Old List: A 1800 after 6 games, B 2200 after 6 games, C 2000 after 6 games
New List: A, B, C, D have played 8 each.

In reality, some machines have played more games than others, and as a
result instead of minimizing the function I have above, the goal
should be to minimize a similar but more complex function, which
accounts for the variability in confidence in ratings.

On another note, a somewhat reasonable function would be to restrict
minimization of difference to those players who do not have new
results. (This wouldn't work when most of the pool plays regularly,
but is great for a computer list, since most of the computers don't
really play any more and their skill level is assumed to remain
constant.) This particular more complex function would assign
identical ratings to players on the list who haven't played since the
last list. Clearly, this wasn't the function I had in mind when I
wrote my quoted paragraph (at the top of the message.)

Okay, I mentioned an assumption earlier, and here's the place I'll
elaborate on it. I am assuming time-independence of the player's
skills. This wouldn't work in a system where player's abilities
evolved, and that's why the models Bruce came up with would break my
assertions. In those systems, the order in which the games are played
directly affects the allocation of rating points to each player.

SSDF is rating computers, and to the best of my knowledge has also
made the time-independence assumption. With 'learning features' in
some of today's new programs, this assumption just became invalid. It
is unlikely that in the immediate term there is much effect, but in
the long term a time-dependent system probably should be used.

Also invalidated is the presumption that the quantity of games in
matches and the order in which matches are played does not affect the
ratings. Before MChess Pro 6.0's entry into the list this December,
it did not matter in the slightest, but now it may make a difference
of some ELO points because a program may discover a variation is good
against player A and use it against player B, or discover a variation
is good against player B and use it against player A. A player may
play A 10 times followed by B 10 times, and results could be different
than if it had played "A,B,B,A,A,B,B,A,A,B,B,A,A,B,B,A,A,B,B,A" (10
times each still, but the games are interspersed). This is quite a
different effect from Rebel 8.0 playing games before MChess Pro 6.0,
because Rebel 8.0 isn't a book learner. One could argue that it's
unfair to new learners that MChess Pro 6.0 will already have had
learning practice by the time they arrive, and so they might not get
the chance to bust MChess Pro's 1.e4 e5 2.Bc4 or something. One could
argue that as it was released earlier it is proper for it to have
already accomplished some learning. Someone could mention that as
more than one copy of the software is in testing use, some versions
will have learned things that others haven't, which is less than
ideal. The list goes on.

Whether these things are something that requires immediate attention
depends on if the 'learning' features make a significant change to the
real ELO of the program. If it matters only 2 or 3 points over 800
games then it is nothing to worry about, and can safely be dealt with
by a small adjustment to the margin of error. If the strength
difference exceeds the margin of error (~30 ELO) over 800 games than
SSDF testing policy is about to become a lot more complicated, and
I'll apologize to Rolf Tueschen in advance.

--
Dave Gomboc
drgo...@a.stu.athabascau.ca

Rolf W. Tueschen

unread,

Dec 13, 1996, 3:00:00 AM12/13/96

to

mcl...@prima.ruhr.de (mclane) wrote:

>drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>>60 points ahead of the nearest competitor *on the same hardware* was
>>big progress at a time when all the top program were within 10 points
>>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
>>new list.) There's always "congratulations" from the chairman when
>>there's a new number one. You're trying to insinuate bias then there
>>isn't any.

>Rebel8 IS a great progress. Your are right and Rolf is wrong.

----------------------------------------------------------------------------------------------------------------

To give precision:

I n e v e r talked about Rebel8 as bad prg.
On the contrary I read your big article about Rebel8. And I trusted you
that Rebel8 was a strong prg.

Here however we discuss SSDF. And in detail the quoted part refers to my
critics of Thoralf in CSS who said Rebel8 demonstrates a clear advance in
*computer chess wisdom*.

That in fact is stupid. At least Deep Blue and others shouldn't be
forgotten. *I* don't know the results of Shredder vs. Rebel8. I don't know
yet exact scores of Rebel8 against the other updates. Against Genius4 he
lost!
But that wouldn't change your verdict that Rebel8 surely is a strong prg.

I couldn't test or play Rebel8 for myself.

It's so bad that Ed cheated with his surprise action. Because as new user
you had to buy further programs to do converting.
As concerning this some here in the group didn't know that. But in the
surprise action this tool was NOT included. I got this from Ed personally.
So with all my cbf material I repeat Ed trapped me in a little cheating as
other *Kaffefahrten*- or *Take part you always win action*-organizer do
too.
Important to know that all that could be said because Ed did NOT inform
about these costs in advance. So I still call it cheating. Period.
Ed?

brucemo

unread,

Dec 14, 1996, 3:00:00 AM12/14/96

to

a bean wrote:

> | 2. Solving certain tactical positions quickly may be a good
> | indicator of speed chess ability but may not translate
> | well into 40/2 games.
>
> Exactly why we need this - the purpose of a CAT test is to determine the
> relative playing strengths for various time controls and positional types.

My provider has decided to implement one of these "you must have more new
stuff than quoted stuff", so I can't quote some of the parts of this that I
want to respond to, since I don't intend to write a heck of a lot.

For determining the strength of a program, you can't beat playing a real
game. There are numerous problems with test suites.

You can complain about hidden agendas of the suite builder if a program you
favor performs poorly. You might have a point, maybe the suite builder does
want to favor another program, and has tuned the suite for that program. On
the other hand, perhaps you are wrong. In either case, I predict a huge
angry thread in here.

It would be difficult to make a suite that conforms perfectly to the demands
of the average game. The tendency is to put in positions that probably won't
come up in practical play. One of the suites has a Q vs P ending, where one
the superficially most obvious move draws, while another move wins. If you
have a tablebase you get this right, if you don't, you probably get it right
but maybe you don't. In practical play this might come up once in 500 games,
but in this suite it counts for like 1/30 of your total score. There's
another one I've seen that's designed to trip up null move programs in
endings. The position is very much contrived, it is not likely to happen in
practical play. Many of the suites have examples of brilliant attacks, but
in human vs computer play the computer is more often defending against wild
attacks.

Just getting the right proportion of opening vs middlegame vs ending
positions in a suite would be hard.

You'd have to be careful to make very certain that a program did something
for the right reason, and this might be kind of hard. For instance, if the
point of a positional problem is to know that the bishop pair is worth more
than giving the opponent a doubled pawn, how do you determine that the
program knows that the bishop pair is worth more than a doubled pawn, rather
than that it doesn't know that a doubled pawn is bad in any case?

And of course, there are suites where the "best move" is really a matter of
taste.

So I be more prone to trust the results of real games, which is not to say
there aren't problems with THAT method, too.

bruce

a bean

unread,

Dec 14, 1996, 3:00:00 AM12/14/96

to

brucemo <bru...@nwlink.com> wrote in article <32B2EE...@nwlink.com>...
| a bean wrote:

| > Exactly why we need this - the purpose of a CAT test is to determine
the
| > relative playing strengths for various time controls and positional
types.
|

|| For determining the strength of a program, you can't beat playing a real

| game.

How many games would it take to expose a program to enough positional and
tactical elements?

Most of the moves in a game are either fairly obvious or a "matter of
taste" - and therefore don't help at all in terms of evaluating the
strength of a given program.

|In either case, I predict a huge
| angry thread in here.

Amused "yes" - Angry "no".
Why so much resistance to good ideas?

| It would be difficult to make a suite that conforms perfectly to the
demands
| of the average game.

It doesn't have to.

The tendency is to put in positions that probably won't
| come up in practical play. One of the suites has a Q vs P ending, where
one
| the superficially most obvious move draws, while another move wins. If
you
| have a tablebase you get this right, if you don't, you probably get it
right
| but maybe you don't. In practical play this might come up once in 500
games,
| but in this suite it counts for like 1/30 of your total score.

I agree - and that is why the test positions need to be categorized.

There's
| another one I've seen that's designed to trip up null move programs in
| endings. The position is very much contrived, it is not likely to happen
in
| practical play. Many of the suites have examples of brilliant attacks,
but
| in human vs computer play the computer is more often defending against
wild
| attacks.

Categorizing positions again is the answer.

| Just getting the right proportion of opening vs middlegame vs ending
| positions in a suite would be hard.

It is not necessary.

The purpose of the test is NOT to form a mathematical prediction as to OTB
results,
but rather to be able to objectively compare the speed and accuracy of
chess programs in certain types of positions.

| You'd have to be careful to make very certain that a program did
something
| for the right reason, and this might be kind of hard. For instance, if
the
| point of a positional problem is to know that the bishop pair is worth
more
| than giving the opponent a doubled pawn, how do you determine that the
| program knows that the bishop pair is worth more than a doubled pawn,
rather
| than that it doesn't know that a doubled pawn is bad in any case?

I agree - The positions need to have clear solutions or there would not be
an objective way to credit the answers. In the above example it is not
enough to say that the bishop pair is worth more than taking on the doubled
pawn - it would have to be proven by demonstrating a forced variation
leading to advantage. If the test position is too fuzzy to produce this
type of solution then it would have to either be modified to do so or
thrown out altogether. Otherwise we would have.....

| And of course, there are suites where the "best move" is really a matter
of
| taste.

What would you say if correct answers on the SAT where a "matter of taste"?
These types of positions could NOT be used at all in order to keep the
results objective.

| So I be more prone to trust the results of real games, which is not to
say
| there aren't problems with THAT method, too.

I very much trust the results - the problem is that -
the results don't always tell you what you want to know!

Each game is unique, and therefore the results derived are also then
unique.
The point of testing with standardized positions is that it is
STANDARDIZED,
and can then be used as a basis for comparison. (not necc. predicting OTB
performance!)

I will try to demonstrate some weaknesses of relying on game results to
form opinions of chess playing ability.

[ProgramA defeats ProgramB in a 10 game match 6-4] - assuming equal
hardware.

What does this REALLY tell us about the ACTUAL chess playing strengths of
these programs?

What if I told you that ProgramA used KillerBook to win 3 games and that 3
other games were won in theoretically drawn endgames by the use of endgame
tablebase. Once both programs were out of book though - ProgramB clearly
demonstrated a much higher degree of raw chess playing ability - and was
able to easily out-perform and out-calculate its opponent in most types of
positions.

Which program would you rather have to analyze a complex middlegame?
Which program would make a better endgame sparing partner?

And now the really important question:

What did the game result score (6-4) have to do with answering any of the
last 2 questions?!

Game results don't tell the whole story.

The only way that I could qualify the game results would be to actually
play over the games, and when certain key or problem type positions were
encountered - I would observe how each program handled them.

This is in effect - a sloppy form of testing by position anyway!

Rolf W. Tueschen

unread,

Dec 14, 1996, 3:00:00 AM12/14/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>Yes, I agree.

---------------------------------------------------------------------------------------------------------------
see No 1 below
---------------------------------------------------------------------------------------------------------------

-------

>4) Make the RU value for each player their new RI, and iterate until
> the ratings converge.

>5) If you have done a previous list, it is probably a good idea to
>find the best fit between the previous list and this one, so that the
>ratings are assured to be relatively level from list to list.

----------------------------------------------------------------------------------------------------------
see No 2 below
----------------------------------------------------------------------------------------------------------

>A couple of simplifications have been made in step 5.. really it is
>assuming that all ratings were equally reliable, as in the case

> Old List: A 1800 after 6 games, B 2200 after 6 games, C 2000 after 6 games
> New List: A, B, C, D have played 8 each.

>In reality, some machines have played more games than others, and as a
>result instead of minimizing the function I have above, the goal
>should be to minimize a similar but more complex function, which
>accounts for the variability in confidence in ratings.

>On another note, a somewhat reasonable function would be to restrict
>minimization of difference to those players who do not have new
>results. (This wouldn't work when most of the pool plays regularly,
>but is great for a computer list, since most of the computers don't
>really play any more and their skill level is assumed to remain
>constant.) This particular more complex function would assign
>identical ratings to players on the list who haven't played since the
>last list. Clearly, this wasn't the function I had in mind when I
>wrote my quoted paragraph (at the top of the message.)

---------------------------------------------------------------------------------------------------------------
see No 3 below
---------------------------------------------------------------------------------------------------------------

I'd like to give three questions of understanding to be able to judge on
this detailed explanation. Thanks for possible answers.

Note please, that the important second part of the above quoted post isn't
of interest for me because I did and still claim that SSDF did cheat under
conditions before *learning* came up.

No 1

Is assumption true that value for RO depends a lot on Elos of the prgs AND
how many games are played against low or high leveled Elos?

(Remember my critics of SSDF proceduring of DIFFERENT game numbers and
different adversairy prgs that are played by a certain prg?)

No 2

Regarding my claim that there could be no such time independence strength
value of a certain prg as such could someone show some evidence for the
truth of the following?

A certain prg P tested at a certain time against n other prgs with a
certain Elo E1 as result.

Then after some time same P tested against n-5 prgs AND some new n+5 prgs
with a E2 result.

And the assumption made by SSDF and author of former post is E1 = E2?

And if NOT directly then a certain complex such and such function would be
utilized to make them equal?

No 3

And especially because it *happened* that some prgs played different
quantity of games a *complex correction* was needed?

(Remember my claim that it was intollerable to match prgs vs different
other prgs AND more so with different quantity?)

What this complex functions looks like?

May I add finally my former claim that Elo or SSDF results didn't/ and
couldn't give a certain *real* value for the strength of a prg?
I meant that each calculated score only could show a value in relation to
other tested prgs at a certain time.

And what I did never understand why SSDF whose staff IS in close contact
with their testers couldn't distribute a balanced testing proceduring
before hand instead of inventing some *complex* corrections afterwards.

Rolf W. Tueschen

unread,

Dec 14, 1996, 3:00:00 AM12/14/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>Actually I thought it was rather impolite. I am not saying that
>someone is 'shit', as you put it. I am genuinely unclear as to
>whether he was trying to say something of importance in that paragraph
>or not! As for the "don't even ... bother posting", that is more a
>reflection of the high esteem I hold of the series of posts spanning a
>two-month period (so far!) by Rolf Tueschen which attempted to drag
>Ed Schroder through the mud than it pertains to the SSDF discussion,
>though the threads do cross from time to time.

......

>Dave Gomboc
>drgo...@a.stu.athabascau.ca

---------------------------------------------------------------------------------------------------------

Very funny and highest level for sure.

Take that:

Ed Schroder did violate massively the rules of etiquette of newsgroups.
Ed Schroder did promotion in the absence of concurrents in SSDF.
Ed Schroder did cheating with his *surprise action*.
Ed Schroder did mean and mediocre behind the curtain war via email.
Ed Schroder did post exact wordings of emails.
Ed Schroder did cheat with his 40 games called killer games.
Ed Schroder did cheat with the matches on his webpage (Doubles)
Ed Schroder did cheat on BT position testers. (See CSS 1994)
Ed Schroder did threaten to sue a newbie for reporting this.
Ed Schroder did writings of ultimate surrender for over one week.

So who is responsible for the mud?
The reporter?

Tim Mirabile

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

Peter Fendrich <j...@vd.volvo.se> wrote:

>So far so good but soon we are facing a new, really bad, problem:
> Programs will learn and change their strength during time and this
> will make the SSDF approach inappropriate.
>Even worse:
> Different individuals of the same program will have different history
>and
> therefore have different strength and all is changing over time :-<
>
>Every copy of a program will be an 'individual' by itself.
>Another con: this affect the way testing can be made.
>The balancing and the order the of the opponents will be more important
>than before. If some humans are playing the program for fun now and then
>this will have impact on the strength as well.
>
>I don't think this is of any importance right now and not in the nearest
>future, but learning alg's will be better. When the same program with
>and without learning shows completely different strengths this is a
>real
>problem for the SSDF testing.
>
>I have no good ideas how to deal with this and solutions are welcome!
>Hopefully the leading team of SSDF still will listen to my suggestions
>when I have any... :)
>
>To hell with learning ... ;-)

Learning should probably be disabled during testing, but if the programs are
released with info already in the learning files, this should be allowed to be
used. Essentially you want it to play each game as if it were the first one out
of the shrink wrap.

--
Tim Mirabile <t...@mail.htp.com> - http://www.webcom.com/timm/
Visit my homepage for information on USCF & FIDE rated chess on Long Island.
TimM on the Free Internet Chess Server - telnet://fics.onenet.net:5000/
ICD/Your Move Chess & Games - http://www.icdchess.com/
The opinions of my employers are not necessarily mine, and vice versa.

Chris Whittington

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

TUESCHEN.MEDIZ...@t-online.de (Rolf W. Tueschen) wrote:
>

> mcl...@prima.ruhr.de (mclane) wrote:
>
> >drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
>

> >>60 points ahead of the nearest competitor *on the same hardware* was
> >>big progress at a time when all the top program were within 10 points
> >>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
> >>new list.) There's always "congratulations" from the chairman when
> >>there's a new number one. You're trying to insinuate bias then there
> >>isn't any.
>
> >Rebel8 IS a great progress. Your are right and Rolf is wrong.
>

> ----------------------------------------------------------------------------------------------------------------
>
> To give precision:
>
> I n e v e r talked about Rebel8 as bad prg.
> On the contrary I read your big article about Rebel8. And I trusted you
> that Rebel8 was a strong prg.
>
> Here however we discuss SSDF. And in detail the quoted part refers to my
> critics of Thoralf in CSS who said Rebel8 demonstrates a clear advance in
> *computer chess wisdom*.
>
> That in fact is stupid. At least Deep Blue and others shouldn't be
> forgotten. *I* don't know the results of Shredder vs. Rebel8. I don't know
> yet exact scores of Rebel8 against the other updates. Against Genius4 he
> lost!
> But that wouldn't change your verdict that Rebel8 surely is a strong prg.
>
> I couldn't test or play Rebel8 for myself.
>
> It's so bad that Ed cheated with his surprise action. Because as new user
> you had to buy further programs to do converting.
> As concerning this some here in the group didn't know that. But in the
> surprise action this tool was NOT included. I got this from Ed personally.
> So with all my cbf material I repeat Ed trapped me in a little cheating as
> other *Kaffefahrten*- or *Take part you always win action*-organizer do
> too.
> Important to know that all that could be said because Ed did NOT inform
> about these costs in advance. So I still call it cheating. Period.
> Ed?

I'll reply for Ed.

He is too polite to say what he really thinks.

FUCK-OFF.

Chris Whittington

Ed Schröder

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to TUESCHEN.MEDIZ...@t-online.de

From: TUESCHEN.MEDIZ...@t-online.de (Rolf W. Tueschen)

Perhaps you can explain what you mean with the word "cheating"?
I get the impression you do not understand the meaning of that word.

Cheating sounds as fraud.
Or in German "Betrug" (if I remember well)

: It's so bad that Ed cheated with his surprise action. Because as new

: user you had to buy further programs to do converting.

The DBUTIL.EXE program does not convert anything and you do not need it
to run Rebel8. It was for faithful Rebel users. We always give existing
customers something extra.

The "surprise" mailing was sent to about 1500 people by private email.
Nobody complained about cheating.

Offering people a discount is no cheating!

Some people wanted to be removed from the mailing list which we will do
of course. Others received 2-4 copies... :) Sorry about that again...

: Important to know that all that could be said because Ed did NOT

: inform about these costs in advance. So I still call it cheating.
: Period.
: Ed?

No period at all...

The mailing was a complete info, perhaps you misunderstood.
I have included the original email you got from me.
Perhaps you should read again, this time more carefully.

- Ed -

: Rolf Tueschen

Sorry for the spam folks but Rolf Tueschen has mentioned this subject
now 3 times as cheating...

------------------------------------------------------------------------
Dear Rebel user,

The new REBEL 8.0 will be released on September 1, 1996

Till November 30, 1996 we have planned a VERY special action!

REBEL 7.0 users...
If you are a satisfied Rebel7 user you probably want to upgrade to the
new Rebel 8.0 chess program against upgrade price. Normally you have to
return your original Rebel7 diskette to your retailer. But now you can
KEEP your original Rebel7 diskette AND get our database utility program
DBUTIL.EXE (HFL 49,00 DM 49,00 $29.95) for free!

NON Rebel7 users...
Also you (most of you have Rebel Decade) are able to get the new
Rebel 8.0 (HFL 199,00 DM 199,00 $149.00) against UPGRADE PRICE! (discount
is 40-50%). Don't ask us how this is possible, just take your profit!

How is this all possible?
We have created a special UPGRADE CHEQUE which you can send to your local
retailer. This is your PROOF to get the new Rebel 8.0 against upgrade
price by your local retailer. We like to mail you this cheque to your
postal address but of course we need your home address in this case. (the
Internet is a wonderful place but in this case we still need the postman!)

What to do...
Email us your postal address (if you are interested of course!) and we
will mail you the UPGRADE CHEQUE as soon as possible. We know some of you
already emailed us your postal address, please do it again otherwise we
(maybe) make mistakes and that is the last thing we want.

Important remarks...

1) Please remember that this action is valid TILL November 30, 1996!!
After that date the upgrade cheque is not valid anymore!

2) On our home page you can see a list of our world wide dealers (see
chapters "Dealer List" and "Email your Dealer"). So you can contact your
retailer directly from our home page by email or get their phone number.
They all are happy to answer all your remaining questions.

3) If you want more information about the new Rebel 8.0 we suggest you
also to take a look on our home page see chapter "Rebel 8.0 announcement
and screen shots". Here you will find 12 screen shots and a short
description of the 25 new features above Rebel 7.0

4) We have also compared Rebel Decade with the new Rebel 8.0 Besides
from the playing strength, Rebel 8.0 is about 200 ELO points STRONGER,
we counted 50 extra possibilities above Rebel Decade!

The database utility DBUTIL.EXE ...
As already said Rebel7 users can keep their original diskette and also
become the database utility program for free. Here is a short list of
the 11 extra database functions:

1) Import and Export games which enables you to make a database as big
as you want (a wide range of selections are available!).
2) Recover damaged databases.
3) Remove double games in a database (very fast!)
4) Delete games in a database (a wide range of selections are available).
5) Rename up to 1000 player names in ONE session done by a predefined
TEXT file. (make player names unique / uniform)
6) Insert / change up to 1000 Elo ratings of specific players in ONE
session done by a predefined TEXT file. (make Player Elo ratings uniform).
7) Compress a database. etc. etc.

Be sure you upgrade to the new Rebel 8.0 before November 30, 1996!
If you upgrade after November 30, 1996 you need to return your original
diskette to you local retailer and you also will miss the database
utility program!

Best regards,

- Ed Schroder -

_______________________________________________________________
Schroder BV http://www.xs4all.nl/~rebchess/
Software Development
P.O. Box 6365 mailto:rebc...@xs4all.nl
7401JJ Deventer
The Netherlands
_______________________________________________________________

Tim Mirabile

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

brucemo <bru...@nwlink.com> wrote:

>a bean wrote:
>
>> | 2. Solving certain tactical positions quickly may be a good
>> | indicator of speed chess ability but may not translate
>> | well into 40/2 games.
>>
>> Exactly why we need this - the purpose of a CAT test is to determine the
>> relative playing strengths for various time controls and positional types.
>

>My provider has decided to implement one of these "you must have more new
>stuff than quoted stuff", so I can't quote some of the parts of this that I
>want to respond to, since I don't intend to write a heck of a lot.

Sorry if this is off topic, but how do these things work? Are you sure it's not
your Netscape [X-Mailer: Mozilla 2.01 (Win95; U)] enforcing that? I remember
that I had the same thing using Netscape 1.1, and when I switched to Free Agent,
it let me post any way I wanted. Now it seems to me that if the news server had
to enforce it, it could be pretty complicated to determine what is quoted and
what is not, especially since many newsreaders use different quoting characters.
At least at the editing level, Netscape can determine which lines you touched
and which you didn't. I used to fool it by removing the quoting characters in
front of the blank lines. This is much better than what some people do around
here, which is to add a lot of garbage to the end of the message, which kind of
defeats the purpose of saving bandwidth.

mclane

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>:-)

>Ok.

I have not called anybody a racist because of the reasons you
mentioned.
Chris called me a fascists/racist because I said that this is
APE-chess and he said that APE is a sign that I am thinking british or
austrians are apes.
But: if anybody who uses the word APE for something is a racists.....

So maybe you mix up what I said and what not.

>#
>#>They post margins of error which reflect the number of games played by
>#>each hardware and software combination. Why that is not good enough
>#>for you is a mystery to me. Your "fundamental base of statistics" as
>#>written is some of that bollocks I mentioned earlier.
>#
>#Statistic is not a science. It is a religion.
>#There are no FUNDAMENTAL BASE OF STATISTICS.
>#They are as wrong as the idea that forces influence material parts so
>#that the apple falls down the earth.
>#The apple is nbot falling because there is a force called gravitation.
>#And there is no fundamental base of .... that RT has not understand.
>#Your statement is a prejudice against RT. You hide it because you
>#don't say: Rolf - you are not well-educated enough to understand.
>#But instead you say: you don't have the fundamental Base of
>#statistics.

>You really feel that statistics is a religion, and not a science?
>I guess you're entitled to your opinion. In my view, statistics CAN be
>done rigoursly, and often is.

Science is a religion itself.
There is no objectivity. You can only try to be very near, within a
certain range, called error-margin.
But still you will never reach the truth.

In my opinion statistics is not to use to find out what will happen,
but to interprete why something has happened before.
So statistics do not show us anything that is not known before.
As an example see ssdf-list. They do not show me anything I do not
know months before. But they proof my ideas very well.
If you believe that statistics shows us something that has not
happened, then you could do the same influence on the future by going
into a church and burn a candle-light. It would have the same effect.

>#
>#Thats prejudicel. Not more. Not less.
>#Without any attack: you should read a book about
>#latest partical-physics and mathzematical-theorems instead of praying
>#ideas of closed statistical-rule-bases. Because these fundamental
>#stuff is out of an old paradigm called classisistic science.
>#This war was lost a couple of years before Einstein found out his
>#theories of relativity. This is a couple of years ago. 90 or 100 years
>#ago. So YOU are living in the wrong century, not RT.
>#All these mechanical ideas are wrong since the time Einstein made hs
>#first attack on them.

>I think you've wandered off somewhere above...

Off- topic, thats for sure.
But the theories of Einstein are still amazingly important to our
paradigm.

It was impollite, but still harmless in relation to my own postings.

>I am not saying that
>someone is 'shit', as you put it. I am genuinely unclear as to
>whether he was trying to say something of importance in that paragraph
>or not!

Right. That was my point. You are GENUINELY UNCLEAR.
Thats what I always say: some people here are IMPOLITE but in the same
way they hide their opinion with DISCLAIMERS and behave like the
Heisenbergsche Unschaerferelation.

> As for the "don't even ... bother posting", that is more a
>reflection of the high esteem I hold of the series of posts spanning a
>two-month period (so far!) by Rolf Tueschen which attempted to drag
>Ed Schroder through the mud than it pertains to the SSDF discussion,
>though the threads do cross from time to time.

I confirm.

brucemo

unread,

Dec 15, 1996, 3:00:00 AM12/15/96

to

Tim Mirabile wrote:

> Sorry if this is off topic, but how do these things work? Are you sure it's not
> your Netscape [X-Mailer: Mozilla 2.01 (Win95; U)] enforcing that? I remember
> that I had the same thing using Netscape 1.1, and when I switched to Free Agent,
> it let me post any way I wanted.

I have no idea how it works but it's bugging the heck out of me. I think it's my
provider doing it, since I haven't changed my config lately, and I know they've been
trying to get their news server running better. So someone apparently turned on a cool
"feature".

bruce

Dave Gomboc

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

In article <32B2EE...@nwlink.com>, brucemo <bru...@nwlink.com> wrote:
#a bean wrote:
#
#> | 2. Solving certain tactical positions quickly may be a good
#> | indicator of speed chess ability but may not translate
#> | well into 40/2 games.
#>
#> Exactly why we need this - the purpose of a CAT test is to determine the
#> relative playing strengths for various time controls and positional types.
#
#My provider has decided to implement one of these "you must have more new
#stuff than quoted stuff", so I can't quote some of the parts of this that I
#want to respond to, since I don't intend to write a heck of a lot.

My provider does the same thing.. it is annoying. It is handled at
the news server, not the editor level, so I find myself making a small
macro in Emacs every time to change the first character in a line to
an octothorpe ('#').. why do they have to make things difficult? It's
not as if people won't figure out a way around it anyway.. sigh.

#For determining the strength of a program, you can't beat playing a real
#game. There are numerous problems with test suites.
#
#You can complain about hidden agendas of the suite builder if a program you
#favor performs poorly. You might have a point, maybe the suite builder does
#want to favor another program, and has tuned the suite for that program. On
#the other hand, perhaps you are wrong. In either case, I predict a huge
#angry thread in here.
#
#It would be difficult to make a suite that conforms perfectly to the demands
#of the average game. The tendency is to put in positions that probably won't
#come up in practical play. One of the suites has a Q vs P ending, where one
#the superficially most obvious move draws, while another move wins. If you
#have a tablebase you get this right, if you don't, you probably get it right
#but maybe you don't. In practical play this might come up once in 500 games,
#but in this suite it counts for like 1/30 of your total score. There's
#another one I've seen that's designed to trip up null move programs in
#endings. The position is very much contrived, it is not likely to happen in
#practical play. Many of the suites have examples of brilliant attacks, but
#in human vs computer play the computer is more often defending against wild
#attacks.
#
#Just getting the right proportion of opening vs middlegame vs ending
#positions in a suite would be hard.
#
#You'd have to be careful to make very certain that a program did something
#for the right reason, and this might be kind of hard. For instance, if the
#point of a positional problem is to know that the bishop pair is worth more
#than giving the opponent a doubled pawn, how do you determine that the
#program knows that the bishop pair is worth more than a doubled pawn, rather
#than that it doesn't know that a doubled pawn is bad in any case?
#
#And of course, there are suites where the "best move" is really a matter of
#taste.
#
#So I be more prone to trust the results of real games, which is not to say
#there aren't problems with THAT method, too.
#
#bruce

I don't understand why people don't test positions in modern opening
theory. Maybe they do, but they keep quiet about it? :-) My reasons
for thinking this would be a good idea:

1) Many positions have been well-explored by humans in the last 40
years. Definate judgements have taken place in many cases (although
in some it is true that people have simply moved on due to the whims
of fashion without reaching a verdict). There are many positions
between moves 10 and 20 today which have had extensive analysis
devoted to them. Programs which attempt to be more 'intelligent' in
their search can especially benefit, as their quiescence search should
be compared with how grandmasters have continued in real games, and
how their analysis runs in books. (Why does my program not consider
this ! move in the quiescance search? Does my program end its
q-search too early in this position? Not early enough? The
grandmaster just said "26. Be5 and White wins." Why is this not clear
to my program?)

2) Being able to play these positions well is correlated with playing
chess well. It is well and good to create a large opening book, but
this isn't really sufficient. Players need to know how to conduct a
game when a non-standard, but not obviously bad, move is played. The
program needs to be able to play in the correct manner, whether that
means a small material investment to open lines on the kingside, or
ensuring that minor pieces are not traded and that their isolated pawn
isn't blockaded. I think if a program is able to correctly handle the
"transition from the opening to the middlegame" (TOM), as Ignacio
Marin refers to it, the program will be able to handle the entire
middlegame well. The benefit of testing these sorts of things with
positions from opening theory is that idea 1) applies, in other words
a decision as to whether the program is playing in a proper manner is
easy to make, because grandmasters have already pored over the
position.

3) Domain proficiency may be developed. It is very common in computer
chess for a program to avoid certain types of positions simply because
it does not understand them. Testing opening positions in precisely
the positions where it is weakest will do the most to improve its
overall standard of play. Sure, an opening book may be programmed
where such lines are avoided as much as possible, but often some
characteristics of positions that are avoided will be present in
positions that are played to. If a program could play the Black side
of the Hedgehog well, it probably would be better able to cope with
other positions where flexible piece placement and latent dynamism
compensate for a space disadvantage. A more well-rounded computer
program could actually play into these sorts of positions versus
computer opponents, and as it has already been programmed to deal with
such a situation in a proper manner, it would stand an excellent
chance of outplaying the opponent.

4) Theoretical novelties may be found and used in tournament play. If
a position has been misanalysed by human annotators, the machine will
achieve a better position out of the TOM than it often otherwise
would. As White, it might secure a clear edge, or as Black, solid
equality (or better!). This is more likely to occur in positions
which modern Grandmaters avoid for various reasons. (Examples would
be the King's Gambit, because White has safer ways of playing for the
advantage, or the Yugoslav Dragon or Poisoned Pawn Najdorf, simply
because many Grandmasters don't deeply study positions they will rarely
reach.)

5) Return on the time invested is both short-term and long-term.
Short term, because a program will immediately play better chess in
the kind of positions that you happen to be working with. Work with
the King's Indian Defence will help the machine in other positions
with similar features (for instance it is common for Black to place a
bishop on g7 and play ...f5 in some variations of the Ruy Lopez.)
Long-term, because as the program understands more and more types of
positions, it can choose freely between them as the situation demands.
Your tournament opening book becomes a lot wider. When faced with the
choice between a promising attack and a secure positional plus it can
select based on the objective merit of the specific position, as
opposed to the program's predilection for one course or the other
based upon the author's 0.25 pawn suggestion to take one course or
another "because it can't play that other type of position well". I
understand that 'teaching it to play some sorts of positions well
might cause it to play other sorts of positions not as well'.. my
opinion is that with a good deal of work, a program can be taught to
understand what the essential features of a position are.

"You seem to think it's a pretty easy thing to do. Why don't you try
this yourself, and get back to us!" Yes, I'd like that very much. If
I get to graduate school, I probably will give a computer chess
program a good shot. Until then, I am more concerned with getting
through my undergraduate degree (in Computer Science) and paying the
bills. I don't think it's an easy thing to do at all, but I think
it's one important way in which computer chess can progress. I do
understand that in some ways this sort of tuning would be more
difficult with a fast, deep searcher than with a slower one which
thinks more at each node, but it's possible that in many cases what's
really needed is adjustment of some coefficients to terms, so even a
fast searcher can be improved. Piece-square tables come to mind, but
that's not really what I'm thinking of as I write this message...
perhaps they are a first step in some positions with immobile pawns,
but are almost certainly inadequate in the general case.

Anyway, no doubt there are some disadvantages or implausibilities to
my idea which you are itching to post, so I'll not keep you further. ;-)

Dave Gomboc (CFC 2084)
drgo...@a.stu.athabascau.ca

Dave Gomboc

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

In article <58ov5f$5...@news00.btx.dtag.de>,

Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:

#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
#

#>In article <58704j$o...@news00.btx.dtag.de>,
#>Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:
#>#Title: SSDF Rating Listing ---- is cheating
#I snipped my post because of ease of reading.
#---------------------
#
#>As if it had something to do with you being German. More like it was
#>because the contents of posts of yours which are intellgible span the
#>range from vapid speculation to sinister disingenuity, crossing abject
#>stupidity and sheer bollocks en route. How many months do you have to
#>be a regular poster on the net before you're NOT a "newbie"? (Hint:
#>You've posted 188 messages to the rec.games.chess.* hierarchy since
#>September 1st, and probably quite a few more before then.)
#
#Yes, I'm in that business from 1955 on. And wrote zillions of posts. :)
#
#Dave, I stopped claiming newbie status some weeks ago. But I'm still very
#new here compared to others, no?
#
#I'd like to answer your commentary just because you write below some
#interesting stuff. Interesting because I normally write speculation,
#stupidity, bollocks. If I understood you right?
#
#Let me tell you one single argument. Normally you shouldn't begin a debate
#with such heavy verdicts simply because you made yourself sort of fool to
#discuss thereafter, no? So I'll give my answer but afterwards you should
#decide to search for better correspondence. If I rattled your equilibrance
#that much with my statment about SSDF, that you lost some needed abilities
#to read carefully, I apologize of having done this. It was not my
#intention.

You are suggesting that I reply to someone else's posts rather than
yours, because it would make for better correspondence? You are
probably right, but since I've started this reply, I might as well
finish it.

#Right at the beginning you jump in and your nose tipped the dust. Why that?
#The *German* thing WAS mentioned. I always read about german anti SSDF
#gang. Didn't you realize this?

You said that people flamed you and attacked you because you were
German, which from my view is just a load of bollocks. If you can
find it please quote it for me. Generally people base their
flaming/attacking on their opinion of the merits of the other person's
ideas.. I do not think I am an exception in this regard.

#---------------------
#>There's no cheating factor. He's simply stating the obvious (well,
#>apparently not obvious, since it needs to be explained to you): the
#>more games that are played in the SSDF system, the less that future
#>games will affect measurements are taken. Rebel 8.0's rating will
#>continue to shift with all the rest, as games continue to be played.
#>(Exercise: demonstrate to yourself that the rating of a program on the
#>SSDF list can move between lists despite not having any games played
#>between a list and the list previous to it.)
#
#First of all it would be very important if you would be pleased to define
#your notion *games*. That's important for my statement. But as you are so
#well educated I could freely follow that you understood the games as
#equally comparable from which *match* (important second notion) they might
#ever had come. Am I right? Well then my commentary is again that that is
#the first serious error in SSDF listing/statistics. And the second
#therefore is that the matches are in reality of SSDF testings NOT defined.
#You see 1-0 matches and surely a lot of 20 games matches.

Here's where you and I have a big disagreement. *Game* is the
important unit here, not *Match*. When two programs have played 20
games against each other, they have done exactly that - they have
played "a game", "20 times". There is a big difference between this
and the human concept of a 20-game match. None of the multi-game
dynamics ("force a draw this time, I'll win tomorrow with White")
exist between two computer opponents. They don't (yet) even recognise
one opponent from the next. The SSDF list is based on games. The
"match" summary is just a convenient way to package results into a
small space, because humans are interested in 'how did it do against
xxx'?

#Second, I want to point out that it is cheating if you test rather
#unimportent matches in hundreds and the very interesting ones are almost
#hidden in this mass. Look e.g. at the result of Genius4 vs Rebel8. These
#two are concurrents. Not Comet or whatelse. Sorry I play with Comet, I like
#it, but it is not in the same class.

Why is it unfair to test Rebel 8 against the same computers that
Genius 4, MChess Pro 5, etc. were tested against? MChess Pro 6 is
being tested against the same group as well. So will Genius 5 and
Hiarcs 5. In addition, they will presumably be tested against each
other, because they will presumably all be near the top. It is best
to test against as many programs as possible that are within 150
points or so in either direction. (Humans would be 300 points, but
computer vs. computer ratings tend to predict better than human vs.
human ratings, because there are less variable factors from game to
game.)

#--------------------

#>60 points ahead of the nearest competitor *on the same hardware* was
#>big progress at a time when all the top program were within 10 points
#>of each other. (It's only 20 points ahead now, with MChess 6.0 on the
#>new list.) There's always "congratulations" from the chairman when
#>there's a new number one. You're trying to insinuate bias then there
#>isn't any.
#

#Nope, friendly Dave. Don't fuzzle my words to the invisible. You're not
#Spock, no?
#It's not the 60 points. I agree with you if each winner gets
#congretulations. But I will prove this. I assume that you're wrong. They
#didn't applaude each time with such emphazis.
#But my point was another one.
#Karlsson wrote about *chess computer wisdom*. And that is at this very
#moment without tests between the real concurrents simple cheating. Period.

You have a problem with 'concurrent' above. 'Concurrent' means
(roughly) taking place at the same time. As of October 1st, how can
you claim that Genius 5 is concurrent with Rebel 8? You can't,
because they aren't. Rebel 8.0 at that time was concurrent with
Genius 4, Hiarcs 4, etc. Two months from now things will be
different, and if Rebel 8.0 isn't tested a fair bit against other new
programs you can complain then. I can tell you now that you won't
need to, because SSDF _does_ test concurrent programs.

As for computer chess wisdom, I don't read anything more into that
than "obviously Rebel 8.0 is doing something right, we haven't seen a
60-point gap in many years!" Do you?

#----------------------
#>Once technology eases people's lives, they seldom want to "go back to
#>the old way". I don't have the fortune to own a snow blower, but you
#>can bet that once I had it the snow shovel would stay in the garage.
#>Keeping in mind (the obvious idea) that it is easier to autotest than
#>to test manually, do you actually have a point to make here? All I
#>see is some moron taking shots at people who provide what amounts to a
#>public service because he doesn't have anything better to do.
#
#Hey, what's moron, my dicionary iseems so empty ...

(roughly) an adult with the mental development of an 8-12 year old.
All right, I could have given you credit for 16 or so. :-)

#I thought we had passed this first claim of the amateur tester in swedish
#woods who was insulted by talking about his master instrument - the
#autoplayer. You seem completely overlook today's questioning of details of
#that autoplayer. Read Ed. Read Bert. Period.

The only period here is that you don't both to reply to my contention
that it is perfectly acceptable for the SSDF to test software without
autoplayer at a lesser rate due to practical time and financial
constraints. Perhaps you would like to show me where Ed and Bert
disagreed with this. There is a big difference between "unfortunately
you can only play 200 games with program A while you can do 500 games
with program B, so you'll have to wait another couple of months before
we get 200 more" and "it is wrong to do only 200 for this while doing
500 for that! you must do exactly the same number of each or it is
not fair".

#-----------------------
#>They post margins of error which reflect the number of games played by
#>each hardware and software combination. Why that is not good enough
#>for you is a mystery to me. Your "fundamental base of statistics" as
#>written is some of that bollocks I mentioned earlier.
#
#Yes, as concerning your *games*, read above.

I'm glad you don't disagree.

#------------------------
#>Why don't you buy a couple of these machines and test some games.
#>You'll find that it may or may not matter as much as 10 points. In
#>fact, why don't you just go and add the 10 points right now, just to
#>make yourself happy? Do you really know there's such a big difference
#>between 68% against average rating A and 64% against average rating B
#>which is close to A? Back your argument with arithmetic. Skip the
#>excess verbiage, it's tedious to read through.
#
#What do you like most, me buying some machines and test or going back to
#school or ameliorate autoplayer or anwering you at the moment?

I thought I had made it rather clear that if you didn't reply to me I
wouldn't miss you. I guess that makes my preference one of the other
choices. Maybe option three sounds best.. if you program a windows
autoplayer that new windows software can use then we'll hear less
complaining from anyone else who feels that there is some deep
conspiracy to not play as many games against strong software with
windows software so that program xxx can be at the top.

#-----------------------
#>Does this have something to do with chess? Or the SSDF?
#
#You refered my PORSCHE dingdong? Yes, it has something to do with testing
#procedures. Didn't you read that as such? Don't you like horses?
#No serious, you're trying to astonish me.

The Porsche/horse nonsense was exactly that.

#-----------------------
#>Oh, it was a stupid analogy, with an attempt to be comic near the end.
#>I see. Ha ha. Real funny, that. It would be nice if the win
#>percentage was closer to 50%, but statistical aberrations such as
#>these are bound to happen once in a while. Indeed, I would be
#>suspicous if they did not.
#
#If I didn't know I write to Dave I would suppose MB is on the phone. :)
#Or did you translate some passages?
#But serious. You didn't get the point again.
#If I would detect such results I would continue testing those special
#machines. Didn't you?

Why would you continue testing those special results? Sounds like
bias where it isn't warranted. Test everything, those 'special
results' will get updated over time too.

#----------------------
#>You may or may not be trying to say something of relevance in the
#>above paragraph. Don't bother clarifying if you don't feel like it.
#>In fact, you don't even have to bother posting, I won't miss you.
#
#Dave, but it's so simple, just kill my posts. *I* as new -- sorry Dave,
#really -- I never do this because I like all writings of people from
#everywhere. :)

Was "*I* as new" an aborted attempt to claim newbie status once again? :)

>----------------------
#>SSDF tests new versions when they get them. By March's list, you'll
#>be able to see how Rebel 8.0, M-Chess 6.0, Chess Genius 5.0, Hiarcs
#>5.0, and any other new programs do, with a lot of interplay between
#>them in particular. Why would you want to delay publication of
#>results of games? If a programmer beats competitors to market by a
#>couple months, why shouldn't his product appear sooner? Perhaps
#>another programmer decides to delay for a bit and continue to improve
#>his software, eventually yielding a higher-rated product than it would
#>have been had he published it sooner. Why should you not begin
#>testing with the released product? Consumers are especially
#>interested in the latest software available. If you want to see how
#>the top programs do against only each other it is very easy to pull
#>this information out of the SSDF list, which includes a result
#>breakdown by opponent. Really independent testers test what they've
#>got when they get it.
#
#Dave, why, when, would, if. You're really on the slope by now.
#Surely they are alowed to do this. They could do whatever they want.
#Problem remains, if statistically that would be sound. And my point was
#that it was NOT. But we're at the end of our debate. And you still didn't
#mention notion games, notion match, how many games, what pairings and so
#on.
#But that wasn't your favorite intention I see. But *bollock* notion as
#verdict is written much more easily. I understand. You did your job. But
#you've failed. Perhaps we'd meet in another thread. As I announced I
#wouldn't answer you again if you continue in that style, ok? Couldn't we
#change that? I'm still not convinced that I couldn't learn something of
#you.

As I mentioned in another post, the real reason I was so angry when I
replied the first time was because of your claim that people attacked
you because you are German. I think my characterization applied
pretty well to the series of 75+ posts aimed at baiting Ed Schroder
which I reviewed before I posted.

In my opinion you have a misguided notion of statistics, so how am I
supposed to convince you of their soundness? I don't believe in
deliberately not playing games because 'oh it is too good it will win
easily', I don't believe in deliberately playing games because 'oh
this program will give it a hard time, play it some more'. The bias
for selecting opponents right now is largely based on "what's
available?" and "who wants to play what against what?", which is in my
opinion a lot fairer than "okay, we must play 20 games against x, 20
games against y, 10 games against d, 10 games against e, etc." simply
because this latter way of choice is influenced too much by a
program's current results. You're never going to get rid of bias
entirely but introducing a huge amount of it by restricting interplay
amongst new software and software 6-12 months older would only hinder
the situation. Why should a new program be tested only, or even
largely, against just two or three others that are soon to be
released? It should be tested against many programs. If you want to
see how it does in relation to those particular 2 or 3 opponents, you
can always check the "match" result.. there is no need to prevent the
software from playing other opponents.

Sheesh, MChess Pro 6.0 played 20 games against Comet, just like Rebel
8.0. You might think that there's no important information in the
fact that Rebel drubbed it 19.5-0.5, while MChess settled for
12.5-7.5, but you'd be wrong. Or you might think that it makes a 40
point rating difference because it drives up Rebel's winning
percentage, and you'd be wrong. I don't really know what you think,
other than that you don't think that Rebel should have played Comet.
With the combined score of 32-8 for the two programs vs Comet it's
clear that it is not in their league, but it is not out of the realm
of possibility that Comet could win a game against them.

#Rolf Tueschen
#
#----*Man is unable not to know what he knows.* Leibowitz----

Dave Gomboc
drgo...@a.stu.athabascau.ca

Dave Gomboc

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

In article <E2HB4...@news.prima.ruhr.de>,

mclane <mcl...@prima.ruhr.de> wrote:
#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

#>#>Once technology eases people's lives, they seldom want to "go back to
#>#>the old way". I don't have the fortune to own a snow blower, but you
#>#>can bet that once I had it the snow shovel would stay in the garage.
#>#>Keeping in mind (the obvious idea) that it is easier to autotest than
#>#>to test manually, do you actually have a point to make here? All I
#>#>see is some moron taking shots at people who provide what amounts to a
#>#>public service because he doesn't have anything better to do.

#>#
#>#Come on. Thats unfair.

#
#>Actually, I thought that calling people racist because they disagree
#>with you was unfair. In fact, it was what pissed me off and caused me
#>to write my earlier response in the first place.
#
#I have not called anybody a racist because of the reasons you
#mentioned.
#Chris called me a fascists/racist because I said that this is
#APE-chess and he said that APE is a sign that I am thinking british or
#austrians are apes.
#But: if anybody who uses the word APE for something is a racists.....
#
#So maybe you mix up what I said and what not.

I was referring to something that Rolf said, not something you said.

#>#>They post margins of error which reflect the number of games played by

#>#>each hardware and software combination. Why that is not good enough
#>#>for you is a mystery to me. Your "fundamental base of statistics" as
#>#>written is some of that bollocks I mentioned earlier.

#>#
#>#Statistic is not a science. It is a religion.

#>#There are no FUNDAMENTAL BASE OF STATISTICS.
#>#They are as wrong as the idea that forces influence material parts so
#>#that the apple falls down the earth.
#>#The apple is nbot falling because there is a force called gravitation.
#>#And there is no fundamental base of .... that RT has not understand.
#>#Your statement is a prejudice against RT. You hide it because you
#>#don't say: Rolf - you are not well-educated enough to understand.
#>#But instead you say: you don't have the fundamental Base of
#>#statistics.
#
#>You really feel that statistics is a religion, and not a science?
#>I guess you're entitled to your opinion. In my view, statistics CAN be
#>done rigoursly, and often is.
#
#Science is a religion itself.
#There is no objectivity. You can only try to be very near, within a
#certain range, called error-margin.
#But still you will never reach the truth.
#
#In my opinion statistics is not to use to find out what will happen,
#but to interprete why something has happened before.
#So statistics do not show us anything that is not known before.
#As an example see ssdf-list. They do not show me anything I do not
#know months before. But they proof my ideas very well.
#If you believe that statistics shows us something that has not
#happened, then you could do the same influence on the future by going
#into a church and burn a candle-light. It would have the same effect.

It is interesting that you are okay with descriptive statistics but
not with inferential statistics.

#>#

#>#Thats prejudicel. Not more. Not less.

#>#Without any attack: you should read a book about
#>#latest partical-physics and mathzematical-theorems instead of praying
#>#ideas of closed statistical-rule-bases. Because these fundamental
#>#stuff is out of an old paradigm called classisistic science.
#>#This war was lost a couple of years before Einstein found out his
#>#theories of relativity. This is a couple of years ago. 90 or 100 years
#>#ago. So YOU are living in the wrong century, not RT.
#>#All these mechanical ideas are wrong since the time Einstein made hs
#>#first attack on them.
#
#>I think you've wandered off somewhere above...
#Off- topic, thats for sure.
#But the theories of Einstein are still amazingly important to our
#paradigm.
#
#>#I know that this has not much to do with chess, but your attacks that
#>#RT has no fundamental base of statistics has also nothing to do with
#>#chess. Again: thats a prejudice.

#>#
#>#>Does this have something to do with chess? Or the SSDF?
#>#
#>#Do you want to force him what he should say and what not ?

#>#"I will fight for the right that you can say what you want" or how was
#>#the sentence ?!
#>#
#>#

#>#>Oh, it was a stupid analogy, with an attempt to be comic near the end.

#>#>I see. Ha ha. Real funny, that. It would be nice if the win
#>#>percentage was closer to 50%, but statistical aberrations such as
#>#>these are bound to happen once in a while. Indeed, I would be
#>#>suspicous if they did not.
#>#
#>#

#>#>You may or may not be trying to say something of relevance in the

#>#>above paragraph. Don't bother clarifying if you don't feel like it.
#>#>In fact, you don't even have to bother posting, I won't miss you.
#
#>..,here you are returning to 'earth' :-)
#
#>#Nice language. Not colloquial but also not very friendly.
#>#But now I undersand what politeness is.
#>#Say somebody is shit without saying it!!
#>#Very polite.
#
#>Actually I thought it was rather impolite.
#It was impollite, but still harmless in relation to my own postings.
#
#>I am not saying that
#>someone is 'shit', as you put it. I am genuinely unclear as to
#>whether he was trying to say something of importance in that paragraph
#>or not!
#
#Right. That was my point. You are GENUINELY UNCLEAR.
#Thats what I always say: some people here are IMPOLITE but in the same
#way they hide their opinion with DISCLAIMERS and behave like the
#Heisenbergsche Unschaerferelation.

Clever. I'll restate: "I do not know whether he was trying to say
something of importantance in that paragraph or not". And what is
'Heisenbergsche Unschaerferelation'?

#> As for the "don't even ... bother posting", that is more a
#>reflection of the high esteem I hold of the series of posts spanning a
#>two-month period (so far!) by Rolf Tueschen which attempted to drag
#>Ed Schroder through the mud than it pertains to the SSDF discussion,
#>though the threads do cross from time to time.
#
#I confirm.
#
#>#
#>#>Dave Gomboc
#>#>drgo...@a.stu.athabascau.ca
#>#

#>#You are right- we are 1/2 or even more advance in time. When I wrote

#>#my Rebel8 article in AUGUST, I had Rebel8 before and tested it.
#>#So now in November they discuss what I forgot in august after have
#>#written down the article.

Dave Gomboc
drgo...@a.stu.athabascau.ca

Bill Newton

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
<chr...@cpsoft.demon.co.uk> writes

...snip...

>I'll reply for Ed.
>
>He is too polite to say what he really thinks.
>
>FUCK-OFF.

Tis a great pity Chris that YOU'RE not.... know what I mean?

Too often you send page after page repeating other posts ad nauseam, to
which you add one or two lines of what you obviously consider to be
comments of succinct wit.

Believe me they are not, although they may be halfway there.

The snipped posting above is a particularly odious example of your style
and does you no credit whatsoever.

However your knowledge of chess computing in general is undeniable and
you have in the past produced many informed postings.

Accordingly I sincerely hope that in future you will continue with your
'informed' postings and leave witty comments for folk with the talent
and ability to express themselves in language suitable for a public
forum.

Regards.

--
Bill Newton

en...@lix.intercom.es

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

In article <jgyp$KAd3j...@demon.co.uk>,

Bill Newton <not...@demon.co.uk> wrote:
>
> In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
> <chr...@cpsoft.demon.co.uk> writes
>
> ...snip...
>

> Too often you send page after page repeating other posts ad nauseam, to
> which you add one or two lines of what you obviously consider to be
> comments of succinct wit.
>
> Believe me they are not, although they may be halfway there.

Chris's postings arrive at my server between midnight and 2 am, and he is
responsible for me being short of sleep every day. I am addicted to them,
often making me laugh to tears. Chris: don't
change.

Enrique
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet

Peter Fendrich

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

Tim Mirabile wrote:

- snip -

> Learning should probably be disabled during testing, but if the programs are
> released with info already in the learning files, this should be allowed to be
> used. Essentially you want it to play each game as if it were the first one out
> of the shrink wrap.
>

> --
> Tim Mirabile <t...@mail.htp.com> - http://www.webcom.com/timm/
> Visit my homepage for information on USCF & FIDE rated chess on Long Island.
> TimM on the Free Internet Chess Server - telnet://fics.onenet.net:5000/
> ICD/Your Move Chess & Games - http://www.icdchess.com/
> The opinions of my employers are not necessarily mine, and vice versa.

That's one way of doing it. Then you will not measure the strength of
the
learning feature itself. If the learning is more or less the same for
all
programs this approach wouldn't miss much.
I have a fealing that even theese features will vary...
Learning will be part of the programs strength.

--
J-P Fendrich

Harald Faber

unread,

Dec 17, 1996, 3:00:00 AM12/17/96

to

Hello Rolf,

I think I have an interesting aspect to tell you:

Snipped interesting story

RWT> A real serious non cheating staff would have said, look Eddie, thatAs
RWT> nice, but weAll wait till chrismas. Then we could match Reb8 against
RWT> Genius5, Mchess6, HIARCS5 and so on. Because we are really independent
RWT> testers. What would it help you if you could tell the world youAve beaten
RWT> all those oldtimers? Wer zuletzt lacht, lacht am besten. No?

You cannot expect that there mustn't be tests before te other new versions
are on market. Forget about it, you can only test with programs which are
available, so what is wrong with this?

BTW: Rebel8 is no Porsche relating to your oldtimer-comparison
What if someone invents a car which is cheaper and much better than all
available cars, does he compete with oldtimers, too? No is the right
answer...

Ciao and see ya
Harald
--

mclane

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

Bill Newton <not...@demon.co.uk> wrote:

>In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
><chr...@cpsoft.demon.co.uk> writes

>...snip...

>>I'll reply for Ed.

>>
>>He is too polite to say what he really thinks.
>>
>>FUCK-OFF.

>Tis a great pity Chris that YOU'RE not.... know what I mean?

>Too often you send page after page repeating other posts ad nauseam, to
>which you add one or two lines of what you obviously consider to be
>comments of succinct wit.

I like Chris comments. He has a good overview about the market, the
scene, chess programming, and other important things.

But of course - I am not unbiased.

>Believe me they are not, although they may be halfway there.

>The snipped posting above is a particularly odious example of your style

>and does you no credit whatsoever.

>However your knowledge of chess computing in general is undeniable and
>you have in the past produced many informed postings.

Yes.

>Accordingly I sincerely hope that in future you will continue with your
>'informed' postings and leave witty comments for folk with the talent
>and ability to express themselves in language suitable for a public
>forum.

Hm. These are not nice * regards *

>Regards.

Wouldn't it be better to say something else instead of double-moral
REGARDS ?!
>--
>Bill Newton

Maybe: use BYE BYE.

mclane

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

en...@lix.intercom.es wrote:

>Chris's postings arrive at my server between midnight and 2 am, and he is
>responsible for me being short of sleep every day.

Yes, like all britains he has a nice black humor.
Monty Python is nothing against him.

mclane

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>In article <E2HB4...@news.prima.ruhr.de>,
>mclane <mcl...@prima.ruhr.de> wrote:
>#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

>#So maybe you mix up what I said and what not.

>I was referring to something that Rolf said, not something you said.

Sorry - so I have mixed up something. Sorry.

Could you comment on this, I don't see the point due to language
problems or whatever (it is again 2h a.m. in germany!)

snip...

>#>I am not saying that
>#>someone is 'shit', as you put it. I am genuinely unclear as to
>#>whether he was trying to say something of importance in that paragraph
>#>or not!
>#
>#Right. That was my point. You are GENUINELY UNCLEAR.
>#Thats what I always say: some people here are IMPOLITE but in the same
>#way they hide their opinion with DISCLAIMERS and behave like the
>#Heisenbergsche Unschaerferelation.

>Clever. I'll restate: "I do not know whether he was trying to say
>something of importantance in that paragraph or not". And what is
>'Heisenbergsche Unschaerferelation'?

It says that you cannot say anything about anything without changing
it due to the try to find out about it:
Individual subatomar particles cannot be foreseen.
You can only speak in terms of groups or in statistics about them.
And there we have this damned statistics again!
The german physician Werner Heisenberg was the inventor of this
Unschärferelation (I don't know the english term in the moment).
1927 he started the philosophical debate , because in other words his
theory said that the conciousness influences the materia DIRECTLY
without using forces.
The physician is no longer the observer, but the participant of the
world, so these empirical physicians are now like us:
only participants!
Welcome in the club.

I wanted to say that you are as indistinct like particles in their
sphreres.

I know that you said that you don't know if his posting was important
or not, but this is a very nice saying that it was not important.
Isn't it.

What I mean is: there are many ways not to say something impolite but
to say something that is also not nice and anybody understand : this
is an attack.

>Dave Gomboc
>drgo...@a.stu.athabascau.ca

Chris and I like more the DIRECT speech. I read that Ed also prefers
this language.

Bill Newton

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
<chr...@cpsoft.demon.co.uk> writes

...snip...

>I'll reply for Ed.
>
>He is too polite to say what he really thinks.
>
>FUCK-OFF.

Tis a great pity Chris that YOU'RE not.... know what I mean?

Too often you send page after page repeating other posts ad nauseam, to
which you add one or two lines of what you obviously consider to be
comments of succinct wit.

Believe me they are not, although they may be halfway there.

The snipped posting above is a particularly odious example of your style
and does you no credit whatsoever.

However your knowledge of chess computing in general is undeniable and
you have in the past produced many informed postings.

Accordingly I sincerely hope that in future you will continue with your

'informed' postings and leave witty comments for folk with the talent
and ability to express themselves in language suitable for a public
forum.

Regards.

--
Bill Newton

Chris Whittington

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

en...@lix.intercom.es wrote:
>
> In article <jgyp$KAd3j...@demon.co.uk>,
> Bill Newton <not...@demon.co.uk> wrote:
> >

> > In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
> > <chr...@cpsoft.demon.co.uk> writes
> >
> > ...snip...
> >

> > Too often you send page after page repeating other posts ad nauseam, to
> > which you add one or two lines of what you obviously consider to be
> > comments of succinct wit.
> >
> > Believe me they are not, although they may be halfway there.
>

> Chris's postings arrive at my server between midnight and 2 am, and he is

> responsible for me being short of sleep every day. I am addicted to them,

> often making me laugh to tears. Chris: don't
> change.

No chance of that :)

Chris Whittington

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

en...@lix.intercom.es wrote:
>
> In article <jgyp$KAd3j...@demon.co.uk>,
> Bill Newton <not...@demon.co.uk> wrote:
> >
> > In article <85065876...@cpsoft.demon.co.uk>, Chris Whittington
> > <chr...@cpsoft.demon.co.uk> writes
> >
> > ...snip...
> >
> > Too often you send page after page repeating other posts ad nauseam, to
> > which you add one or two lines of what you obviously consider to be
> > comments of succinct wit.
> >
> > Believe me they are not, although they may be halfway there.
>
> Chris's postings arrive at my server between midnight and 2 am, and he is
> responsible for me being short of sleep every day. I am addicted to them,
> often making me laugh to tears. Chris: don't
> change.

No chance of that :)

Only if my wife gets her way and removes my bollocks :(

graham_douglass

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

In article <32B2EE...@nwlink.com>, brucemo says...

>
>a bean wrote:
>
>> | 2. Solving certain tactical positions quickly may be a good

>> | indicator of speed chess ability but may not translate

>> | well into 40/2 games.

>>
>> Exactly why we need this - the purpose of a CAT test is to determine the

>> relative playing strengths for various time controls and positional types.
>

>My provider has decided to implement one of these "you must have more new

>stuff than quoted stuff", so I can't quote some of the parts of this that I

>want to respond to, since I don't intend to write a heck of a lot.

This is very bad. To avoid too much reading, I like to look at the end of a
thread, and pick up the conversation from reading one post only. The "more new
than quoted" rule is crass stupidity with, as far as I can see, no redeeming
benefits.

>
>For determining the strength of a program, you can't beat playing a real

>game. There are numerous problems with test suites.
>

>You can complain about hidden agendas of the suite builder if a program you

>favor performs poorly. You might have a point, maybe the suite builder does

>want to favor another program, and has tuned the suite for that program. On

>the other hand, perhaps you are wrong. In either case, I predict a huge

>angry thread in here.
>

>It would be difficult to make a suite that conforms perfectly to the demands

>of the average game. The tendency is to put in positions that probably won't

>come up in practical play. One of the suites has a Q vs P ending, where one

>the superficially most obvious move draws, while another move wins. If you

>have a tablebase you get this right, if you don't, you probably get it right

>but maybe you don't. In practical play this might come up once in 500 games,

>but in this suite it counts for like 1/30 of your total score. There's

>another one I've seen that's designed to trip up null move programs in

>endings. The position is very much contrived, it is not likely to happen in

>practical play. Many of the suites have examples of brilliant attacks, but

>in human vs computer play the computer is more often defending against wild

>attacks.

>
>Just getting the right proportion of opening vs middlegame vs ending

>positions in a suite would be hard.
>

>You'd have to be careful to make very certain that a program did something

>for the right reason, and this might be kind of hard. For instance, if the

>point of a positional problem is to know that the bishop pair is worth more

>than giving the opponent a doubled pawn, how do you determine that the

>program knows that the bishop pair is worth more than a doubled pawn, rather

>than that it doesn't know that a doubled pawn is bad in any case?
>

>And of course, there are suites where the "best move" is really a matter of

>taste.

>
>So I be more prone to trust the results of real games, which is not to say

>there aren't problems with THAT method, too.

It is often said that nothing less than a REAL war sorts out the good generals
from the bad ones.

The SSDF list is the nearest thing we have to a real war.

Bad things happen in real wars. The side you expect to win doesn't always win.

People do things they're not supposed to. For example, in the 1940 Battle of
Britain, the best historical evidence is that the carpet bombing of cities
started by accident (the UK bombed Berlin in retaliation for the accidental
bombing of a UK city by a German plane. After that, Germany stopped the
successful tactic of bombing British air bases, and started bombing UK cities
instead - a tactic that certainly saved the RAF).

You might regard "cooked opening books" as a bad thing - but it is up to you to
find a way of defending.

You see, the further you get from a "real war", the less you are really learning
about the real relative strengths of the programs in the real world.

The SSDF is a good list - the only way it could be improved is by including
humans in the list as well! This is done, to a small extent, in Eric Hallsworth's
list.

>
>bruce

Bill Newton

unread,

Dec 18, 1996, 3:00:00 AM12/18/96

to

In article <8508306...@dejanews.com>, en...@lix.intercom.es writes

>Chris's postings arrive at my server between midnight and 2 am, and he is
>responsible for me being short of sleep every day. I am addicted to them,
>often making me laugh to tears. Chris: don't
>change.

I fear lack of sleep is telling on you if a vulgar two word obscenity
such as Chris posted can make you 'laugh to tears'.

All I can say is you have my sympathy and my sincere hope that you
acquire a more adult sense of humour as you grow up.

--
Bill Newton

Churak

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <f48mZNAl...@demon.co.uk>,
Bill Newton <not...@demon.co.uk> rambled at Chris:

>Accordingly I sincerely hope that in future you will continue with your
>'informed' postings and leave witty comments for folk with the talent
>and ability to express themselves in language suitable for a public
>forum.

I strongly disagree. Chris' posts are a hoot, even when I disagree with what
he says.

I *do* agree, though, that Chris needs to learn how to "snip". :-)

-- Steve Lopez

=============================================
lummra khatunikh churak himrukkal gual.
lummra dlanmukoi hiweshma tahen gual.
lum surimtokoi gual.
lum brufenul tsulajun muni.
tusmriremra dlanmukoi lakun ssiya!
=============================================

Churak

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <$Ayu3GAR...@demon.co.uk>,
Bill Newton <not...@demon.co.uk> wrote:

>All I can say is you have my sympathy and my sincere hope that you
>acquire a more adult sense of humour as you grow up.

If you're going to hold your nose so high in the air, Bill, please do us all a
favor and at least blow it once in a while. Nobody wants to look at *that*!

Chris Whittington

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

mcl...@prima.ruhr.de (mclane) wrote:
>
> drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
>

> >In article <E2HB4...@news.prima.ruhr.de>,
> >mclane <mcl...@prima.ruhr.de> wrote:
> >#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
>
> >#So maybe you mix up what I said and what not.
>
> >I was referring to something that Rolf said, not something you said.
>

> Sorry - so I have mixed up something. Sorry.
>

> Could you comment on this, I don't see the point due to language
> problems or whatever (it is again 2h a.m. in germany!)
>
> snip...
>

> >#>I am not saying that
> >#>someone is 'shit', as you put it. I am genuinely unclear as to
> >#>whether he was trying to say something of importance in that paragraph
> >#>or not!
> >#
> >#Right. That was my point. You are GENUINELY UNCLEAR.
> >#Thats what I always say: some people here are IMPOLITE but in the same
> >#way they hide their opinion with DISCLAIMERS and behave like the
> >#Heisenbergsche Unschaerferelation.
>
> >Clever. I'll restate: "I do not know whether he was trying to say
> >something of importantance in that paragraph or not". And what is
> >'Heisenbergsche Unschaerferelation'?

> It says that you cannot say anything about anything without changing
> it due to the try to find out about it:
> Individual subatomar particles cannot be foreseen.
> You can only speak in terms of groups or in statistics about them.
> And there we have this damned statistics again!
> The german physician Werner Heisenberg was the inventor of this
> Unschärferelation (I don't know the english term in the moment).

Heisenberg Uncertainty Principle.

There are many formulations. One simple to understand usage is the
idea that you cant measure something without changing the thing
you are measuring.

Eg. bounce a photon of a small particle in order to see the particle,
and the particle goes flying off in another direction.

Or stick a thermometer in something to measure its temperature and
the temperature of the something is affected by the different
start temperature of the thermometer. etc. etc. etc.

As Thorsten points out, this principle of physics joined the
philosophical debate that there was no such thing as a neutral
observer. The observer affects and becomes a part of the thing he
is observing.

Schrodinger's cat is a thought experiment relating to this.
The cat in the box is in an undetermined state. It is the act of
observing it which kills it or leaves it alive.

Or take the SSDF. They measure chess program ratings. But now the
existence of the SSDF rating list has affected the programs
themselves ....... :)

Chris Whittington

Bill Newton

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <f48mZNAl...@demon.co.uk>, Bill Newton

<not...@demon.co.uk> writes

>>I'll reply for Ed.
>>
>>He is too polite to say what he really thinks.
>>
>>FUCK-OFF.
>
>Tis a great pity Chris that YOU'RE not.... know what I mean?

...snip...

Sorry about 'repeated' postings appearing from me recently. Not intended
and I dont know why its happening!

Regards.

--
Bill Newton

Bill Newton

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <8508306...@dejanews.com>, en...@lix.intercom.es writes
>Chris's postings arrive at my server between midnight and 2 am, and he is
>responsible for me being short of sleep every day. I am addicted to them,
>often making me laugh to tears. Chris: don't
>change.

I fear lack of sleep is telling on you if a vulgar two word obscenity
such as Chris posted can make you 'laugh to tears'.

All I can say is you have my sympathy and my sincere hope that you

acquire a more adult sense of humour as you grow up.

--
Bill Newton

Enrique Irazoqui

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

>en...@lix.intercom.es writes
Bill Newton <not...@demon.co.uk> wrote

>>Chris's postings arrive at my server between midnight and 2 am, and he is
>>responsible for me being short of sleep every day. I am addicted to them,
>>often making me laugh to tears. Chris: don't
>>change.

>I fear lack of sleep is telling on you if a vulgar two word obscenity
>such as Chris posted can make you 'laugh to tears'.

Fear not! This "two word obscenity" didn't make me laugh. Just a big smile.
You have a reading problem, though. I said "addicted to them", plural.

>All I can say is you have my sympathy and my sincere hope that you
>acquire a more adult sense of humour as you grow up.

Your sympathy and your sincere hope are hypocritical, self-righteous and
empty-minded.

With all my sympathy, Enrique

>Bill Newton

graham_douglass

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <85092232...@cpsoft.demon.co.uk>, Chris says...
{snip}

>> Chris's postings arrive at my server between midnight and 2 am, and he is
>> responsible for me being short of sleep every day. I am addicted to them,
>> often making me laugh to tears. Chris: don't
>> change.
>

>No chance of that :)
>
>Only if my wife gets her way and removes my bollocks :(

Anything we can do to help her? :-)

>
>Chris Whittington

graham_douglass

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <E2L8J...@news.prima.ruhr.de>, mcl...@prima.ruhr.de says...
{snip}

>Yes, like all britains he has a nice black humor.
>Monty Python is nothing against him.

Monty Python is drivel.

Bill Newton

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

Bill Newton posted:

>Accordingly I sincerely hope that in future you will continue with your
>'informed' postings and leave witty comments for folk with the talent
>and ability to express themselves in language suitable for a public
>forum.

In article <59a4ga$j...@news3.texas.net>, Churak <bays...@intrepid.net>
writes

>I strongly disagree. Chris' posts are a hoot, even when I disagree with what
>he says.

Let me see if I'm understanding you correctly:

Chris Whittington posted:

>I'll reply for Ed.<

>He is too polite to say what he really thinks.<

>FUCK-OFF.<

You apparently consider this comment to be a 'hoot'

Perhaps you'll understand better when your acne fades! :-)

>I *do* agree, though, that Chris needs to learn how to "snip". :-)

Already your acne starts to fade! :-)

Hang looser, be happy.

Regards.

>

--
Bill Newton

Bill Newton

unread,

Dec 19, 1996, 3:00:00 AM12/19/96

to

In article <59a4lc$j...@news3.texas.net>, Churak <bays...@intrepid.net>
writes

>In article <$Ayu3GAR...@demon.co.uk>,
> Bill Newton <not...@demon.co.uk> wrote:
>

>>All I can say is you have my sympathy and my sincere hope that you
>>acquire a more adult sense of humour as you grow up.
>

>If you're going to hold your nose so high in the air, Bill, please do us all a
>favor and at least blow it once in a while. Nobody wants to look at *that*!
>
>-- Steve Lopez

I am assuming from your convoluted comment above that you condone the
use of obscene language in this public forum and are criticising me for
condemning the use of same!

Well, I'm not offering any apologies!

Incidentally I'm not as 'heavy' as you might think on this issue but I
do believe that it takes very little effort for contributors to make
their point without the use of obscene language.

But maybe I have to rest my case there for now and learn to accept that
anybody using obscene language in a public forum to make their point,
just lacks the necessary skills / sense of decency, to express their
view any other way :-)

Regards.

--
Bill Newton

mclane

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

Chris Whittington <chr...@cpsoft.demon.co.uk> wrote:

>mcl...@prima.ruhr.de (mclane) wrote:
>>
>> drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
>>

>Heisenberg Uncertainty Principle.

>There are many formulations. One simple to understand usage is the
>idea that you cant measure something without changing the thing
>you are measuring.

>Eg. bounce a photon of a small particle in order to see the particle,
>and the particle goes flying off in another direction.

>Or stick a thermometer in something to measure its temperature and
>the temperature of the something is affected by the different
>start temperature of the thermometer. etc. etc. etc.

>As Thorsten points out, this principle of physics joined the
>philosophical debate that there was no such thing as a neutral
>observer. The observer affects and becomes a part of the thing he
>is observing.

>Schrodinger's cat is a thought experiment relating to this.
>The cat in the box is in an undetermined state. It is the act of
>observing it which kills it or leaves it alive.

>Or take the SSDF. They measure chess program ratings. But now the
>existence of the SSDF rating list has affected the programs
>themselves ....... :)

That was exactly what I mean ! Thanks to my personal translator
into oxford-english.... :-)

Churak

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

In article <gEOp2ZAgwXuyEw0$@demon.co.uk>,

Bill Newton <not...@demon.co.uk> wrote:
>Bill Newton posted:
>
>>Accordingly I sincerely hope that in future you will continue with your
>>'informed' postings and leave witty comments for folk with the talent
>>and ability to express themselves in language suitable for a public
>>forum.
>
>In article <59a4ga$j...@news3.texas.net>, Churak <bays...@intrepid.net>
>writes
>
>>I strongly disagree. Chris' posts are a hoot, even when I disagree with what
>>he says.
>
>Let me see if I'm understanding you correctly:
>
>Chris Whittington posted:
>
>>I'll reply for Ed.<
>
>>He is too polite to say what he really thinks.<
>
>>FUCK-OFF.<
>
>You apparently consider this comment to be a 'hoot'

Your interpretation. Perhaps I should have said "I find MOST of Chris'
comments to be a hoot". Or perhaps an exact percentage, such as "I find 97.65%
of Chris' comments to be a hoot". The later may be too anal-retentive even for
you, but somehow I doubt it.

>
>Perhaps you'll understand better when your acne fades! :-)

I'm 36 years old with a clear complexion, but I do find both your presumption
and your implication humorous.

>
>>I *do* agree, though, that Chris needs to learn how to "snip". :-)
>
>Already your acne starts to fade! :-)
>
>Hang looser, be happy.

I grin frequently, especially when reading the rec.games.chess newsgroups. :-)

Churak

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

In article <MbMSRdAp...@demon.co.uk>,

Bill Newton <not...@demon.co.uk> wrote:
>In article <59a4lc$j...@news3.texas.net>, Churak <bays...@intrepid.net>
>writes

>>In article <$Ayu3GAR...@demon.co.uk>,

>> Bill Newton <not...@demon.co.uk> wrote:
>>
>>>All I can say is you have my sympathy and my sincere hope that you
>>>acquire a more adult sense of humour as you grow up.
>>
>>If you're going to hold your nose so high in the air, Bill, please do us all
a
>>favor and at least blow it once in a while. Nobody wants to look at *that*!
>>
>>-- Steve Lopez
>
>I am assuming from your convoluted comment above that you condone the
>use of obscene language in this public forum and are criticising me for
>condemning the use of same!

Nope. Just busting your chops.

>
>Well, I'm not offering any apologies!

Good. I wasn't expecting one.

>
>Incidentally I'm not as 'heavy' as you might think on this issue but I
>do believe that it takes very little effort for contributors to make
>their point without the use of obscene language.

No argument there. However, you *did* manage to repost his comment a couple of
times without any editing (such as strategically-placed asterisks); in my
book, that constitutes guilt by association. :-) It at least makes you as
guilty as Chris.

>
>But maybe I have to rest my case there for now and learn to accept that
>anybody using obscene language in a public forum to make their point,
>just lacks the necessary skills / sense of decency, to express their
>view any other way :-)

Possibly. It's also possible that the use of such language reflects a high
degree of frustration on the part of the poster. It's also possible that the
one using such language has taken the words of Gen. George S. Patton to heart:
"When I want it to stick, I give it to them hard and dirty".

Chris certainly got *your* attention. :-)

It's a large world. Anything's possible.

Bill Newton

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

In article <01bbedbe$0ef30f80$e9ec...@10.0.1.1>, Enrique Irazoqui
<en...@lix.intercom.es> writes

>Fear not! This "two word obscenity" didn't make me laugh. Just a big smile.
>You have a reading problem, though. I said "addicted to them", plural.

I'm a patient person Enrique so rather than criticising your impulsive
comment may I first of all suggest that you try very hard to grasp the
point of my original post.

For example.

The issue in question IS one of 'THEM' to which you say you are
addicted. O.K.?

Bill Newton posted:

>>All I can say is you have my sympathy and my sincere hope that you
>>acquire a more adult sense of humour as you grow up.

Enrique posted:

>
>Your sympathy and your sincere hope are hypocritical, self-righteous and
>empty-minded.

Nope! My comments are sincere. I'm simply stating that 'I' really would
like to see you acquire a more adult sense of humour as you grow up.
Wouldn't you?

>With all my sympathy, Enrique

Thats very kind of you. Thank you ;-)

Bye for now.

Regards.
--
Bill Newton

Rolf Czedzak

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

Chris Whittington wrote: <85092232...@cpsoft.demon.co.uk>

CW> Only if my wife gets her way and removes my bollocks :(

'Bollocks' is not in my (roman catholic) dictionary, so my best
wishes are with You. ;-))

CW> Chris Whittington

Rolf

Bill Newton

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

In article <59cm17$p...@news3.texas.net>, Churak <bays...@intrepid.net>
writes

>Nope. Just busting your chops.

I'm English Steve, not too sure what this phrase means. Is it
complimentary to me? :-)

>No argument there. However, you *did* manage to repost his comment a couple of
>times without any editing (such as strategically-placed asterisks); in my
>book, that constitutes guilt by association. :-) It at least makes you as
>guilty as Chris.

You find him guilty eh!? :-)

Seriously though I believe it was Impossible to make the point without
quoting Chris but I always made sure it was attributed to him!

>Possibly. It's also possible that the use of such language reflects a
>high degree of frustration on the part of the poster.

Yep, I'll buy that.

>It's also possible that the one using such language has taken the
>words of Gen. George S. Patton to heart:
>"When I want it to stick, I give it to them hard and dirty".

Hang on a minute... your quoting George S Patton .... wasn't he the
American General that beat up on an American soldier who was suffering
from battle fatigue? Hmnnnnn.....Was that giving it to him hard and
dirty?.... Did he make it stick? ....... Only joking, just my rhetorical
rambling but if you want to talk soldiers I'll stick with Montgomery :-)

>
>Chris certainly got *your* attention. :-)

Chris always gets my attention, he can produce good stuff when he puts
his mind to it..... if only he would ALWAYS put his mind to it.

>It's a large world. Anything's possible.

Well..... almost anything! :-)

Regards.

--
Bill Newton

Bill Newton

unread,

Dec 20, 1996, 3:00:00 AM12/20/96

to

In article <59ckr1$p...@news3.texas.net>, Churak <bays...@intrepid.net>
writes

>
>Your interpretation. Perhaps I should have said "I find MOST of Chris'
>comments to be a hoot".

Yep, it always helps if you post what you MEAN to say :-).

>Or perhaps an exact percentage, such as "I find 97.65%
>of Chris' comments to be a hoot". The later may be too anal-retentive even for
>you, but somehow I doubt it.

Tut tut..... sigh

>
>I'm 36 years old with a clear complexion, but I do find both your presumption
>and your implication humorous.

Well thank you, I'm pleased you saw the intended humorous side of things
as sometimes I'm too 'acid' for my own good but perhaps I'll learn to
mellow out as I grow up :-)

>I grin frequently, especially when reading the rec.games.chess newsgroups. :-)

I bet you do too, I know I often SQTM ( smile quietly to myself) when
reading comments. (That acronym will NEVER catch on!)

Enrique Irazoqui

unread,

Dec 21, 1996, 3:00:00 AM12/21/96

to

Bill Newton <not...@demon.co.uk> escribió en artículo
<LuxDMOAd...@demon.co.uk>...

> In article <01bbedbe$0ef30f80$e9ec...@10.0.1.1>, Enrique Irazoqui
> <en...@lix.intercom.es> writes
> >Fear not! This "two word obscenity" didn't make me laugh. Just a big
smile.
> >You have a reading problem, though. I said "addicted to them", plural.
>
> I'm a patient person Enrique so rather than criticising your impulsive
> comment may I first of all suggest that you try very hard to grasp the
> point of my original post.
>
> For example.
>
> The issue in question IS one of 'THEM' to which you say you are
> addicted. O.K.?

No. If you pay all the attention you are capable of, you may understand the
following. Chris tried to stop the postings of someone that was
continuously insulting Ed. It was tried before by several people and it
didn't work. This is the issue. Accusations and insults are offending, but
you choose to pick on Chris's "two word obscenity" and play the role of
moral vigilante. Mind your own sense of morals and humor and play
patronizing with your great grand children.

End of thread.

Enrique

mclane

unread,

Dec 22, 1996, 3:00:00 AM12/22/96

to

"Enrique Irazoqui" <en...@lix.intercom.es> wrote:

>No. If you pay all the attention you are capable of, you may understand the
>following. Chris tried to stop the postings of someone that was
>continuously insulting Ed. It was tried before by several people and it
>didn't work. This is the issue. Accusations and insults are offending, but
>you choose to pick on Chris's "two word obscenity" and play the role of
>moral vigilante. Mind your own sense of morals and humor and play
>patronizing with your great grand children.

>End of thread.

>Enrique
Hello Enrique,

have a nice week-end and a good Christmas-time...

I completely agree with you statement.

Moritz Berger

unread,

Dec 22, 1996, 3:00:00 AM12/22/96

to

On 14 Dec 1996 06:08:08 GMT, TUESCHEN.MEDIZ...@t-online.de
(Rolf W. Tueschen) wrote:
>Very funny and highest level for sure.
>
>Take that:
>
>Ed Schroder did violate massively the rules of etiquette of newsgroups.
Ed announced new products here. This is strictly chess related and
customary among almost all professional programers (Marty Hirsch,
Richard Lang, Mark Uniacke to name a few). By fighting personal wars
(nobody wants to listen to your rating) in a public, unmoderated
newsgroup, you violate the spirit of the usenet newsgroups. You are
perceived by many as a detrimentous offender. Nobody has to defend
against informations (call it ads if you have to) about new releases
of chess programs. Ed Schröder has always been modest about the gain
in playing strength, before the SSDF results were available he only
told us that he hoped to gain 20-30 points on the SSDF list. I don't
call this cheating ...

>Ed Schroder did promotion in the absence of concurrents in SSDF.
Ed Schröder published a chess playing program (Rebel 8) when he
finished it. The SSDF ratings that he refered to instead of own
guesses (he could have claimed much higher numbers and still have
been within the error margin of the list, so nobody would have called
this obvious cheating) seem to indicate that it played very strong,
so he quoted them because the SSDF is an independent organization and
if you tell people that your figures are SSDF ratings (some other
folks quote those ratings as absoulte "ELO" numbers, which is
completely wrong, since they aren't) everybody can figure out oneself
what to make out of it.

In your opinion, all programmers _have_ to release their programs at
the same time, maybe April 1st would suit you? Rolf, you are sick.
What makes you think that the chess playing strength of programs
depends on their release dates? This could only affect opening
preparation (e.g. killer books), which don't matter much anyway
(according to Marty Hirsch).

>Ed Schroder did cheating with his *surprise action*.
As far as I remember, you complained about being unable to use
ChessBase data without buying additional tools. This is not true, you
don't have to buy anything else: Rebel has had a .cbf to Rebel
converter for ChessBase files included for free since the Rebel 6.0
release. If you don't like this, you can also export .PGN files from
ChessBase and import them in Rebel (version 7 and up). Rebel 8 also
writes .PGN files. Oh well, Rebel 8 also reads and writes .EPD files.

>Ed Schroder did mean and mediocre behind the curtain war via email.
no comment. (since this is really private stuff between you and him)

>Ed Schroder did post exact wordings of emails.
He posted only his own words, not your replies. Please feel free to
correct me if I'm wrong.

Let me tell you what you did: You snipped his replies out of context
and made up your own version which made Ed look very cynical. Again,
his "try harder Rolf, try harder" was qouted (without any indication
of the different context) below some things you copied from another
article. I call this manipulation and a plain, simple lie of the
worst kind. You are not a newbie, you are a criminal.

>Ed Schroder did cheat with his 40 games called killer games.
He merely presented facts. He revealed every single detail about the
testing conditions (or you would never have known which autoplayer
settings he used etc.). Killer opening lines are played (by
definition) out of the opening book, so if games are discarded after
leaving the book, this wouldn't make any difference in finding killer
book lines. Even if a program that played killer lines would be able
to improve even more beyond those killer lines using its learning
function, this would not change the fact that the killer lines where
there beforehand. I hope that I made my point clear. Everybody can
accept his prepositions and accept the results (if not the
conclusions). Or you could e.g. say something like "no, I'm Rolf
Tueschen, I don't believe a single word from Ed Schröder". But you
have to understand that this is your personal conclusion. You may
present your evidence (if any ...) and line of thought (if any ...),
but you have to accept if most people don't agree with you. This
should finish the topic, but no, you keep calling Ed names _without_
any evidence at several other occasions.

>Ed Schroder did cheat with the matches on his webpage (Doubles)
see above. Either accept his prepositions or don't accept them. It's
up to you. Much more important than the autoplayer mode are the games
themselves that reveal much about the playing style and strength of
the programs involved. Ed Schröder is the only programmer that is
willing to reveal that much information about his program, please
note that he didn't hesitate to publish the negative score of Rebel 7
against Genius 3. He gives potential customers more informations
about his program than anybody else. He even gives a Rebel Decade
version away for free which resembles the Rebel 8 GUI so everybody
can try out the look and feel themselve.

>Ed Schroder did cheat on BT position testers. (See CSS 1994)
no comment. Your invaluable source carries its contradiction in itself.

>Ed Schroder did threaten to sue a newbie for reporting this.
he threatened to sue somebody who attacked him in a libelous way. If
I would kill you (to be sure: you wouldn't be worth it), could I also
claim "newbie" status in you opinion? I have never killed somebody
before, so this would be my first "murderer" experience. The
important thing is that others would judge my behaviour according to
certain standards I failed to comply with. The same thing is
happening to you: Everybody keeps telling you that you are way below
the standards both of decent human beings (a very subjective
definition ..) and this newsgroup (which is sometimes fairly low,
anyway, but you're about the absoulte zero mark) with your BIG topic.

>Ed Schroder did writings of ultimate surrender for over one week.
he hoped that you would be able to come back to your senses. His
logic was faulty in this point, since coming back to something means
that there is something to revert to.

>So who is responsible for the mud?
YOU. Everybody here strongly agreed. Maybe somebody would have told
you if anybody would like you to continue on your favourite
preoccupation (throwing mud at Ed Schröder). Nobody did. Instead,
many told you that they would like you to stop.

>The reporter?
Maybe it's all your parents fault ... I really don't want to know.

>
>
>Rolf Tueschen
>
>----*Man is unable not to know what he knows.* Leibowitz----
>

Bill Newton

unread,

Dec 22, 1996, 3:00:00 AM12/22/96

to

Bill Newton

unread,

Dec 22, 1996, 3:00:00 AM12/22/96

to

lensp...@aol.com

unread,

Dec 23, 1996, 3:00:00 AM12/23/96

to

In article <59b7l3$p...@lana.zippo.com>, Graham Douglass writes:

>>Yes, like all britains he has a nice black humor.
>>Monty Python is nothing against him.
>
>Monty Python is drivel.

I don't know about you, but I am ready to turn this newsgroup off as far
as my FlashSession scan, because, apart from one or two people with no
commercial interests, all I have read is a bunch of back-biting,
face-spitting flaming among a bunch of idiots who may be able to write a
chess program, but whose parents FAILED MISERABLY in teaching them some
etiquette. Maybe Ed was right in leaving. I don't think I would put up
with being flamed continuously, because there is nothing to stop the
insults from returning.

I subscribed to this group to hopefully find out some of the latest tips
and tricks in programming, not to put up with a bunch of political BS.

I'm not replying to you personally, but I am getting sick and tired of
having to wade through BS just to read what I am interested in.

And I agree on Monty Python, I outgrew them YEARS ago....

Bill Newton

unread,

Dec 23, 1996, 3:00:00 AM12/23/96

to

In article <01bbef4e$ffd11280$c2ec...@10.0.1.1>, Enrique Irazoqui
<en...@lix.intercom.es> writes

>No. If you pay all the attention you are capable of, you may understand the
>following. Chris tried to stop the postings of someone that was
>continuously insulting Ed. It was tried before by several people and it
>didn't work. This is the issue. Accusations and insults are offending, but
>you choose to pick on Chris's "two word obscenity" and play the role of
>moral vigilante. Mind your own sense of morals and humor and play
>patronizing with your great grand children.

Oh dear, you're upset AND confused!

Allow me to suggest that you re-read your own writing as reproduced
above and you'll see that you're now stating that " accusations and
insults are offending"

Ergo, CWs original posting was offensive.

That 'is' and always has been 'my' point.

>End of thread.

Not quite ;-)

Let logic be your bedfellow.

Regards.

--
Bill Newton

Rolf Tueschen

unread,

Dec 23, 1996, 3:00:00 AM12/23/96

to

Bill Newton <not...@demon.co.uk> quoted:

>In article <01bbef4e$ffd11280$c2ec...@10.0.1.1>, Enrique Irazoqui
><en...@lix.intercom.es> writes
>>No. If you pay all the attention you are capable of, you may understand the
>>following. Chris tried to stop the postings of someone that was
>>continuously insulting Ed. It was tried before by several people and it
>>didn't work. This is the issue. Accusations and insults are offending, but
>>you choose to pick on Chris's "two word obscenity" and play the role of
>>moral vigilante. Mind your own sense of morals and humor and play
>>patronizing with your great grand children.

and commented:

>Oh dear, you're upset AND confused!

>--
>Bill Newton
----------------------------------------------------------------------------------------------------------

Enrique, now please make a stop.
Virtually s.th. like a mindstop or in reality a thought regeneration.

I heard that you were really one of a few senior experts here in this group.

Now I want to make it very clear to you what's the Ed-help-activity of some
writers all about.
You know what? It's really nonsense.
I think that you should be able to understand that
attacks/questions/assumptions/hypothesis are NOT necessarily insults, ok? Or
did you refer on some unwritten royal ettiquette??

I hope you remember our short email intermezzo. I thought that you were a
little bit too much on the side of Ed. Not speaking of taking sides between me
and him, there I consider your behavior as ok. But as concerning Ed's actions
here on the net and in private mail you are completely wrong. He was NOT honest
at all. He was simply moneymaking, ok? And for reaching his goal he achieved to
confuse SSDF and a lot of rgcc. Sure, he gave some support on Rebel8 too. But
his main activities were attacks and insults on collegues and diagreeable
critics. *I* have proof of his mean actions via mail.

You was also confused by him. You wrote, keep away for awhile, I'll try to
clearify the killer thing with them. And what came out? Nothing. I never read
something clearifying. So the only thing I remember is my time-out.

After your actual activities for the help of Ed, I can no longer keep quite.

You try to participate in a mean game of a very confused person. For the sake
of his Rebel Ed is running wild. And after being put in the same boat with me
by the royal court of SSDF he got depressive and quit the group. Crying like a
14 year old mother-son. And exactly at the moment of this behavior you jumped
into the court. :)

Now I ask you as defender of Ed, where did I insult the big Eddie? If you
couldn't answer precisesly I would be forced to assume that you are giving your
name for mean actions for the sake of one special business man here, called Ed
Schroder. Schroder BV. And this company did misuse the usenet. Period.

Rethink all what you conserved (you once mentioned you hadn't followed the
discussion (!)) of the writings of both sides. And please make a new statement.
Otherwise you'll be washed away with Ed. You won't achieve a stop of *a poster
here in this group*. I'll give you my word.
From the beginning of all my questions you were not visible. You did never
comment my writings. Now suddenly you play advocate for Ed. What has changed
your mind? Ed's unrealistic wishes?

I hope you got it. I repeat. I want to read from *you* where I did insult Ed.
Ok? If you could further show me where I did go wrong I will apologize as quick
as possible. That's for sure. But let me humbly give you a trivial hint.
Insults couldn't be judged only by their spelling/language. You have to analyse
the content. :)

Now please speak it all out. That's freedom in internet.

BTW nevertheless I wish you some peaceful days coming

Harald Faber

unread,

Dec 23, 1996, 3:00:00 AM12/23/96

to

Hello Rolf,

I think I have an interesting aspect to tell you:

RWT> It's so bad that Ed cheated with his surprise action. Because as new user
RWT> you had to buy further programs to do converting.

Absolutely rubbish.
TB.EXE converts cbf-format into Rebel-format, and it is included somewhere
on 1st or 2nd disk
What you perhaps mean is DBUTIL which has only the possibility to kill
doubles in databases, the rest is implemennted in Rebel8.

RWT> As concerning this some here in the group didn't know that. But in the
RWT> surprise action this tool was NOT included. I got this from Ed
RWT> personally. So with all my cbf material I repeat Ed trapped me in a
RWT> little cheating as other *Kaffefahrten*- or *Take part you always win
RWT> action*-organizer do too.

Rubbish, read above.

RWT> Important to know that all that could be said because Ed did NOT inform
RWT> about these costs in advance. So I still call it cheating. Period.
RWT> Ed?

You are not able to read ads, aren't you? There are many features, also
for importing databases, explained.

Don't write about things you don't know.

Ciao and see ya
Harald
--

Rolf Tueschen

unread,

Dec 27, 1996, 3:00:00 AM12/27/96

to

Harald...@p21.f2.n1.z1001.fidonet.org (Harald Faber) wrote:

>I think I have an interesting aspect to tell you:

>RWT> It's so bad that Ed cheated with his surprise action. Because as new user
>RWT> you had to buy further programs to do converting.

>Absolutely rubbish.
>TB.EXE converts cbf-format into Rebel-format, and it is included somewhere
>on 1st or 2nd disk
>What you perhaps mean is DBUTIL

>RWT> Important to know that all that could be said because Ed did NOT inform

>RWT> about these costs in advance. So I still call it cheating. Period.
>RWT> Ed?

>You are not able to read ads, aren't you? There are many features, also
>for importing databases, explained.

>Don't write about things you don't know.

>Ciao and see ya
>Harald
>--

--------------------------------------------------------------------------------------------------------------------------

Sorry, but that's more complicated than Harald Faber thought. And I think that
Schroder BV knew better what was all about the tools for complete use of
Rebel8. As I said Schroder BV sent email with exact differentiation.

And I'd like to repeat my annoyance regarding such boasting posts *I think I
have an interesting ...*, *You're not able to read*, *Don't write about things
you don't know.*
No, that's not the style of open conversation.
*Ciao and see ya*. That's sort of hypocritical r/c bull.

Look, first try to understand the whole story from August on. And then rethink
all your conclusions. But you have to keep the history in mind. If now Schroder
BV changed policy it's good, but back in August/september Schroder BV himself
wrote me about the reasons for not including the needed tools. He wrote, that
this was important for the *normal* Rebel client. So, what YOU think as ALWAYS
included, is obviously NOT true. For a killing doubles tool alone around 48 DM
extra costs? Please explain that.

So I repeat, and I never did say more, that Schroder BV DID CHEAT with his
surprise action in August 1996. Period.
I think I could read *Ed*, whereas *ads* that I don't have, I couldn't read
them.

Now, Harald, *ciao and see ya*, come back with your corrections. ^H^a^r ^a^l^d.

mclane

unread,

Dec 28, 1996, 3:00:00 AM12/28/96

to

Harald...@p21.f2.n1.z1001.fidonet.org (Harald Faber) wrote:

>Hello Rolf,

>I think I have an interesting aspect to tell you:

>Don't write about things you don't know.

Come on. Peace! Chess instead WAR.
They always ask ME if I have a personal "vendetta".
I will not ask you the same. Also I will not ask RT.

I am your opinon. But is your statement any help to our problem.
In my opinon hate will produce hate.

I will give a poem, sorry that it is in german, but maybe you and Rolf
understand it:

Das Wort

Sprich nie ein liebeloses Wort,
es ist nicht nur ein leerer Schall.
Du sendest es zwar von dir fort,
doch bleibt es bei dir überall.

Es geht mit dir, wohin du gehst,
begleitet dich auf Schritt und Tritt,
und ob du es auch nicht verstehst,
es nimmt sogar noch andre mit.

So wächst die liebelose Schar,
die nichts als Böses von dir spricht,
und was zuerst ein Wort nur war,
das wird zum Spruch einst im Gericht.

>Ciao and see ya
>Harald
>--

In my opinon this poem is also written FOR ME, because I also forget
to often HOW I write.
What do you think, Harald ?

mclane

unread,

Dec 29, 1996, 3:00:00 AM12/29/96

to

Dave Gomboc

unread,

Dec 29, 1996, 3:00:00 AM12/29/96

to

In article <58tg88$m...@news00.btx.dtag.de>,

Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:
>drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
>

>>Actually I thought it was rather impolite. I am not saying that
>>someone is 'shit', as you put it. I am genuinely unclear as to
>>whether he was trying to say something of importance in that paragraph
>>or not! As for the "don't even ... bother posting", that is more a
>>reflection of the high esteem I hold of the series of posts spanning a
>>two-month period (so far!) by Rolf Tueschen which attempted to drag
>>Ed Schroder through the mud than it pertains to the SSDF discussion,
>>though the threads do cross from time to time.
>
>......
>
>>Dave Gomboc
>>drgo...@a.stu.athabascau.ca
>
>---------------------------------------------------------------------------------------------------------

>
>Very funny and highest level for sure.
>
>Take that:
>
>Ed Schroder did violate massively the rules of etiquette of newsgroups.

No. 5 years ago I would have said so, but commercial presence in
newsgroups is so common today that it is difficult for those who did
not "link up" in the last couple of years to imagine a time when a few
brave commercial posters were loudly flamed. Today several program
authors post at irregular intervals in rec.games.chess.computer, and
they are welcomed by (most of) the participants.

However, a smear campaign executed over 75 posts (probably 100 now),
such as the one you continue to commit, is still considered
exceedingly poor manners and reflects poorly upon its perpetrator(s).

>Ed Schroder did promotion in the absence of concurrents in SSDF.

You still have a problem with '2 1 '. It is not surprising that
if you are using a language that is not native to you that mistakes in
its usage will occur. That is not a problem either. What is a
problem is that you understand 'concurrent' in a proper manner but
choose to use it incorrectly anyway. As of October 15, 1996, Rebel
8.0 is not concurrent with Chess Genius 5.0. Rebel 8.0 is a product
which had been available to the public for some time. Chess Genius
5.0 was the probable name of a software program eagerly expected by
some of the chess community. They were *not* concurrent. Please get
this through your thick skull. (Mclane: I hope I'm not being too
polite this time. :-)

>Ed Schroder did cheating with his *surprise action*.

Exactly what cheating is this? He never stated that a double-game
remover was included as part of Rebel 8.0. Any functions he
advertised as part of Rebel 8.0 were implemented in the final release.
Feel free to correct me if I'm in error.

>Ed Schroder did mean and mediocre behind the curtain war via email.

This seems to tied in with your next line, so I'll answer there.

>Ed Schroder did post exact wordings of emails.

I have seen a couple of letters from Ed to *you* posted on this
newsgroup by Ed. I have not seen any letters that you sent him posted
on this newsgroup by Ed. It is not a breach of etiquette to post and
mail your own email. There is some debate over whether it is a breach
of etiquette to post someone else's mail - people have different
opinions on this - but since he hasn't done even that (please locate a
message in which he has if you disagree) it is a moot point.

>Ed Schroder did cheat with his 40 games called killer games.

Perhaps the word you use in German hasn't translated very well. I saw
a bunch of games, and played through 10 or so. They looked like
killer lines to me. Marty Hirsch posted regarding one example,
explaining that in one particular position his opening book
specialist, Sandro Neechi, had come up with a novelty and included
various lines which were not known to theory at the time. Games in
such a variation could be considered to be 'killer' games, though
since the relevance of the variation extends to a wide range of
computer and human opponents (as opposed to just a couple of
competitors) I would classify it instead as quality opening
preparation. While it is a mistake to refer to that particular line
as a killer line, it is not a mistake to refer to a very high
percentage of the games presented as evidence as killer lines. It is
fairer to say that Ed Schroder has made a (small!) error rather than
Ed Schroder is cheating. Why small? Because if you remove the one
game from his list of examples, his claim is still worth serious
debate.

>Ed Schroder did cheat with the matches on his webpage (Doubles)

The conditions under which the matches take place have been stated on
the web page and in this newsgroup more than once. There is no
cheating happening. If you do not agree with Ed's methodology, you
can find an alternate one at the SSDF site which will count 'double
games' which should relieve your concerns.

>Ed Schroder did cheat on BT position testers. (See CSS 1994)

Ed has already posted an explanation regarding this, but it was a
while ago.. perhaps you were not around when it was posted. I suppose
that he might post it again, or he might just consider it a waste of
time, as you are likely to manufacture more mud from it anyway. If
he feels the latter, I wouldn't blame him.

Is CSS not the same magazine that 0 a lot of bad press in this
newsgroup recently? From 0 posts it appears that it is not very
reputable. I do not have a subscription to the magazine, and if I did
I would have to get my mother (born in Vienna) to read it to me since
I do not know German, so I can't comment further regarding how
independent from commercial interests it is.

>Ed Schroder did threaten to sue a newbie for reporting this.

Did Ed Schroder threaten to sue someone who had been mudslinging for
50+ posts? Bummer, too bad he didn't go through with it. There is a
difference between the freedom to express one's opinions and
presenting this as fact, a line which you have crossed many times
(including in the post I am replying to.)

>Ed Schroder did writings of ultimate surrender for over one week.

You'll have to explain this one to me. I don't know what you mean in
your sentence.

>So who is responsible for the mud?

>The reporter?

The creator of the mud.. that would be you.

>Rolf Tueschen
>
>----*Man is unable not to know what he knows.* Leibowitz----
>

Dave Gomboc
drgo...@a.stu.athabascau.ca

Dave Gomboc

unread,

Dec 29, 1996, 3:00:00 AM12/29/96

to

In article <58te6s$m...@news00.btx.dtag.de>,

Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:

#drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:
#
#>In article <32B049...@nwlink.com>, brucemo <bru...@nwlink.com> wrote:
#>>Dave Gomboc wrote:
#>>
#>>> There's no cheating factor. He's simply stating the obvious (well,
#>>> apparently not obvious, since it needs to be explained to you): the
#>>> more games that are played in the SSDF system, the less that future
#>>> games will affect measurements are taken. Rebel 8.0's rating will
#>>> continue to shift with all the rest, as games continue to be played.
#>>> (Exercise: demonstrate to yourself that the rating of a program on the
#>>> SSDF list can move between lists despite not having any games played
#>>> between a list and the list previous to it.)
#>>
#>>This depends a lot on how they do the rating system.
#
#>Yes, I agree.
#
#>[Bruce's remarks regarding standard ELO setups deleted]
#
#>>Is the SSDF list vulnerable to any of this, or do you guys use a
#>>completely different rating system or something like that?
#
#>I was confident when I posted my earlier message that the SSDF treat
#>all games as one large dataset, i.e. no "update" calculations are done
#>from one list to the next. Instead ratings for the entire table are
#>solved for based on all results in the table in roughly the manner I
#>will detail below:
#
#>1) Collect all results of players against each other, and record them
#>in a large 2-d array.
#
#>2) Assign every player an identical initial rating (RI) rating (we'll
#>pick 2000 for now).
#
#>3) Assign an 'updated rating' (RU) based upon the performance rating formula,
#>which is RO + 400 * (W-L)/G, where RO is average opponent rating
#> W is number of wins
#> L is number of losses
#> G is number of games played
#
#> To calculate average opponent rating, you need to 'scale', e.g.
#
#> If I play a 2200 once, a 2000 twice, and a 1900 once, my RO is 2025.
#
#> 2200*1 + 2000*2 + 1900*1
#> ------------------------ = 2025 = RO
#> 1+2+1
#
#---------------------------------------------------------------------------------------------------------------
#see No 1 below
#---------------------------------------------------------------------------------------------------------------
#
#-------
#
#>4) Make the RU value for each player their new RI, and iterate until
#> the ratings converge.
#
#>5) If you have done a previous list, it is probably a good idea to
#>find the best fit between the previous list and this one, so that the
#>ratings are assured to be relatively level from list to list.
#
#----------------------------------------------------------------------------------------------------------
#see No 2 below
#----------------------------------------------------------------------------------------------------------
#
#>A couple of simplifications have been made in step 5.. really it is
#>assuming that all ratings were equally reliable, as in the case
#
#> Old List: A 1800 after 6 games, B 2200 after 6 games, C 2000 after 6 games
#> New List: A, B, C, D have played 8 each.
#
#>In reality, some machines have played more games than others, and as a
#>result instead of minimizing the function I have above, the goal
#>should be to minimize a similar but more complex function, which
#>accounts for the variability in confidence in ratings.
#
#>On another note, a somewhat reasonable function would be to restrict
#>minimization of difference to those players who do not have new
#>results. (This wouldn't work when most of the pool plays regularly,
#>but is great for a computer list, since most of the computers don't
#>really play any more and their skill level is assumed to remain
#>constant.) This particular more complex function would assign
#>identical ratings to players on the list who haven't played since the
#>last list. Clearly, this wasn't the function I had in mind when I
#>wrote my quoted paragraph (at the top of the message.)
#
#---------------------------------------------------------------------------------------------------------------
#see No 3 below
#---------------------------------------------------------------------------------------------------------------
#
#
#I'd like to give three questions of understanding to be able to judge on
#this detailed explanation. Thanks for possible answers.
#
#Note please, that the important second part of the above quoted post isn't
#of interest for me because I did and still claim that SSDF did cheat under
#conditions before *learning* came up.
#
#No 1
#
#Is assumption true that value for RO depends a lot on Elos of the prgs AND
#how many games are played against low or high leveled Elos?
#
#(Remember my critics of SSDF proceduring of DIFFERENT game numbers and
#different adversairy prgs that are played by a certain prg?)

The value for RO does depends on the opponent's ELO ratings, where RO
is weighted (as above) because some opponents are played more than
others.

#
#No 2
#
#Regarding my claim that there could be no such time independence strength
#value of a certain prg as such could someone show some evidence for the
#truth of the following?
#
#A certain prg P tested at a certain time against n other prgs with a
#certain Elo E1 as result.
#
#Then after some time same P tested against n-5 prgs AND some new n+5 prgs
#with a E2 result.
#
#And the assumption made by SSDF and author of former post is E1 = E2?
#
#And if NOT directly then a certain complex such and such function would be
#utilized to make them equal?

No, the assumption is not that they are equal. The assumption is that
if you have the set of programs V1 (includes 200 different E1s) and
the set of programs V2 (includes 200 different E2s) than the means of
V1 and V2 will be close to the same. Even this statement is probably
inaccurate, but it is closer to the spirit of what is happening.
Basically, any particular program may score better or worse with the
substitution of some 5 programs for some 5 other programs, but overall
if the programs that are added in are about the same strength as the
programs taken out then the average rating will remain the same.

In normal practice, the five added in are stronger than the five taken
out, but it is ideal that the mean of V1 be the same as the mean of
V2, so a constant is added (or subtracted) to V2 so that the mean of
V2 is the same as the mean of V1.

#No 3
#
#And especially because it *happened* that some prgs played different
#quantity of games a *complex correction* was needed?
#
#(Remember my claim that it was intollerable to match prgs vs different
#other prgs AND more so with different quantity?)
#
#What this complex functions looks like?

I have spent too much time already detailing the mathematical
processes by which SSDF are (and could be) calculated. It's probably
best if you ask someone in the Mathematics and Statistics department
at your local university. They will probably speak German, so you're
more likely to achieve satisfaction with their explanation than with
mine. Personally, I don't feel that a more accurate function than I
described is really required.. the margins of error are currently too
high, and the increase in accuracy is hidden by the lack of precision.
It would be more worthwhile if each software program played about 7000
games instead of 700, as the margin of error would be correspondingly
reduced.

What I am saying basically is that because SSDF's margin of error is
usually around 30 points for the top programs, it is not especially
meaningful to report the last digit of their results. It is totally
justifiable to round each result to the nearest 10 (or 5, as FIDE does
it) and report that rating instead: such practice would reduce the
amount of "Wow! Program XXX is 2 points better! YYY is not as good!"
that goes on. It is also perfectly acceptable (and from a theoretical
standpoint, better) to report that last digit, and heck, even tenths
of a point (i.e. 2406.4 +28.4 -29.1). The margins of error are
provided beside the rating, so readers (should) know that the extra
accuracy doesn't mean much. To say the program is 2405 +-30 is quite
sufficient.

#May I add finally my former claim that Elo or SSDF results didn't/ and
#couldn't give a certain *real* value for the strength of a prg?
#I meant that each calculated score only could show a value in relation to
#other tested prgs at a certain time.
#
#And what I did never understand why SSDF whose staff IS in close contact
#with their testers couldn't distribute a balanced testing proceduring
#before hand instead of inventing some *complex* corrections afterwards.

A calculated score in relation to other scores is exactly what rating
systems are made of. Three tournaments ago, my (CFC - national)
rating was 2029. After two tournaments from then it went to 2084,
which is what you will find on the (Alberta) rating list at
www.chesscanada.org. When they rate the tournament I played in last
weekend it will probably go up to 2125-2130. None of those numbers
mean anything by themselves. When you compare these numbers you
notice that I am playing chess better than I was before. It's only in
relation to the other numbers that a rating means anything.

The SSDF could 'balance' out their games more, but as long as the
sample is reasonably random then it's okay. Remember that most rating
systems don't have the ability to control who plays who, and how many
times. They just do the best job they can with the results they are
provided with - and they do a pretty good job.

#
#Rolf Tueschen
#
#----*Man is unable not to know what he knows.* Leibowitz----
#

Dave Gomboc
drgo...@a.stu.athabascau.ca

Cpsoft

unread,

Dec 29, 1996, 3:00:00 AM12/29/96

to

Dave Gomboc <drgo...@a.stu.athabascau.ca> wrote in article <5a4qug$2...@aurora.cs.athabascau.ca>...
> In article <58tg88$m...@news00.btx.dtag.de>,
> Rolf W. Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote:
> >drgo...@a.stu.athabascau.ca (Dave Gomboc) wrote:

> >
>
> >Ed Schroder did promotion in the absence of concurrents in SSDF.
>
> You still have a problem with '2 1 '. It is not surprising that
> if you are using a language that is not native to you that mistakes in
> its usage will occur. That is not a problem either. What is a
> problem is that you understand 'concurrent' in a proper manner but
> choose to use it incorrectly anyway. As of October 15, 1996, Rebel
> 8.0 is not concurrent with Chess Genius 5.0. Rebel 8.0 is a product
> which had been available to the public for some time. Chess Genius
> 5.0 was the probable name of a software program eagerly expected by
> some of the chess community. They were *not* concurrent. Please get
> this through your thick skull. (Mclane: I hope I'm not being too
> polite this time. :-)
>

Germans use the word 'concurrent' to mean 'competitor' as in Rebel
and Genius are competitive/concurrent in the market place.
Ie. competitive/concurrent for each other's business.

Although quite what the original statement "> >Ed Schroder did promotion in the absence of concurrents in SSDF." means in this context remain unclear.

Aside to Gomboc: agree with the rest of your post, though :)

Chris Whittington

Cpsoft

unread,

Dec 30, 1996, 3:00:00 AM12/30/96

to

Rolf Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in article <5a1a21$e...@news00.btx.dtag.de>...
> Harald...@p21.f2.n1.z1001.fidonet.org (Harald Faber) wrote:

>
> >I think I have an interesting aspect to tell you:
>

> >RWT> It's so bad that Ed cheated with his surprise action. Because as new user
> >RWT> you had to buy further programs to do converting.
>
> >Absolutely rubbish.
> >TB.EXE converts cbf-format into Rebel-format, and it is included somewhere
> >on 1st or 2nd disk
> >What you perhaps mean is DBUTIL
>
> >RWT> Important to know that all that could be said because Ed did NOT inform
> >RWT> about these costs in advance. So I still call it cheating. Period.
> >RWT> Ed?
>
> >You are not able to read ads, aren't you? There are many features, also
> >for importing databases, explained.
>

> >Don't write about things you don't know.
>

> >Ciao and see ya
> >Harald
> >--

> --------------------------------------------------------------------------------------------------------------------------
>
> Sorry, but that's more complicated than Harald Faber thought. And I think that
> Schroder BV knew better what was all about the tools for complete use of
> Rebel8. As I said Schroder BV sent email with exact differentiation.
>
> And I'd like to repeat my annoyance regarding such boasting posts *I think I
> have an interesting ...*, *You're not able to read*, *Don't write about things
> you don't know.*
> No, that's not the style of open conversation.
> *Ciao and see ya*. That's sort of hypocritical r/c bull.
>
> Look, first try to understand the whole story from August on. And then rethink
> all your conclusions. But you have to keep the history in mind. If now Schroder
> BV changed policy it's good, but back in August/september Schroder BV himself
> wrote me about the reasons for not including the needed tools. He wrote, that
> this was important for the *normal* Rebel client. So, what YOU think as ALWAYS
> included, is obviously NOT true. For a killing doubles tool alone around 48 DM
> extra costs? Please explain that.
>
> So I repeat, and I never did say more, that Schroder BV DID CHEAT with his
> surprise action in August 1996. Period.
> I think I could read *Ed*, whereas *ads* that I don't have, I couldn't read
> them.
>
> Now, Harald, *ciao and see ya*, come back with your corrections. ^H^a^r ^a^l^d.
>

Teuschen, nobody cares about this endless drivel.

FUCK-OFF

Chris Whittington

>
>
>

Jean Peter Fendrich

unread,

Dec 30, 1996, 3:00:00 AM12/30/96

to

Dave Gomboc wrote:

- snip -

>>So who is responsible for the mud?
>>The reporter?

>The creator of the mud.. that would be you.

- snip -

In Sweden we have this saying about 'throwing pearls to the pigs'.
The deeper meaning of this saying shouldn't be too hard to catch.
I think this is what you are doing right here ... :)

--
J-P Fendrich

Harald Faber

unread,

Dec 30, 1996, 3:00:00 AM12/30/96

to

Hello mclane,

m> From: mcl...@prima.ruhr.de (mclane)
m> Subject: Re: SSDF didn't stop cheating in mags CSS and ER
m> Organization: Prima e.V. Dortmund

m>
m> >Hello Rolf,
m> >Don't write about things you don't know.
m>
m> Come on. Peace! Chess instead WAR.
m> I am your opinon. But is your statement any help to our problem.
m> In my opinon hate will produce hate.

It was no hate, just direct. read below.

m> Sprich nie ein liebeloses Wort,
m> es ist nicht nur ein leerer Schall.
m> Du sendest es zwar von dir fort,
m> doch bleibt es bei dir nberall.

Du solltest die Umlaute weglassen. ;-)

m> In my opinon this poem is also written FOR ME, because I also forget
m> to often HOW I write.
m> What do you think, Harald ?

Ich halte mich schon zurueck, aber ich war eben der Meinung, dass das sein
musste. Und wenn ich aergerlich geworden waere, haette ich schon mit
entsprechender Grossschrift und negativen Smileys darauf hingewiesen.