How can Deep Blue lose?
Shadow Warrior
I thought I'd heard that computers such as Deep Blue could look ahead
1000's of moves. If this is so, how can a computer like Deep Blue lose
any games?
brucemo
By making bad moves against good players, same as anyone else.
bruce
George R. Barrett
Chris Carson wrote:
>
> Shadow Warrior wrote:
> >
> > I thought I'd heard that computers such as Deep Blue could look ahead
> > 1000's of moves. If this is so, how can a computer like Deep Blue lose
> > any games?
>
> (see http://www.chess.ibm.com/home/html/b.html)
>
> This translates into somewhere between 12 and 14 ply (half moves).
> This is based on an average of 3 minutes per move time control
> and an average of typically 36 moves to examine per position
> and a typical Alph-Beta search (note: Deep Blue team claims to
> do a full width search for the first few ply's).
>
Well, let's do the math and find out if the Deep Blue team is full of
it...
Let n = number of ply (half moves), then for 36 moves per position this
gives:
36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
Deep Blue's team is full of dark and nasty unmentionables.
[snip]
> Best Regards,
> Chris Carson email: chris-...@ti.com
--
_____________________________________________________________________
George R. Barrett
Electrical Engineering : Systems `` Without insight, method is
University of Michigan, Ann Arbor largely useless. ''
_____________________________________________________________________
"We make our world significant by the courage of our questions and
the depth of our answers." -- Carl Sagan
Chris Carson
Shadow Warrior wrote:
>
> I thought I'd heard that computers such as Deep Blue could look ahead
> 1000's of moves. If this is so, how can a computer like Deep Blue lose
> any games?
Actually, Deep Blue looks at up to 200 million positions per second.
(see http://www.chess.ibm.com/home/html/b.html)
This translates into somewhere between 12 and 14 ply (half moves).
This is based on an average of 3 minutes per move time control
and an average of typically 36 moves to examine per position
and a typical Alph-Beta search (note: Deep Blue team claims to
do a full width search for the first few ply's).
This means that Deep Blue sees 6 or 7 moves ahead (more or less
depending on the position and time constraints, it could see many
more moves ahead depending on the positon and time, but I do not
recall any claim of 1000's of moves ahead. I think DB has been
disassembled, Deep Blue Jr. has made a couple of appearances
earlier this summer.
I am not a Deep Blue expert, but maybe this helps.
Best Regards,
Chris Carson
--
Robert Hyatt
George R. Barrett (grba...@eecs.umich.edu) wrote:
: Chris Carson wrote:
: >
: > >
: > > I thought I'd heard that computers such as Deep Blue could look ahead
: > > 1000's of moves. If this is so, how can a computer like Deep Blue lose
: > > any games?
: >
: > Actually, Deep Blue looks at up to 200 million positions per second.
: > (see http://www.chess.ibm.com/home/html/b.html)
: >
: > This translates into somewhere between 12 and 14 ply (half moves).
: > This is based on an average of 3 minutes per move time control
: > and an average of typically 36 moves to examine per position
: > and a typical Alph-Beta search (note: Deep Blue team claims to
: > do a full width search for the first few ply's).
: Well, let's do the math and find out if the Deep Blue team is full of
: it...
: Let n = number of ply (half moves), then for 36 moves per position this
: gives:
: 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
you left out alpha/beta. This is a pure minimax tree size. Alpha
beta will produce trees of the following side:
D=depth, W=width or branching factor, N=number of terminal positions.
for D even:
N=W^(D/2)+W^(D/2) = 2*W^(D/2)
for D odd:
n=W^floor(D/2)+W^ceil(D/2)
So alpha/beta will reduce the size of the tree to approximately the square
root of the size of the full minimax tree...
: or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
: Deep Blue's team is full of dark and nasty unmentionables.
Using the alpha/beta number, you get N~=13 or so, which is right in the
right ballpark. Of course, if they used null-move R=2, they could get to
at *least* 15 (2 plies more on average)...
Randy Baker
George R. Barrett <grba...@eecs.umich.edu> wrote in article
<33F22D89...@eecs.umich.edu>...
> Chris Carson wrote:
> >
> > Shadow Warrior wrote:
> > >
> > > I thought I'd heard that computers such as Deep Blue could look ahead
> > > 1000's of moves. If this is so, how can a computer like Deep Blue
lose
> > > any games?
> >
> > Actually, Deep Blue looks at up to 200 million positions per second.
> > (see http://www.chess.ibm.com/home/html/b.html)
> >
> > This translates into somewhere between 12 and 14 ply (half moves).
> > This is based on an average of 3 minutes per move time control
> > and an average of typically 36 moves to examine per position
> > and a typical Alph-Beta search (note: Deep Blue team claims to
> > do a full width search for the first few ply's).
> >
>
> Well, let's do the math and find out if the Deep Blue team is full of
> it...
>
> Let n = number of ply (half moves), then for 36 moves per position this
> gives:
>
> 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
>
> Deep Blue's team is full of dark and nasty unmentionables.
You forget that alpha-beta can optimistically reduce the branching factor
to about 6 (square root of minimax branching factors), so the "proof" lies
elsewhere...
--
Randy Baker (remove Z from address in email replies)
>
>
> [snip]
>
> > Best Regards,
> > Chris Carson email: chris-...@ti.com
>
Robert Hyatt
Shadow Warrior (wad...@hotmail.com) wrote:
: Shadow Warrior wrote:
: >
: > I thought I'd heard that computers such as Deep Blue could look ahead
: > 1000's of moves. If this is so, how can a computer like Deep Blue lose
: > any games?
: Thanks for all the input. Just a couple more questions, then I promise I
: will leave you guys alone.
: Would Deep Blue win if there were no time constraints?
If db had no time constraints, but the opponent did, it would be bad. But
if neither has any time constraints, it would have no special advantage, other
than a computer can focus on a game for hours or days without stopping, while
a human can't..
: What was the function of the Deep Blue team? Were they adjusting A.I. on
: the fly?
They adjusted things between rounds it seems. Not to mention the years
of work to get it to the level it is currently at.
: Programming bugs aside, would Deep Blue require strategy of any kind, or
: could it win by brute force?
No. Search won't do it alone. They spent a lot of time on evaluation
to play at the level they are at.
: Finally, any idea how many moves ahead someone like Kasparov is able to
: see?
"enough" is the common answer, but not "enough" when playing DB it seemed. :)
: Thanks in advance from a chess rookie,
: Shadow Warrior
Howard Exner
Shadow Warrior <wad...@hotmail.com> wrote in article
<33F31F...@hotmail.com>...
> Shadow Warrior wrote:
> >
> > I thought I'd heard that computers such as Deep Blue could look
ahead
> > 1000's of moves. If this is so, how can a computer like Deep Blue
lose
> > any games?
Unplug the bastard.
Stephen L. Holliday
Although Deep Blue can look ahead 1000's of moves, it doesn't know anything
about chess strategy, forming a play, even some lost positions. They
especially have trouble with understanding a spacial advantage. And don't
forget, every program has it's bugs.
Deep Blue's program is probably not as bug ridden as Windows 95, but I bet
there are a few.
Shadow Warrior <wad...@hotmail.com> wrote in article
Tom Kerrigan
George R. Barrett (grba...@eecs.umich.edu) wrote:
: Well, let's do the math and find out if the Deep Blue team is full of
: it...
: Let n = number of ply (half moves), then for 36 moves per position this
: gives:
: 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
: or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
: Deep Blue's team is full of dark and nasty unmentionables.
Yes, you're very, very wrong. Please shut up.
Cheers,
Tom
George R. Barrett
> : >
> : > Shadow Warrior wrote:
> : > > I thought I'd heard that computers such as Deep Blue could look ahead
> : > > 1000's of moves. If this is so, how can a computer like Deep Blue lose
> : > > any games?
> : > Actually, Deep Blue looks at up to 200 million positions per second.
> : > (see http://www.chess.ibm.com/home/html/b.html)
> : >
> : > This translates into somewhere between 12 and 14 ply (half moves).
> : > This is based on an average of 3 minutes per move time control
> : > and an average of typically 36 moves to examine per position
> : > and a typical Alph-Beta search (note: Deep Blue team claims to
> : > do a full width search for the first few ply's).
>
> : Well, let's do the math and find out if the Deep Blue team is full of
> : it...
>
> : Let n = number of ply (half moves), then for 36 moves per position this
> : gives:
>
> : 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
>
> beta will produce trees of the following side:
I've gotten at least two replys that say I left out alpha/beta.
No, I didn't. Well not intentionally. I understood the original poster
to say that the Deep Blue team claimed 12-14 plys at full width.
Looking back, I see it actually says they claim to do full width for the
"first few plys". Whatever that means: 4 plys? 5 plys? 6 plys?
Clearly, if I had realized how ambiguous the original poster's
re-statement of the Deep Blue team's claims were, I wouldn't have
responded with a mathematical equation. That would be absurd.
Tom Kerrigan telling me to "shut up" is just rude. To the others whose
comments simply pointed out my misunderstanding of the claim, thanks.
>
> D=depth, W=width or branching factor, N=number of terminal positions.
>
> for D even:
>
> N=W^(D/2)+W^(D/2) = 2*W^(D/2)
>
> for D odd:
>
> n=W^floor(D/2)+W^ceil(D/2)
>
> So alpha/beta will reduce the size of the tree to approximately the square
> root of the size of the full minimax tree...
>
> : or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
> : Deep Blue's team is full of dark and nasty unmentionables.
>
> Using the alpha/beta number, you get N~=13 or so, which is right in the
> right ballpark. Of course, if they used null-move R=2, they could get to
> at *least* 15 (2 plies more on average)...
--
Shadow Warrior
Shadow Warrior wrote:
>
> I thought I'd heard that computers such as Deep Blue could look ahead
> 1000's of moves. If this is so, how can a computer like Deep Blue lose
> any games?
Thanks for all the input. Just a couple more questions, then I promise I
will leave you guys alone.
Would Deep Blue win if there were no time constraints?
What was the function of the Deep Blue team? Were they adjusting A.I. on
the fly?
Programming bugs aside, would Deep Blue require strategy of any kind, or
could it win by brute force?
Finally, any idea how many moves ahead someone like Kasparov is able to
see?
Robert Hyatt
: >
: > George R. Barrett (grba...@eecs.umich.edu) wrote:
: > : Chris Carson wrote:
: > : >
: > : > >
: > : > > I thought I'd heard that computers such as Deep Blue could look ahead
: > : > > 1000's of moves. If this is so, how can a computer like Deep Blue lose
: > : > > any games?
: > : > Actually, Deep Blue looks at up to 200 million positions per second.
: > : > (see http://www.chess.ibm.com/home/html/b.html)
: > : >
: > : > This translates into somewhere between 12 and 14 ply (half moves).
: > : > This is based on an average of 3 minutes per move time control
: > : > and an average of typically 36 moves to examine per position
: > : > and a typical Alph-Beta search (note: Deep Blue team claims to
: > : > do a full width search for the first few ply's).
: > : >
: >
: > : Well, let's do the math and find out if the Deep Blue team is full of
: > : it...
: >
: > : Let n = number of ply (half moves), then for 36 moves per position this
: > : gives:
: >
: > : 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
: >
: > you left out alpha/beta. This is a pure minimax tree size. Alpha
: > beta will produce trees of the following side:
: I've gotten at least two replys that say I left out alpha/beta.
: No, I didn't. Well not intentionally. I understood the original poster
: to say that the Deep Blue team claimed 12-14 plys at full width.
: Looking back, I see it actually says they claim to do full width for the
: "first few plys". Whatever that means: 4 plys? 5 plys? 6 plys?
: Clearly, if I had realized how ambiguous the original poster's
: re-statement of the Deep Blue team's claims were, I wouldn't have
: responded with a mathematical equation. That would be absurd.
DB actually searches full-width to whatever depth they reach. Last time
I sat across the table from them they were doing 11 plies at a couple of
million nodes per second. This was 11 full-width plies plus a heck of
a lot of other extensions to drive this even deeper... However, if you
factor 200M X 300 seconds (200M nodes per sec x 300 secs for an average
move) yu get 60,000 M nodes. If you use this for N below and back-solve,
13 seems reasonable...
: Tom Kerrigan telling me to "shut up" is just rude. To the others whose
Chris Carson
George R. Barrett wrote:
>
[SNIP]
>
> Looking back, I see it actually says they claim to do full width for the
> "first few plys". Whatever that means: 4 plys? 5 plys? 6 plys?
> Clearly, if I had realized how ambiguous the original poster's
> re-statement of the Deep Blue team's claims were, I wouldn't have
> responded with a mathematical equation. That would be absurd.
>
George,
I could not find the article that said Deep Blue searched full
width for the first few ply's. I remember reading it, but I
must not have saved the URL. :( Perhaps Deja News. I will look.
I am not a member of the Deep Blue Team and I do not work for IBM.
I just used the published 200 million positions per second
as published on http://www.chess.ibm.com/home/html/b.html and
plugged into the Alpha/Beta equations. My guess for full width
(it is only a guess) is 3 or 4 ply. This would be a small node
count compared to 200 million nodes per second. I think one
game with some Deep Blue analysis was published on this news group.
I do not have a copy, but I will search Deja News. I think only
part of the game with analysis was published, but if I recall,
the ply depth looked to be in this ball park.
Deep Blue is a great achievement. Deep Blue does not play perfect
chess (although I would not have a chance against it, so it might
seem perfect from my side of the board). :)
--
Robert Hyatt
Chris Carson (chris-car...@ti.com) wrote:
: George,
: I could not find the article that said Deep Blue searched full
: width for the first few ply's. I remember reading it, but I
: must not have saved the URL. :( Perhaps Deja News. I will look.
: I am not a member of the Deep Blue Team and I do not work for IBM.
: I just used the published 200 million positions per second
: as published on http://www.chess.ibm.com/home/html/b.html and
: plugged into the Alpha/Beta equations. My guess for full width
: (it is only a guess) is 3 or 4 ply. This would be a small node
: count compared to 200 million nodes per second. I think one
: game with some Deep Blue analysis was published on this news group.
: I do not have a copy, but I will search Deja News. I think only
: part of the game with analysis was published, but if I recall,
: the ply depth looked to be in this ball park.
unless they have changed the search drastically since I last saw and
talked with them, they were simply full-width to some depth, followed
by a quiescence search.. IE, DB has never been selective at all, except
if you choose to count the q-search (captures) as selective since all
moves are not followed there of course...
13-14 plies fits 200M nodes per second easily. Crafty at 100K typically
hits 12 in the middlegame using null-move. 300K would be 13, 900K would
be 14, 2.7M (round to 3M) would be 15, 9M would be 16, 27M would be 17,
81M would be 18, 243M would be 19... damn I want to build some hardware. :)
jeff kish
does anyone khow how well deep blue or any other program plays with out
there databases. i felt that the datbases, not the moves it can look at
per second, will make computers unbeatable. has any one done any kind of
study on how important databases are
Tom Kerrigan
George R. Barrett (grba...@eecs.umich.edu) wrote:
: Tom Kerrigan telling me to "shut up" is just rude. To the others whose
: comments simply pointed out my misunderstanding of the claim, thanks.
It certainly was, but nowhere near as screwed up as you saying the DB
team is full of BS because of your seriously flawed math.
Next time, present your math and ASK why it doesn't agree with whatever.
That way, you might learn something, and you won't be an a-hole doing it.
Cheers,
Tom
Mike Coffin
> George R. Barrett (grba...@eecs.umich.edu) wrote:
>
> Tom Kerrigan telling me to "shut up" is just rude. To the others whose
> comments simply pointed out my misunderstanding of the claim, thanks.
>
kerr...@deimos.frii.com (Tom Kerrigan) replied:
> It certainly was, but nowhere near as screwed up as you saying the DB
> team is full of BS because of your seriously flawed math.
>
> Next time, present your math and ASK why it doesn't agree with whatever.
> That way, you might learn something, and you won't be an a-hole doing it.
George: now that you've had a lesson in netiquette from Tom Kerrigan,
you are probably ready to proceed to your marriage counciling session
with O.J. Simpson and sportsmanship lessons from Mike Tyson.
Cheers,
-mike
brucemo
> : Tom Kerrigan telling me to "shut up" is just rude. To the others whose
> : comments simply pointed out my misunderstanding of the claim, thanks.
> It certainly was, but nowhere near as screwed up as you saying the DB
> team is full of BS because of your seriously flawed math.
> Next time, present your math and ASK why it doesn't agree with whatever.
> That way, you might learn something, and you won't be an a-hole doing it.
Tom, you need to calm down.
bruce
Stephen L. Holliday
Here (in my humble opinion) are the answere to your questions:
If Deep Blue had no time constraints then neither would it's opponent. This
would be a huge advantage for humans, giving them the time to spot deep
tactics (Computers are much stronger against humans in blitz than in longer
games)
I don't feel I know enough (does anybody?) about the secretive deep blue
team to answer your second question, except that deep blue is not an A.I.
program - it's just a giant chess calculator.
Every time Deep Blue wins, it's by sheer brute force. The only time it uses
strategy is in the opening (is that strategy?)
Kasparov claims to look up to 20-30 moves ahead, sometimes more if there
are a lot of forcing moves. While I don't want to say anything bad about
such a great chess player, I seriously doubt this. Looking so far ahead is
almost entirely guesswork, based mostly on intuition. Most strong players
usually only look a few moves ahead unless the position is very sharp.
If you are so interested in deep blue I suggest you go to IBM's Club
Kasparov's web page at http://www.club-kasparov.com
Jason the Great
Shadow Warrior <wad...@hotmail.com> wrote in article
> Shadow Warrior wrote:
> >
> > I thought I'd heard that computers such as Deep Blue could look ahead
> > 1000's of moves. If this is so, how can a computer like Deep Blue lose
> > any games?
>
Robert Hyatt
Komputer Korner (kor...@netcom.ca) wrote:
: If Fritz can finish 11 ply in 3 minutes surely Deeper Blue can get to
: 16.
: --
the suggestion is reasonable. The only thing wrong with it is that
Fritz uses null-move, R=2 to get to that depth. DB does not use null-move
*period*. Second, DB uses singular extensions, which causes the tree to
grow along narrow pathways, restricting the depth along other uninteresting
paths. This reduces their depth as well. I'd make a wild guess of
somewhere between 13-16 for the middlegame for them.
But as I said, I'd love to see crafty at 200M, just to see what it would
do. :)
Komputer Korner
Robert Hyatt wrote:
skipped
-Others wrote
-: I've gotten at least two replys that say I left out alpha/beta.
-: No, I didn't. Well not intentionally. I understood the original
poster
-: to say that the Deep Blue team claimed 12-14 plys at full width.
-: Looking back, I see it actually says they claim to do full width for
the
-: "first few plys". Whatever that means: 4 plys? 5 plys? 6 plys?
-: Clearly, if I had realized how ambiguous the original poster's
:- re-statement of the Deep Blue team's claims were, I wouldn't have
-: responded with a mathematical equation. That would be absurd.
-DB actually searches full-width to whatever depth they reach. Last
time
-I sat across the table from them they were doing 11 plies at a couple
of
-million nodes per second. This was 11 full-width plies plus a heck of
-a lot of other extensions to drive this even deeper... However, if you
-factor 200M X 300 seconds (200M nodes per sec x 300 secs for an average
-move) yu get 60,000 M nodes. If you use this for N below and
back-solve,-13 seems reasonable..
If Fritz can finish 11 ply in 3 minutes surely Deeper Blue can get to
16.
--
Best regards
Komputer Korner
The inkompetent komputer
If you see a 1 in my email address, take it out.
Note that my true email is still kor...@netcom.ca
I don't often check the email of the sympatico address.
Komputer Korner
Robert Hyatt wrote:
>
> Chris Carson (chris-car...@ti.com) wrote:
-: George,
-
-: I could not find the article that said Deep Blue searched full
-: width for the first few ply's. I remember reading it, but I
-: must not have saved the URL. :( Perhaps Deja News. I will look.
-: I am not a member of the Deep Blue Team and I do not work for IBM.
-: I just used the published 200 million positions per second
-: as published on http://www.chess.ibm.com/home/html/b.html and
-: plugged into the Alpha/Beta equations. My guess for full width
-: (it is only a guess) is 3 or 4 ply. This would be a small node
-: count compared to 200 million nodes per second. I think one
-: game with some Deep Blue analysis was published on this news group.
-: I do not have a copy, but I will search Deja News. I think only
-: part of the game with analysis was published, but if I recall,
-: the ply depth looked to be in this ball park.
-unless they have changed the search drastically since I last saw and
-talked with them, they were simply full-width to some depth, followed
-by a quiescence search.. IE, DB has never been selective at all, except
-if you choose to count the q-search (captures) as selective since all
-moves are not followed there of course...
-13-14 plies fits 200M nodes per second easily. Crafty at 100K
typically
-hits 12 in the middlegame using null-move. 300K would be 13, 900K
would
-be 14, 2.7M (round to 3M) would be 15, 9M would be 16, 27M would be 17,
-81M would be 18, 243M would be 19... damn I want to build some
hardware. :)
Are you now saying that Deeper Blue hits 18 ply in the middlegame
based on the above numbers?
Robert Hyatt
Komputer Korner (kor...@netcom.ca) wrote:
: Robert Hyatt wrote:
: >
: > Chris Carson (chris-car...@ti.com) wrote:
No. They are at least 2-3 plies behind Crafty for a given nodes-per-
second speed, because (a) they don't use null-move at all and (b) they
do *far* more search extensions than I do. They may be 5-6 plies behind
crafty for all I know. *if* crafty could search 200M nps of course...
Joe McCaughan
: : Well, let's do the math and find out if the Deep Blue team is full of
: : it...
: : Let n = number of ply (half moves), then for 36 moves per position this
: : gives:
: : 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
: : or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
: : Deep Blue's team is full of dark and nasty unmentionables.
: Yes, you're very, very wrong. Please shut up.
I think it's time to sequester Tom again...
This type of conduct is not needed here.
--Joe McCaughan
shi...@best.com
: Cheers,
: Tom
Rolf Tueschen
shi...@best.com (Joe McCaughan) wrote:
*This* type is not.
But calling for homicide or knee-shooting is obviously *not* that
important, I understand....
(Just to be precisely. I'm talking about the same Tom kid. He's the only
American demonstrating such an ugly face in rgcc. We Germans have Czub.
I think you have to choose the right moment to intervene. And homicide
is better suited than simple shut ups. I was the addressee in the other
case. Therefore my 23 cts. Not to flame you...)
brucemo
Joe McCaughan wrote:
>
> Tom Kerrigan (kerr...@deimos.frii.com) wrote:
> : George R. Barrett (grba...@eecs.umich.edu) wrote:
>
> : : Well, let's do the math and find out if the Deep Blue team is full of
> : : it...
>
> : : Let n = number of ply (half moves), then for 36 moves per position this
> : : gives:
>
> : : 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
>
> : : or n = log(3.6e10)/log(36), hence n ~= 6.78 ply. Ok, that proves it.
> : : Deep Blue's team is full of dark and nasty unmentionables.
>
> : Yes, you're very, very wrong. Please shut up.
>
> I think it's time to sequester Tom again...
> This type of conduct is not needed here.
>
Perhaps it would have been enough to simply correct the original poster,
rather than telling him to shut up.
"All the Right Moves", Carl Ebeling, p. 14, gives the following formula
for determining how many nodes are required to search to depth D using
the alpha-beta algorithm given optimal move ordering.
The formula is:
bf^((d+1)/2) + bf^(d/2) - 1
I wrote an Excel spreadsheet to calculate this for various depths,
assuming a branching factor of 35, which is commonly used for chess.
I assumed that the "d/2" and "(d+1)/2" sub-expressions should be
integerized, since I don't think for D=3 that you want to be calculating
35^1.5. If I'm wrong, the following is wrong.
"Nodes" is the nodes required to finish this ply, "Total" is the sum of
the nodes required to finish this ply and the nodes required to finish
each previous ply, in order to take into account iterative deepening.
A straight alpha-beta program using alpha-beta and no other pruning
would do 36 billion nodes in 3 minutes, which means that it wouldn't
make it through ply 13. This is in agreement with the poster's math, if
you add in Carl Ebeling's comment that alpha-beta pruning allows you to
go twice as deep in the same time.
I don't know what kind of pruning DB does, but I know that they have
done at least some expensive extensions in the past.
I hesitate to say that this chart simulates DB, since we know so little
about what they do.
bruce
-------
Depth Nodes Total
-------- --------------- ---------------
1 35
2 69 104
3 1,259 1,328
4 2,449 3,708
5 44,099 46,548
6 85,749 129,848
7 1,543,499 1,629,248
8 3,001,249 4,544,748
9 54,022,499 57,023,748
10 105,043,749 159,066,248
11 1,890,787,499 1,995,831,248
12 3,676,531,249 5,567,318,748
13 66,177,562,499 69,854,093,748
Peter Coleman
LOL! (Really. I just had to show my appreciation.)
Howard Exner wrote in article
<01bca8d6$624e5620$4125...@dlcwest.dlcwest.com>...
>Unplug the bastard.
Peter Coleman
First I've heard of this, but if it's true I agree with your statement. I
believe the longest ever mating sequence in a a game is of this magnitude
(and impossible to prove was seen at the time) and obviously this is
considerably more forcing than any normal position, let alone the sort of
closed positions that computers have traditionally been poor at (for just
the same reason). Excluding "forced" moves then around 6 moves (12 ply) is
probably the maximum for full analysis.
Stephen L. Holliday wrote in article <5t12hu$gi4$1...@gte2.gte.net>...
Robert Hyatt
brucemo (bru...@seanet.com) wrote:
: Perhaps it would have been enough to simply correct the original poster,
: rather than telling him to shut up.
: "All the Right Moves", Carl Ebeling, p. 14, gives the following formula
: for determining how many nodes are required to search to depth D using
: the alpha-beta algorithm given optimal move ordering.
: The formula is:
: bf^((d+1)/2) + bf^(d/2) - 1
: I wrote an Excel spreadsheet to calculate this for various depths,
: assuming a branching factor of 35, which is commonly used for chess.
: I assumed that the "d/2" and "(d+1)/2" sub-expressions should be
: integerized, since I don't think for D=3 that you want to be calculating
: 35^1.5. If I'm wrong, the following is wrong.
You are correct. The "actual" formula is:
N = W^floor(D/2) + W^ceil(D/2)
the mathematical "floor" function is integer division rounded *down*
"ceil" is integer divison rounded *up*... the D+1 does that in the
formula you gave...
: "Nodes" is the nodes required to finish this ply, "Total" is the sum of
Rolf Tueschen
kerr...@deimos.frii.com (Tom Kerrigan) wrote:
>Rolf Tueschen (TUESCHEN.MEDIZ...@t-online.de) wrote:
>: (Just to be precisely. I'm talking about the same Tom kid. He's the only
>: American demonstrating such an ugly face in rgcc. We Germans have Czub.
>: I think you have to choose the right moment to intervene. And homicide
>: is better suited than simple shut ups. I was the addressee in the other
>: case. Therefore my 23 cts. Not to flame you...)
>Yes, yes, we know who you're talking about. The problem here is that this
>is a newsgroup for computer chess, and I'd like to see interesting
>conversation and information exchange going on. More often than not, I
>see crap. You, for example, going WAY off-topic to personally attack Ed,
>which simply WAS NOT warranted. This guy calling the DB work BS because
>of his own flawed math. KK writing huge posts about programming aspects
>he clearly doesn't understand. Czub brining every single thing he sees
>on television into the discussions. All of this behavior can not result
>in anything productive and it just makes this newsgroup that much less
>pleasant to read. Hence, my posts. Think about it.
Well, I try.
Nazi propaganda. And when someone posts such stuff, and note, if this
one is a German, then I'm ready to talk with this guy seriously. And I
don't want to be disturbed by Ed, who didn't even read what's all about,
but who then insulted me for having criticised Czub's nazi stuff. I dont
want to be disturbed by *you*, little boy, wishing me dead and some
stuff like this. Can't you see this through??
Czub nazi songs ° Tueschen flames him ° Eddie insults Tueschen °
Tueschen attacks Eddie ° ----- Tom attacks Tueschen for having flamed Ed
Schroder, Tom thinks it would be better if Tueschen was dead (!) or Tom
wanted to shoot Tueschen's knees. Hey Tom, youre not connected to Lt
Calley somehow? If you never heard of, visit a local library, ok?
You may be good in maths. But applying logic to real life situations,
isn't your thing yet. Just my 24 cts.
>Cheers,
>Tom
Dan Kirkland
In article <8p6k9ho...@Eng.Sun.COM> Mike Coffin,
<m...@Eng.Sun.COM> writes:
>George: now that you've had a lesson in netiquette from Tom Kerrigan,
>you are probably ready to proceed to your marriage counciling session
>with O.J. Simpson and sportsmanship lessons from Mike Tyson.
Okay, now that I have been able to pick myself up from the floor...
This was VERY funny Mike... (And sadly, right on target...)
dan
Tom Kerrigan
Rolf Tueschen (TUESCHEN.MEDIZ...@t-online.de) wrote:
: (Just to be precisely. I'm talking about the same Tom kid. He's the only
: American demonstrating such an ugly face in rgcc. We Germans have Czub.
: I think you have to choose the right moment to intervene. And homicide
: is better suited than simple shut ups. I was the addressee in the other
: case. Therefore my 23 cts. Not to flame you...)
Yes, yes, we know who you're talking about. The problem here is that this
is a newsgroup for computer chess, and I'd like to see interesting
conversation and information exchange going on. More often than not, I
see crap. You, for example, going WAY off-topic to personally attack Ed,
which simply WAS NOT warranted. This guy calling the DB work BS because
of his own flawed math. KK writing huge posts about programming aspects
he clearly doesn't understand. Czub brining every single thing he sees
on television into the discussions. All of this behavior can not result
in anything productive and it just makes this newsgroup that much less
pleasant to read. Hence, my posts. Think about it.
Cheers,
Tom
Rickard Westman
In article <33F30990...@eecs.umich.edu>, George R. Barrett wrote:
>> : Well, let's do the math and find out if the Deep Blue team is full of
>> : it...
>>
>> : Let n = number of ply (half moves), then for 36 moves per position this
>> : gives:
>>
>> : 36^n = 200e6 x 60 (seconds/minute) x 3 (minutes)
>>
>> you left out alpha/beta. This is a pure minimax tree size. Alpha
>> beta will produce trees of the following side:
>
>I've gotten at least two replys that say I left out alpha/beta.
>No, I didn't. Well not intentionally. I understood the original poster
>to say that the Deep Blue team claimed 12-14 plys at full width.
There is no contradiction between performing a full width search
and implementing alpha-beta pruning. Alpha-beta pruning is an
exact technique which only prunes nodes that cannot change the
outcome of the minimax search.
Many find it hard to believe that some nodes in the tree can be
ignored altogether without degrading the quality of the search, but
it can easily be demonstrated by an example: If a naive minimax
search would lead to the computation of
min(max(1,2,3),max(1,6,A,B,C),max(3,D,E,F)),
the alpha-beta algorithm would avoid the computation of variables A
through F, since the final result of the computation would be 3 no
matter what value these variables had. Avoiding those computations
could have required a search of a very deep tree, so the savings
are substantial. In the optimal case, you can almost double the
search depth by using alpha-beta pruning.
>Tom Kerrigan telling me to "shut up" is just rude.
>To the others whose comments simply pointed out my
>misunderstanding of the claim, thanks.
I think your understanding of the claim was just fine. But your
assumption that every tree node needs to be investigated in a full
width search was incorrect. Now, Tom was rude, but it is also rude
to post disparaging remarks about people without being sure they
deserve it. Miss Manners says: What goes around comes around.
--
Rickard Westman <ri...@ida.liu.se>
"Beware of the panacea peddlers: Just because you
wind up naked doesn't make you an emperor."
- Michael A Padlipsky
mclane
kerr...@deimos.frii.com (Tom Kerrigan) wrote:
> We've seen all sorts of posts about them from you and keep
>trying to explain them to you but you just don't get it. More than not
>getting it, you confuse and misquote everything, apparently so you can
>throw accusations around. Or how about asymmetry? You clearly have no idea
>what asymmetry is, but you think it's bad because of something vague
>Vincent (noted loser) posted months ago.
That is what I try to tell him for a couple of weeks now. He claims
something (e.g. about CSTal) without explaining what HE understands
under the word asymmetrie. I have no idea what he understands under
asymmetrie. I have tried to find out by reading his posts. But I don't
understand them semantically. I thought this had something to do with
the fact that my english is not the best, but now you tell me, you
don't understand it either. So it is not the language, it is something
else.
>I'd have to say that you're a fairly useful computer chess guy. You provide
>all sorts of technical support for a dozen programs, and my guess is that
>your lists help out all sorts of people (even if I don't think they should
>be posted on the newsgroup). The problem is that you just don't know as
>much as you think you do, and that just really annoys a lot of people here.
My problem is: I am now attacked by him only because I posted that I
don't read his posts because they are boring/redundant and wrong to
me, from my subjective point of view.
Hell - all this shit mosquito-talking. Why can some people only always
differenciate between friends / enemies. I am not KK's enemy.
I don't understand why he has to call anybody who says something
against him as mosquito's or something alike. Maybe Tom we should not
waste so much time fighting against Newton, Tueschen, KK and and and !
Skipping their posts should be a better idea than really discussing
with them. It makes no sense at all.
They will not understand you. They will not see your comments neutral.
They see an insult in anything. And they cannot stand losing. Childish
reactions appear and this goes over and over and over.
Stefan Meyer-Kahlen has come back out of his 6 weeks holiday and will
maybe now join us here.
I told him about the mega-shredder/virtual-thread.
He will maybe reread the stuff and comment on it. But what if Bob does
not believe in Stefans opinon ? :-)
>Cheers,
>Tom
Tom - do we see us in Paris ??
That would be nice.
Jonathan Berry
In article <5t8a2j$9mf$1...@europa.frii.com>,
kerr...@deimos.frii.com (Tom Kerrigan) wrote:
:This Mirror Scandal thing was your idea, wasn't it?
Tom is addressing KK = Komputer Korner (= Al) here.
:Every programmer thinks it's a total non-issue, and can come
:up with dozens of explanations for programs failing the "mirror
:principle" in his sleep, but you think that some sort of
:scandal is going on.
The basis of Science is reproducibility. KK has presented
instances which are identical except trivially, yet most
programs treat them differently. I agree with Tom's conclusion
that it is not a scandal (unless the move recommended is
different), but the discussion of the subject has been
thought-provoking.
: [snip]
I've noticed that I do better against some programs (e.g.,
Rebel Decade) when I play Black. OK, William of Occam would
suggest that that the most likely explanation is that J.Berry
prefers to play Black, but is it possible that some programs
are mildly or even unknowingly optimized to play Black, because
that's the colour the computer gets by default? And might not
such an observation help a programmer or two?
--
cheers,
Jonathan
Komputer Korner
Tom Kerrigan wrote:
> . KK writing huge posts about programming aspects
> he clearly doesn't understand. Czub brining every single thing he sees
>
> on television into the discussions. All of this behavior can not
> result
> in anything productive and it just makes this newsgroup that much less
>
> pleasant to read. Hence, my posts. Think about it.
>
> Cheers,
> Tom
I would like to know what HUGE post I have written about a programming
aspect that I didn't understand. I have made some statements that were
quickly corrected and I certainly don't post on programming details as I
have said many times that I am not a computer chess programmer.
Tom Kerrigan
This Mirror Scandal thing was your idea, wasn't it? Every programmer thinks
it's a total non-issue, and can come up with dozens of explanations for
programs failing the "mirror principle" in his sleep, but you think that
hash tables. We've seen all sorts of posts about them from you and keep
trying to explain them to you but you just don't get it. More than not
getting it, you confuse and misquote everything, apparently so you can
throw accusations around. Or how about asymmetry? You clearly have no idea
what asymmetry is, but you think it's bad because of something vague
Vincent (noted loser) posted months ago.
I'd have to say that you're a fairly useful computer chess guy. You provide
all sorts of technical support for a dozen programs, and my guess is that
your lists help out all sorts of people (even if I don't think they should
be posted on the newsgroup). The problem is that you just don't know as
much as you think you do, and that just really annoys a lot of people here.
Cheers,
Tom
Komputer Korner (kor...@netcom.ca) wrote:
Komputer Korner
Tom Kerrigan wrote:
> This Mirror Scandal thing was your idea, wasn't it? Every programmer
> thinks
> it's a total non-issue, and can come up with dozens of explanations
> for
> programs failing the "mirror principle" in his sleep, but you think
> that
> some sort of scandal is going on. If you don't like that example, try
> hash tables. We've seen all sorts of posts about them from you and
> keep
> trying to explain them to you but you just don't get it. More than not
>
> getting it, you confuse and misquote everything, apparently so you can
>
> throw accusations around. Or how about asymmetry? You clearly have no
> idea
> what asymmetry is, but you think it's bad because of something vague
> Vincent (noted loser) posted months ago.
>
Since I was the first one to jump on the asymmetry issue and have had
it explained to me lots of times, I am fairly sure I understand it
enough to argue that it is a patch for evaluation functions that aren't
good enough. The danger of asymmetry is that if programmers don't
temporarily take out the asymmetry before they test new evaluation
function code, they skew their evauation function. They have to take out
their asymmetry patch and then do the testing and then put in a new
asymmetry patch back in. I wonder how many do this? And for some
programs like CSTAL this is impossible, because asymmetry is everywhere
in the code. The issue for the user depending on the evaluation score to
judge a position is another serious one and that is why I argue that the
analysis mode should always be asymmetry free. Users want to know what
their program thinks in 100's of a pawn untainted by an asymmetry patch.
Some people never look at the evaluation score but a lot of others do.
The mirror scandal is perhaps unavoidable with hash tables and those
programs using bit mapping. My only point was that it seemed logical to
assume that a program would have the same absolute value evaluation
looking at the same position but with the opposite side to move if the
position was a mirror of the other as in 1.e4e5 as compared to 1.e3e5
2.e4 The hash table issue was confused by Chris Whittington's
"drunken sailor" post but as it turns out, having an overloaded hash
table is no worse than not having a hash table at all, since a fall back
move ordering algorithm kicks in whenever a hash table gets overloaded.
I don't pretend to know one thousandth as much you programmers know. All
I am doing is seeking the truth and relaying it to the great "unwashed".
mclane
Komputer Korner <kor...@netcom.ca> wrote:
> And for some
>programs like CSTAL this is impossible, because asymmetry is everywhere
>in the code.
What makes me worried is, Chris and I have the code, and you claim
something in a very strong way. If you would claim it having the code,
but without it ....
>The issue for the user depending on the evaluation score to
>judge a position is another serious one and that is why I argue that the
>analysis mode should always be asymmetry free.
You should maybe send this request to Richard Lang ....
>Users want to know what
>their program thinks in 100's of a pawn untainted by an asymmetry patch.
>Some people never look at the evaluation score but a lot of others do.
>The mirror scandal is perhaps unavoidable with hash tables and those
>programs using bit mapping. My only point was that it seemed logical to
>assume that a program would have the same absolute value evaluation
>looking at the same position but with the opposite side to move if the
>position was a mirror of the other as in 1.e4e5 as compared to 1.e3e5
>2.e4
Logical for somebody not having the experience that this is NOT the
case for a couple of years.
Since 1985 Mephisto Amsterdam you had been able to verify that
there are some "unimportant" world-chess-champion programs who use
asymmetrie. They win almost any championship with this asymmetrie.
And the users have to live with langs not working analysis-mode since
1985.
The reason we have not mentioned this here is, we knew it since we
bought/tested the amsterdam 1985.
So for us it was/still is a normal thing.
> All
>I am doing is seeking the truth and relaying it to the great "unwashed".
!! Big words...
chrisw
--
http://www.demon.co.uk/oxford-soft
Komputer Korner <kor...@netcom.ca> wrote in article
<33F82E74...@netcom.ca>...
> Tom Kerrigan wrote:
>
> > This Mirror Scandal thing was your idea, wasn't it? Every programmer
> > thinks
> > it's a total non-issue, and can come up with dozens of explanations
> > for
> > programs failing the "mirror principle" in his sleep, but you think
> > that
> > some sort of scandal is going on. If you don't like that example, try
> > hash tables. We've seen all sorts of posts about them from you and
> > keep
> > trying to explain them to you but you just don't get it. More than not
> >
> > getting it, you confuse and misquote everything, apparently so you can
> >
> > throw accusations around. Or how about asymmetry? You clearly have no
> > idea
> > what asymmetry is, but you think it's bad because of something vague
> > Vincent (noted loser) posted months ago.
> >
>
> Since I was the first one to jump on the asymmetry issue and have had
> it explained to me lots of times, I am fairly sure I understand it
> enough to argue that it is a patch for evaluation functions that aren't
> good enough.
Total bollocks.
> The danger of asymmetry is that if programmers don't
> temporarily take out the asymmetry before they test new evaluation
> function code, they skew their evauation function.
More bollocks.
> They have to take out
> their asymmetry patch and then do the testing and then put in a new
> asymmetry patch back in.
Yet more bollocks.
> I wonder how many do this?
Ah, some quantification bollocks, KK's favourite.
Exactly 4.566342526262 programmers, probably.
> And for some
> programs like CSTAL this is impossible, because asymmetry is everywhere
> in the code.
Aaarrrggghhh. More assertions of a bollock-ish nature.
> The issue for the user depending on the evaluation score to
> judge a position is another serious one and that is why I argue that the
> analysis mode should always be asymmetry free.
Korner. I want to know the bollocks percentage factor of your posts.
Is it 99.74894644% or 101.87377777 % ?
> Users want to know what
> their program thinks in 100's of a pawn untainted by an asymmetry patch.
Well the dumb materialistic users who think in decimal places and want long
numbers to assure themselves maybe do. But intelligent users realise that
the numbers are bollocks in the first place.
> Some people never look at the evaluation score but a lot of others do.
Try treating the evaluation score as a general guide that is probably
wrong.
> The mirror scandal is perhaps unavoidable with hash tables and those
> programs using bit mapping.
Bollocks !!
> My only point was that it seemed logical to
> assume that a program would have the same absolute value evaluation
Point ? Logic ? Don't believe you.
> looking at the same position but with the opposite side to move if the
> position was a mirror of the other as in 1.e4e5 as compared to 1.e3e5
> 2.e4
Bollocks.
> The hash table issue was confused by Chris Whittington's
> "drunken sailor" post
The only confusion came from you.
> but as it turns out, having an overloaded hash
> table is no worse than not having a hash table at all, since a fall back
> move ordering algorithm kicks in whenever a hash table gets overloaded.
I give up. What relevance is this ?
> I don't pretend to know one thousandth as much you programmers know.
Oh yes you do.
> All
> I am doing is seeking the truth
Like a drunken sailor ?
> and relaying it to the great "unwashed".
Have this weeks Kim il Sung memorial :)
Chris Whittington
Eduardo Suastegui
Rolf Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in article
<5t59n0$7ga$2...@news02.btx.dtag.de>...
> Czub nazi songs ° Tueschen flames him ° Eddie insults Tueschen °
> Tueschen attacks Eddie ° ----- Tom attacks Tueschen for having flamed Ed
> Schroder, Tom thinks it would be better if Tueschen was dead (!) or Tom
> wanted to shoot Tueschen's knees. Hey Tom, youre not connected to Lt
> Calley somehow? If you never heard of, visit a local library, ok?
Are you guys aware of the unwritten rule that when some one uses the term
"Nazi" in a posting the thread is officially over? I think this thread
thankfully qualifies.
Let's start over, shall we? Can't we all get along?
Ed Suastegui
--
http://home.earthlink.net/~esuastegui/eschess
"Once upon a time, chess players would gather at coffee houses.
Now, they ponder chess over a cup of Java."
Per Erik Manne
Jonathan Berry wrote:
>
> : [snip]
>
> I've noticed that I do better against some programs (e.g.,
> Rebel Decade) when I play Black. OK, William of Occam would
> suggest that that the most likely explanation is that J.Berry
> prefers to play Black, but is it possible that some programs
> are mildly or even unknowingly optimized to play Black, because
> that's the colour the computer gets by default? And might not
> such an observation help a programmer or two?
I've had similar experiences. My guess is that it's because
computers are better at defense and counterattack than taking
the initiative and doing something with it.
--
Bergen,
Per Manne
Rickard Westman
In article <4/09z4mip...@islandnet.com>, Jonathan Berry wrote:
>In article <5t8a2j$9mf$1...@europa.frii.com>,
>
>:This Mirror Scandal thing was your idea, wasn't it?
>
>
>:up with dozens of explanations for programs failing the "mirror
>:principle" in his sleep, but you think that some sort of
>:scandal is going on.
>
>instances which are identical except trivially, yet most
>programs treat them differently.
When you change something in an experiment you would normally
expect a different result. A trivial change in the setup often
leads to a trivial change in the result. In this case the trivial
change is entirely *optional* and there is no advantage whatsoever
of implementing it. So I don't see how this would be a problem for
those concerned with reproducability.
>I've noticed that I do better against some programs (e.g.,
>Rebel Decade) when I play Black. OK, William of Occam would
>suggest that that the most likely explanation is that J.Berry
>prefers to play Black, but is it possible that some programs
>are mildly or even unknowingly optimized to play Black, because
>that's the colour the computer gets by default?
The most likely explanation is probably that the opening book of
Rebel Decade suits your style of play better when you play black.
>And might not such an observation help a programmer or two?
Only if they have a special interest in having the program play
better as white against J.Berry.
mclane
Andreas Mader <Ma...@p6.gud.siemens.co.at> wrote:
> After a looooong holiday :) I checked rgcc and I first thought I was
>reading the wrong newsgroup :(
Hi Andreas. Don't be worried. Tueschen is doing the same thing as
usua, and we mainly ignore him. We have found out that this is the
best way. Of course he still repeats and repeats and claims his
whatever threads about ALL PEOPLE HERE ARE CRAZY, despite him
( in the time you were in holiday I saw hi doing it against Ingo,
Bruce, Tom, Enrique and Peter and Reno (do you remember Reno de Caro
and his articles in the 80ties in Computerschach & Spiele) I don't
know for what reason. Maybe because I like these people. Maybe he has
to attack people which I like. That would explain much. )
>Wasn't there a time with chess related topics here?
How ? KK started some ideas about how programs work. He got some echo
from programmers.
Also - as usual - Bob and I exchanged friendly wishes and regards, and
lot of other things. Still the Shredder/Virtual thread was a topic and
Stefan has come back out of holiday too. Maybe HE, the author can say
something about Shredder himself. I could lend him my virtual CD and
than he could comment on the playing strength of both programs.
Maybe, maybe , maybe.
Today I bought again 1 package of Nogger, 3 packages of Magnum and 1
package of another fruit-ice.
I hope this helps against the computer-caused-heat.
It is nice to see you back. I have posted the BS2830 and also again
the Michael Reiter test-suite out of PC-Schach.
We had a discussion about Bednorz' latest effort and Bruce / Bob asked
for a positional/strategical test suite.
I rememberd this old test from 1993.
I am really glad that you are back, so we have one well knowledged
participant here to discuss.
And - unbelievable, Bill Newton has a working Mark 5 and a working MGS
Chafitz !!! I am very jealous ! I declared him war and we will make a
game between his favourite machine (Mark5) and mine (Mephisto III S
68000).
Do you particpiate in Paris ? Do we see us there ??
>Andreas
Rolf Tueschen
"Eduardo Suastegui"
Hi, Eduardo. I must strongly oppose your idea. It might have been
outside your realm of attention. But you simply confuse matters. Please
read my explanation below. I'm sure you'll get the difference. Because
your concluding is 100% wrong.
BTW *best* greetings to sunny Spain. Besides *Enrique* *Irazequo* you're
the only Spaniard visiting this group.
<esc...@aol.com> wrote:
>Rolf Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in article
><5t59n0$7ga$2...@news02.btx.dtag.de>...
>> Czub nazi songs ° Tueschen flames him ° Eddie insults Tueschen °
>> Tueschen attacks Eddie ° ----- Tom attacks Tueschen for having flamed Ed
>> Schroder, Tom thinks it would be better if Tueschen was dead (!) or Tom
>> wanted to shoot Tueschen's knees. Hey Tom, youre not connected to Lt
>> Calley somehow? If you never heard of, visit a local library, ok?
>Are you guys aware of the unwritten rule that when some one uses the term
>"Nazi" in a posting the thread is officially over?
Maybe. But think about this precision!
Nobody -- you got this? Nobody, zero, called another poster here in rgcc
a "nazi". This is very important.
What I did, sorry, but it was me who started this analysis against the
german poster Czub, also called "maclane", was simply to analyse his
ideas of handling people of opposing oppinions here in Germany.
Czub proposed
° Constructing camps to be able to concentrate people like
Scientologists for instance to be able to control them
°° Signing people of opposing opinions with (arm) bandages to be
able to better and faster recognize in the streets / in public
Now, as a German who studied a lot about concentration camps, medical
experiments with those "concentrated" people, about killing them in the
end, and how this all began in Germany over 60 years ago ----
I was determined to have to warn against such "ideas" as mentioned
above. The nazis threw all people of opposing ideas into concentration
camps. But you must know that people with simply different heritage
(gypsies for instance) were trated like "opposing". All these people
were held like criminals in jail. The nazis also signed *all* Jews for
the public life before they were thrown out of Germany, thrown into
these camps and later were killed. But also mentally handicapped and
other ill persons were killed. It's such a long story, it cannot be told
here in a few seconds.
Ideas and methods which are, speaking of Germany, and remember Czub? and
myself we're Germans, the beginning of the whole historical nazi past.
Therefore I warned Czub to continue with these "nazi-like" propagations.
And I asked him to take back these insinuations. But he was deaf...
On the contrary a very famous programmer of this group, the dutch
businessman Ed Schroder jumped on me (!!), criticised me for my critic
against Czub and insulted me very disgustfully. Did you ever saw the
like?
I therefore attacked Schroder, the father of the famous chess program
"Rebel" of supporting the nazi-like postings of Czub. What should it be
meant for instead when Ed Schroder reacted so heavily?
But also Ed Schroder was never called a "nazi". No, he simply supported
Czub with his nazi-like propaganda. Period.
So, dear Eduardo, note for the next post, you hopefully write to this
group, check the past and the development of a thread very carefully
before you come along with a quick verdict. But you were right, simply
calling another opposing poster "a" "nazi", this should be banned from
usenet, simply because it spoils the discussion. But on the opposite
that cannot mean that someone is allowed to propagate ideas which are
like the nazi methods from long ago, *and* this one could not be
criticised, because you then come along and you state the end of the
thread.....
Hope, you got the differentiation.
Best greetings from 30- degree- Germany to how- much- Spain?? :)
Rolf Tueschen
>I think this thread
>thankfully qualifies.
>Let's start over, shall we? Can't we all get along?
Ed, definetly *not*. Read above. First this famous twin of you, Ed
Schroder, must take back his support for Czub in this question. And he
has to excuse for his insults agains me. Ok, Ed? :)
Andreas Mader
Eduardo Suastegui wrote:
> Rolf Tueschen <TUESCHEN.MEDIZ...@t-online.de> wrote in
> article
> <5t59n0$7ga$2...@news02.btx.dtag.de>...
>
> > Czub nazi songs ° Tueschen flames him ° Eddie insults Tueschen °
> > Tueschen attacks Eddie ° ----- Tom attacks Tueschen for having
> flamed Ed
> > Schroder, Tom thinks it would be better if Tueschen was dead (!) or
> Tom
> > wanted to shoot Tueschen's knees. Hey Tom, youre not connected to Lt
>
> > Calley somehow? If you never heard of, visit a local library, ok?
>
> Are you guys aware of the unwritten rule that when some one uses the
> term
> "Nazi" in a posting the thread is officially over? I think this thread
>
> thankfully qualifies.
>
> Let's start over, shall we? Can't we all get along?
>
> Ed Suastegui
>
> --
> http://home.earthlink.net/~esuastegui/eschess
> "Once upon a time, chess players would gather at coffee houses.
> Now, they ponder chess over a cup of Java."
After a looooong holiday :) I checked rgcc and I first thought I was
reading the wrong newsgroup :(
Wasn't there a time with chess related topics here?
Andreas
Tom Kerrigan
Komputer Korner (kor...@netcom.ca) wrote:
: Since I was the first one to jump on the asymmetry issue and have had
: it explained to me lots of times, I am fairly sure I understand it
: enough to argue that it is a patch for evaluation functions that aren't
Well, this shows once again that you have no idea what you're talking about.
Asymmetry is simply another term in an evaluation function, like having a
bonus for a knight. It assumes (for programs I know, anyway) that the
program can't attack as well as the opponent, which is just as valid as
assuming that having an extra knight is good. Evaluations are simply
collections of these assumptions. You can call asymmetry a "patch" as much
as you can call a knight bonus a "patch."
: good enough. The danger of asymmetry is that if programmers don't
: temporarily take out the asymmetry before they test new evaluation
: function code, they skew their evauation function. They have to take out
Where did you pull this out of? Is that place warm and dark?
: 2.e4 The hash table issue was confused by Chris Whittington's
: "drunken sailor" post but as it turns out, having an overloaded hash
His post didn't confuse me. BTW, programs do not have any silly backup
move ordering scheme. They just play the hash move first, if possible.
: I don't pretend to know one thousandth as much you programmers know. All
: I am doing is seeking the truth and relaying it to the great "unwashed".
If the great unwashed are so hot about learning about chess programming,
why don't they just ask a programmer? Just forward their questions to me,
I'll be happy to answer them.
Cheers,
Tom
mclane
TUESCHEN.MEDIZ...@t-online.de (Rolf Tueschen) wrote:
>mcl...@prima.ruhr.de (mclane) wrote:
>>Andreas Mader <Ma...@p6.gud.siemens.co.at> wrote:
>>> After a looooong holiday :) I checked rgcc and I first thought I was
>>>reading the wrong newsgroup :(
>>Hi Andreas. Don't be worried. Tueschen is doing the same thing as
>>usua, and we mainly ignore him. We have found out that this is the
>>best way. Of course he still repeats and repeats and claims his
>>whatever threads about ALL PEOPLE HERE ARE CRAZY,
>Correction:
>Against tis typical Czub lie and cheat I hereby declare
>that I never attacked someone of this group as being "crazy".
No , you THOUGHT you are normal and THEY are crazy.
Your comments about those people here shows EXACTLY and by evidence
/facts (!!) that you really believe all these people are strange and
wrong meanwhile YOU just correct them, or give them a wise hint, or
help them so see better through. Rolf - all those people you comment
here in such a shameful way are professionals ! They do know more
about chess than you. So why do you attack them anyway ? You describe
it, the one is a kid, the other is a spanish guy, the other is from US
and claims you use Bild-Zeitung-style (and you doubt he can judge
because only YOU as a german knows about Bild-Zeitung ?)...
Don't you see that your comments about these honorable people proves
my point that you are crazy ?
Only a fool could believe so many people are wrong and himself is
right.
>If I would think about something like this, Czub came into serious
>consideration of course. :)
>>despite him
>>( in the time you were in holiday I saw hi doing it against Ingo,
>>Bruce, Tom, Enrique and Peter and Reno
>Correction:
>As said above none of these were called crazy from my side.
>a) "Ingo"
>This poster was *asked* for a long time about details of his Dreihirn.
>He *never* clarified this spooky cyborg here in rgcc. PERIOD.
>On the cost of public who pays taxes however he takes part here in rgcc
>to shout his repeated impostor cries for sponsor money....
>This will be checked. Period.
On the cost of public who pays taxes, he takes part here in rgcc ?
Shall we really start such a silly discussion and publish on whose
costs you live your life ???
Is this really YOUR serious critics against Dr. Ingo Althoefer who has
given us many data and results about 3-Hirn ????
You are crazy. And the bad thing with it is: you don't register it.
>b) "Bruce"
>My sole critic was that he tries to back Schroder and even Czub to sue
>me. For the sake of those two I begged him to rethink this. And for all
>to think about correct critic against Czub for his nazi-like propagating
>here in rgcc. Rolf (C) and myself were the only ones who did criticise
>him!
Bruce has shown with all his posts that he is able to fight for
himself and that he is a very individual person.
He has no reason to back me.
Believe me - he is a person himself and does not need to back me.
But you don't like that he comes to similar judgement concerning you
than me and ed ? That is what you don''t like.
>c) "Tom"
>I criticised Tom for publishing *death*-wishes and ideas about shooting
>me in my knees here in this group rgcc.
>No one did criticise this kid for doing this
ever thought why ???
> but when he told someone to
>"shut up" brucemo came on scene immediately and informed him to calm
>down... Unbelievable.
Tom is a nice guy. And you are "younger". And no inch wiser. And i
have seen no program nor any computer-chess-concernoing statement here
from you, despite attacks against any kind of people.
>d) "Enrique"
>This guy who gives his vita as a spanish university professor is in my
>eyes more a domestic help for Schroder but not a genuine poster.
>Mostly he steps right in to back Schroder and Czub in their rantings.
>But I never attacked even him to be crazy. If he's making a living like
>this?
Look - Enrique lives also his own life. But your paranoia thinks he is
more a domestic help for schroeder.
Of course Enrique backs me and Schroeder.
And do you know why ?
Because we have come to the same conclusion about you.
And your comments here show me that you are not able to talk in any
civilized order.
>e) "Peter"
>I don't know who's Peter. But I know that I also didn't call a Peter
>crazy.
The Mchess-operator Peter Schreiner. You forgot your statements about
him ? Hej - you have BSE too.
>f) "Reno"
>I don't know a guy called "Reno". He suddenly appeared and attacked me
>from the californian coast (!) for using a "BILD-Zeitung"-style.... :))
Nobody attacked you. He said you are using Bild-Zeitung-style.
If you would know anything about computerchess, you would remember a
Reno de Caro. But you have no memory for computer-chess, or ?
Only for who attacks you.
But - believe me - it is not important who attacks you.
>He pretended to read this boulevard blood and crime paper daily over
>there. :))
>So, everybody may find his one verdict about this clown. Or better a
>clone (of Czub) because he's almost exactly on Czub's uneducated level
>-- although he pretended to be an educated lecturer. :)
So you don't know Reno - but you know he is a clown and uneducated and
whatever. You insult somebody you don't know ? And only because he has
called your posts in the style of BILD-Zeitung ?
Ha - QED . your comments show exactly me thesis. You are ill Rolf.
>After insulting me here in public he approached by email and wrote about
>his wish to hear more from my side, from my life and so on. ROTFL.
Maybe he thought: this guy is ill, I will try to help him. And what
have YOU done to him for this nice human gesture he had done ?
You insult him...
Brilliant.
>Well, as Heinrich Boell once described one doesn't normally exchange
>emails with such a fishy guy.
I have read many Heinrich Boell-books. Which book do you speak about ?
Heinrich Boell would have never talked to a guy like you.
>But also even this guy I never called crazy.
>So stop this criminal posting, Czub
I shall stop criminal posting ? Read your own comments about the
persons here. I hope the others will also take their consequences
against you.
Maybe one law-suit is not enough and you can fight against 5 or 6 ones
instead of only one.
>BTW, you completely forgot this businessman Schroder who wanted to call
>me to court in Holland for having criticised you for this nazi-like
>propaganda.
Sorry. Ok. I forgot Ed. But he is able to defend himself the best...
>Till today he couldn't come closer to the point. I still wait for an
>official invitation. A train with 24 waggons is ready for departure in
>direction of the Netherland High Courts.
I hope they will not allow you to visit netherlands. It is such a nice
country.
>Mr. Mader should know that Schroder, the more or less papa of Rebel, one
>of the best micros of the *whole* world, this man insulted me very
>heavily for criticising Czub with his nazi-like propagating.
Indeed.
>I still can't understand it. Could you explain me, Mr. Mader?
When he speaks to you Rolf. Many people think they waste their time
talking to you.
>Is it right that you are this poster who falsely attacked me once
>because of time-out as a real newbie? Oh, *sorry*, If I could I wanted
>to take back my wish that you explained me such complicated stuff like
>Schroder's backing Czub for .... (well, you know what I mean.)
>Anyway, best greetings for coming back after your looong vacations.
>Rolf Tueschen
Andreas Mader
mclane wrote:
> And - unbelievable, Bill Newton has a working Mark 5 and a working MGS
>
> Chafitz !!! I am very jealous ! I declared him war and we will make a
> game between his favourite machine (Mark5) and mine (Mephisto III S
> 68000).
>
I sold my "Sargon 2.5" (along with the Steinitz module) a couple of
years ago... I think there was a 6502 processor with 2 MHz working
inside and the search depth on tournament level was 4-5 plies.... Man,
am I getting old! :)
> Do you particpiate in Paris ? Do we see us there ??
I am not going to Paris, a "Nimzo team" will not be there, too.
Instead, a "Hydra team" will show up, because Chrilly Donninger changed
the name of his child. This program is based on Nimzo 3.5, but most of
the routines are new and many new ideas had been implemented. Hydra will
only work with Win 95 (97, NT) and will not wotk under DOS. By now we
are twiddling with the positional parameters (e.g. the brand new
mobility evaluation) and there are many bugs in the program. Hydra is a
"usual" Donninger program: Very fast and (by now) dumb. I have a Pentium
225 MMX and Hydra is doing approx. 180 KNPS on this machine. Maybe this
is going to decrease when more knowledge is implemented. Chrilly also
had a plan for "CHE++", but I think due to time problems this is not
going to happen....
Andreas
Rolf Tueschen
mcl...@prima.ruhr.de (mclane) wrote:
>Andreas Mader <Ma...@p6.gud.siemens.co.at> wrote:
>> After a looooong holiday :) I checked rgcc and I first thought I was
>>reading the wrong newsgroup :(
>Hi Andreas. Don't be worried. Tueschen is doing the same thing as
>usua, and we mainly ignore him. We have found out that this is the
>best way. Of course he still repeats and repeats and claims his
>whatever threads about ALL PEOPLE HERE ARE CRAZY,
Correction:
Against tis typical Czub lie and cheat I hereby declare
that I never attacked someone of this group as being "crazy".
If I would think about something like this, Czub came into serious
consideration of course. :)
>despite him
>( in the time you were in holiday I saw hi doing it against Ingo,
>Bruce, Tom, Enrique and Peter and Reno
Correction:
As said above none of these were called crazy from my side.
a) "Ingo"
This poster was *asked* for a long time about details of his Dreihirn.
He *never* clarified this spooky cyborg here in rgcc. PERIOD.
On the cost of public who pays taxes however he takes part here in rgcc
to shout his repeated impostor cries for sponsor money....
This will be checked. Period.
b) "Bruce"
My sole critic was that he tries to back Schroder and even Czub to sue
me. For the sake of those two I begged him to rethink this. And for all
to think about correct critic against Czub for his nazi-like propagating
here in rgcc. Rolf (C) and myself were the only ones who did criticise
him!
c) "Tom"
I criticised Tom for publishing *death*-wishes and ideas about shooting
me in my knees here in this group rgcc.
No one did criticise this kid for doing this but when he told someone to
"shut up" brucemo came on scene immediately and informed him to calm
down... Unbelievable.
d) "Enrique"
This guy who gives his vita as a spanish university professor is in my
eyes more a domestic help for Schroder but not a genuine poster.
Mostly he steps right in to back Schroder and Czub in their rantings.
But I never attacked even him to be crazy. If he's making a living like
this?
e) "Peter"
I don't know who's Peter. But I know that I also didn't call a Peter
crazy.
f) "Reno"
I don't know a guy called "Reno". He suddenly appeared and attacked me
from the californian coast (!) for using a "BILD-Zeitung"-style.... :))
He pretended to read this boulevard blood and crime paper daily over
there. :))
So, everybody may find his one verdict about this clown. Or better a
clone (of Czub) because he's almost exactly on Czub's uneducated level
-- although he pretended to be an educated lecturer. :)
After insulting me here in public he approached by email and wrote about
his wish to hear more from my side, from my life and so on. ROTFL.
Well, as Heinrich Boell once described one doesn't normally exchange
emails with such a fishy guy.
But also even this guy I never called crazy.
So stop this criminal posting, Czub
>(do you remember Reno de Caro
>and his articles in the 80ties in Computerschach & Spiele) I don't
>know for what reason. Maybe because I like these people. Maybe he has
>to attack people which I like. That would explain much. )
(see above) :))
>>Wasn't there a time with chess related topics here?
>How ?
>Also - as usual - Bob and I exchanged friendly wishes and regards, and
>lot of other things.
Yeah, most of all were shouts that Bob should go back to school to be
able to follow your nazi-like propagation.
BTW, you completely forgot this businessman Schroder who wanted to call
me to court in Holland for having criticised you for this nazi-like
propaganda.
Till today he couldn't come closer to the point. I still wait for an
official invitation. A train with 24 waggons is ready for departure in
direction of the Netherland High Courts.
Mr. Mader should know that Schroder, the more or less papa of Rebel, one
of the best micros of the *whole* world, this man insulted me very
heavily for criticising Czub with his nazi-like propagating.
I still can't understand it. Could you explain me, Mr. Mader?
Is it right that you are this poster who falsely attacked me once
Tom Kerrigan
Rolf Tueschen (TUESCHEN.MEDIZ...@t-online.de) wrote:
: (Another talking trace is your shouting "stop" when this kid Kerrigan
: mumbled things like "Shut up". But when he talks about death and knee
: shooting --- then you take a medium vacation. This is childish. And
: because you still deny the simple fact, you are for me from now on the
: second kid here in rgcc. With son or without.)
I never said a thing about knee shooting. Just taking them out with hammers.
Learn English already.
: Czub's idea appeared in a thread about scientologists, that's right. But
: *therefore* my opposing his idea of building camps to concentrate those
: people is not wrong as such. Not the scientologists interest me, but
Dear Lord God, you want Bruce to protest everything you protest. Maybe he
doesn't read all the posts you do, eh? Maybe his news server broke and he
never got the posts in question? Therefore he's your mortal enemy? Sorry,
I mean a kid with children? It must be very hard for you to make friends.
: If 2+2 are almost 4, you famous Ferret author, then Ed Schroder took
: sides for Czub's ideas. Is this clear? Why should he attack me then? If
: he also were against Czub's nazi-like rantings??????
He attacked you because you flame him every waking moment, is my guess,
and I hope this case goes to trial (or whatever cases do in Holland)
because I'm sure Ed could kick your ass.
Cheers,
Tom
Komputer Korner
If asymmetry was just another term in the evaluation function then it would
not skew the evaluation scores for one side vs the other. .
Asymmetry for king safety is a patch put in after variables like king
saftey have been already been calculated. Asymmetry causes different score
evaluations on the same PV depending on whose side it is to move , assuming
the computer always takes one side when doing the calculating. Your example
of assuming that one side can't attack as well as the other produces the
unequal evaluation scores that asymmetry produces. If the evaluation
function was good enough to begin with, you wouldn't need to make the above
assumption. . I am not saying that the concept is invalid. All I am saying
is that when the program is in analysis mode and giving out score
evaluations, the asymmetry should be turned off. If programmers leave in
the asymmetry when testing they run the danger of skewing the evaluation
function when changes to things like king safety terms have been made.
There is another consideration. When computer speeds get fast enough to
make asymmetry unnecessary for terms like king safety, then leaving the
asymmetry in causes the program to play vastly different than without it
and perhaps weaker in the long run.
On your 2nd point of programs just playing the hash move first without any
move ordering scheme to back up an overloaded hash table; that would
certainly produce the drunken sailor that Chris was talking about.
--
--
Komputer Korner
The inkompetent komputer
Tom Kerrigan <kerr...@deimos.frii.com> wrote in article
<5tfmet$scd$2...@europa.frii.com>...
Andreas Mader
Ingo Althoefer wrote:
> When will Hydra be commercially available ? Will it also have a k-best
> mode ?
>
I don't know the release date, but I think it will be this year.
How should a "k-best mode" work? What is the difference between the
"Fernschach mode" (that is already a feature of Nimzo 3 and 3.5) and a
"k-best mode"?
Best wishes
Andreas Mader
Bill Newton
In article <01bcac16$622e7100$c308...@cpsoft.demon.co.uk>,
chrisw <chr...@cpsoft.demon.co.uk> writes
>Total bollocks.
>More bollocks.
>Yet more bollocks.
>Ah, some quantification bollocks, KK's favourite.
>Bollocks.
These thoughtful quotes were snipped from a single posting authored
by Chris W.
IMO his words filled a much needed gap.
Try a cup of tea and two valium prior to future postings Chris. That
should clear the red mist! ;)
Regards.
--
Bill Newton
Bill Newton
In article <5tg77l$dgh$1...@europa.frii.com>, Tom Kerrigan
<kerr...@deimos.frii.com> writes
>Rolf Tueschen (TUESCHEN.MEDIZ...@t-online.de) wrote:
>This is childish. And
>: because you still deny the simple fact, you are for me from now on the
>: second kid here in rgcc. With son or without.)
>
>I never said a thing about knee shooting. Just taking them out with
>hammers.
>Learn English already.
Tom, comments like your's are juvenile and only serve the purpose
of highlighting a rather less than pleasant aspect of your outlook on
life.
Tis about time you demonstrated a little maturity methinks.
( If I wasn't such a pleasant chap I'd probably be shouting at you and
telling you to 'Grow up' ) :)
>He attacked you because you flame him every waking moment, is my guess,
>and I hope this case goes to trial (or whatever cases do in Holland)
>because I'm sure Ed could kick your ass.
I do not condone the comments attributed to Rolf T in his postings
to Ed. But then again I do not condone Ed's action in taking matters
to court.
Why?
Because Ed is one of the 'top guns' in this group. He is highly
regarded and has nothing to prove or dis-prove.
It maybe significant that a number of other highly regarded
contributors to this group have, in the past, suggested that
Ed should simply ignore Rolf Ts postings if they upset him so much.
IMO that's the only action that has ever been required.
All this court action will do is put money into Lawyer's pockets.
More significantly, whatever happens in court will not alter one jot
the existing good / bad / indifferent opinion, folk already have of both
gentlemen.
So If I were Ed, I'd stop the court action.
And If I were Rolf, I wouldn't hold my breath waiting for an apology
from Ed.
And if they still felt like lightening their wallets, I suspect their
favourite charities could benefit from appropriate donations.
Regards.
--
Bill Newton
Rolf Tueschen
In former thread "How can Deep Blue lose?"
mcl...@prima.ruhr.de (mclane) wrote:
>TUESCHEN.MEDIZ...@t-online.de (Rolf Tueschen) wrote:
>>mcl...@prima.ruhr.de (mclane) wrote:
>>>Andreas Mader <Ma...@p6.gud.siemens.co.at> wrote:
>>>> After a looooong holiday :) I checked rgcc and I first thought I was
>>>>reading the wrong newsgroup :(
>>>Hi Andreas. Don't be worried. Tueschen is doing the same thing as
>>>usua, and we mainly ignore him. We have found out that this is the
>>>best way. Of course he still repeats and repeats and claims his
>>>whatever threads about ALL PEOPLE HERE ARE CRAZY,
>>Correction:
>>Against tis typical Czub lie and cheat I hereby declare
>>that I never attacked someone of this group as being "crazy".
>No , you THOUGHT you are normal and THEY are crazy.
>Your comments about those people here shows EXACTLY and by evidence
>/facts (!!) that you really believe all these people are strange and
>wrong meanwhile YOU just correct them, or give them a wise hint, or
>help them so see better through. Rolf - all those people you comment
>here in such a shameful way are professionals !
I never knew that these people were pro's, that means they were paid for
behaving like this.
Therefore I really do apologize to them and you especially. Hope that I
don't have to go into jail or something? I promise never to think such
wrong (?) thoughts again. They are pro. Basta.
And for helping me out of this jungle I hereby give you one of my rare
Gold Medals. For the best 'Look Through' of the week. Thanks a lot.
>They do know more
>about chess than you. So why do you attack them anyway ? You describe
>it, the one is a kid, the other is a spanish guy,
Sorry, if I ask. But didn't I write "fly"? Spanish *f l y*.
Anyway. I will not continue to correct you in your post below. Because
you are the pro. I'm just a normal IGM.
I will just give a tiny commentary on the Ed.
>>Correction:
>>a) "Ingo"
>>b) "Bruce"
>>c) "Tom"
>ever thought why ???
>>d) "Enrique"
>>e) "Peter"
>>f) "Reno"
ROTFL
DOING THE ED (=DANCING ON BOTH NIPPLES!)
Steffen A. Jakob
lonni...@connect.com (Lonnie ) writes:
> |Instead, a "Hydra team" will show up, because Chrilly Donninger changed
> |the name of his child. This program is based on Nimzo 3.5, but most of
> |the routines are new and many new ideas had been implemented. Hydra will
> |only work with Win 95 (97, NT) and will not wotk under DOS. By now we
> |are twiddling with the positional parameters (e.g. the brand new
> |mobility evaluation) and there are many bugs in the program. Hydra is a
> |"usual" Donninger program: Very fast and (by now) dumb. I have a Pentium
> |225 MMX and Hydra is doing approx. 180 KNPS on this machine. Maybe this
> |is going to decrease when more knowledge is implemented. Chrilly also
> |had a plan for "CHE++", but I think due to time problems this is not
> |going to happen....
>
> Andrea,
> Why doesn't Chilly put a dos ver on CD and do what CG5 did as well? u run a chess
> program through winDOZE u inadvertently run into problems plus the performance loss.
Will "Hydra" be the direct follower of Nimzo (why did Donninger change
the name?), or will "Hydra" come with some Champ-like multimedia
Schnickschnack, too? When will it be available? Are there any
improvements concerning the learning features?
Best wishes,
Yobes.
--
Steffen A. Jakob |"Die Ratten versenken das
s...@gams.at | stinkende Schiff."
+43 1 8176230-18 | (Regina Klotz)
http://www.jakob.at/ |
Andreas Mader
Steffen A. Jakob wrote:
> Will "Hydra" be the direct follower of Nimzo
Good question.... I don't know. Maybe there will be a Nimzo 4 some day,
maybe not....
> (why did Donninger change the name?),
Chrilly Donninger had the famous "Hydra" with the nine heads in mind
when he named his program. I am not happy with this name, too, but names
are not important!
> or will "Hydra" come with some Champ-like multimedia
> Schnickschnack, too?
No multimedia "Schnickschnack" (as far as I know.....)!
> When will it be available?
Available when it is ready! :) (I don't know the date, maybe this year!)
> Are there any
> improvements concerning the learning features?
>
I know Chrilly's new ideas concerning the learning features, but I don't
know if they will be implemented. Sorry that I cannot tell you more
about it... :)
Best wishes
Andreas
Steffen A. Jakob
Hi Andreas!
Andreas Mader <Mader@*p6*.gud.siemens.co.at> writes:
> > Will "Hydra" be the direct follower of Nimzo
>
> Good question.... I don't know. Maybe there will be a Nimzo 4 some day,
> maybe not....
> > (why did Donninger change the name?),
>
> Chrilly Donninger had the famous "Hydra" with the nine heads in mind
> when he named his program. I am not happy with this name, too, but names
> are not important!
I don't care about the name, I only wanted to know if the program will
be Nimzo-like or Champ-like.
> > or will "Hydra" come with some Champ-like multimedia
> > Schnickschnack, too?
>
> No multimedia "Schnickschnack" (as far as I know.....)!
So we can hope that Hydra will be a "better Nimzo". Great!
> > When will it be available?
>
> Available when it is ready! :) (I don't know the date, maybe this year!)
A clear statement! :)
> > Are there any
> > improvements concerning the learning features?
>
> I know Chrilly's new ideas concerning the learning features, but I don't
> know if they will be implemented. Sorry that I cannot tell you more
> about it... :)
You gave me at least the information that Donninger is working on the
learning feature. I didnt expect more.
> Best wishes
> Andreas
^Andreas^Yobes
brucemo
Rolf Tueschen wrote:
>
> [This research result was so important that I created a new threat.
> Formerly it was Re: How Deep Blue could lose...]
>
> brucemo <bru...@seanet.com> wrote:
> >I didn't suggest that Thorsten sue you. I suggested that it would be very
> >strange if he did sue you.
>
> Sorry, evidence speaks another language.
>
> ° Your non-commenting Czub's nazi-like rants *in combination* with
>
> °° your backing of Schroder who from his side backed Czub and insulted
> me for having criticised Czub, *in combination* with
>
> °°° your critic against *me*, who was the only one who stood up and told
>
> Czup "Stop, until here and no more further."
A few years ago I was walking down a sidewalk in a not very nice part of
Seattle, and some guy, who was leaning on a building up ahead of me, began
staggering in my direction, scowling and muttering.
As he got closer to me, he suddenly belched out an obscenity, picked up a
(full) garbage can, and threw it at me, straining with effort, reddened eyes
bulging, sweat- and grime- encrusted hair sticking out in all directions, a
string of drool dripping down onto his greasy shirt.
He missed.
I said, "Jesus Christ! Are you nuts?", and walked away before he could find
another garbage can.
Since these two situations share some similarities, I believe that I will walk
away from this one as well.
Oh, I forgot.
Jesus Christ! Are you nuts?
bruce
Andreas Mader
Lonnie wrote:
> I think Hydra is pretty cool sounding. I remember "Gideon Pro" then
> all of a sudden
> it was Rebel I lived through it :>)
In fact, Hydra was supposed to be a multiprocessor program at first.
Therefore the name "Hydra" ("many" heads -> many processors you have to
battle with). In Paris a "normal" single processor Hydra will
participate. (I hope there is no program called "Herakles" there! :))
> |I know Chrilly's new ideas concerning the learning features, but I
> don't
> |know if they will be implemented. Sorry that I cannot tell you more
> |about it... :)
> |
> What is this?! Unfulfilled foreplay,please tell us!
>
1) "Nicht ueber ungelegte Eier gackern" (I am sure there is an English
phrase for this, too!) I do not want to promise something that maybe
isn't there in the end.
2) I do not have the right to tell everybody Chrilly's new ideas.
Best wishes
Andreas
Andreas Mader
Komputer Korner wrote:
> The difference between Fernschach/correspondence mode and K variation
> mode
> is that in K variation mode you can set the number of main PV's that
> the
> computer will search and return a score, whereas in
> Fernschach/correspondence mode you have to choose the PV's that you
> want
> analyzed. The problem is how do you know which ones are the best?
> Therefore
> K variation mode is much superior to the Fernschach/correspondence
> mode.
>
I think you are wrong!
Check out Nimzo's correspondence mode! You can EITHER set the number of
main PV's OR chose the PV's! You can also exclude some PV's and search
all others! Or you can say: I want Nimzo to search the best 4 PV's AND
e2-e4. And so on and so on....
Best wishes
Andreas
Andreas Mader
Ingo Althoefer wrote:
> Andreas Mader wrote:
> >Ingo Althoefer wrote:
> >> When will Hydra be commercially available ? Will it also have a
> k-best
> >> mode ?
>
> >...
> >How should a "k-best mode" work? What is the difference between the
> >"Fernschach mode" (that is already a feature of Nimzo 3 and 3.5) and
> a
> >"k-best mode"?
>
> k-best mode is meant for the infinite analysing level. ( In Nimzo's
> Fernschach
> mode the times have to be preset. ) In k-best the program is making
> iterative
> deepening. In each iteration the k best moves are computed ( with help
> of a
> modified alpha-beta algorithm ).
>
> The best format for the screen is as follows ( in a fictive example
> with k=3 ):
> ( In this example high values shall begood for the player to move. )
>
> time eval principal line
> 0:57 +0.36 Nf3, Nc6; Bb5, a6; Bxc6, dxc6; Nxe5,
> 1:13 +0.21 Bc4, Nf6; d4, Nxe4; Nf3, Nc6;
> ========================================================
> 0:42 +0.28 f4, exf4; Bc4, Nf6; e5, Ne4; Nf3,
>
> The lines above the ...=======... are the first two results from the
> new
> iteration whereas the one below is still from the previous one.
> On a "usual" screen you do not need such a dashed line. Simply give
> the
> proposals from the new iteration in black and the ones from the old
> iteration
> in grey. By doing this you can omit the search depth information in
> these
> lines.
>
> ( See the strong program Hiarcs where this feature is not solved well.
> In
> Hiarcs the format is
>
> depth time principal line ( with the evals at the ends )
> 8 0:57 Nf3, Nc6; Bb5, a6; Bxc6, dxc6; Nxe5, +36
> 8 1:13 Bc4, Nf6; d4, Nxe4; Nf3, Nc6; +21
> 7 0:42 f4, exf4; Bc4, Nf6; e5, Ne4; Nf3, +28
>
> My main problem with this format is that the important eval data
> are at
> the ends of the lines. So you have to switch with your eyes all the
> time;
> from left to right to left to .... .
> And furthermore, these evals are not directly below each other (
> because
> of different variation lengths ). )
>
> Switching between different values of k is currently relatively best
> solved
> in Fritz where you simply use the "+" and "-" keys.
>
> Hoping for k-best in Hydra, Ingo Althoefer.
I see the point.
To tell you the truth, I don't know if such a mode is planned. I will
speak with Chrilly about it when I next time see him.
Best wishes
Andreas
Rotes Sapiens
On 17 Aug 1997 19:46:08 -0700, jbe...@islandnet.com (Jonathan Berry)
wrote:
>In article <5t8a2j$9mf$1...@europa.frii.com>,
>kerr...@deimos.frii.com (Tom Kerrigan) wrote:
>:Every programmer thinks it's a total non-issue, and can come
>:up with dozens of explanations for programs failing the "mirror
>:principle" in his sleep, but you think that some sort of
>:scandal is going on.
>The basis of Science is reproducibility. KK has presented
>instances which are identical except trivially, yet most
>programs treat them differently. I agree with Tom's conclusion
>that it is not a scandal (unless the move recommended is
>different), but the discussion of the subject has been
>thought-provoking.
>: [snip]
>I've noticed that I do better against some programs (e.g.,
>Rebel Decade) when I play Black. OK, William of Occam would
>suggest that that the most likely explanation is that J.Berry
>prefers to play Black, but is it possible that some programs
>are mildly or even unknowingly optimized to play Black, because
>that's the colour the computer gets by default? And might not
>such an observation help a programmer or two?
So what are you going to do, criticise every chess program because
white always has the first move, and the program might somehow be able
to exploit this?
Does anyone know where this !@#$%! six wheeled robot came from?
mclane
brucemo <bru...@seanet.com> wrote:
>A few years ago I was walking down a sidewalk in a not very nice part of
>Seattle, and some guy, who was leaning on a building up ahead of me, began
>staggering in my direction, scowling and muttering.
>As he got closer to me, he suddenly belched out an obscenity, picked up a
>(full) garbage can, and threw it at me, straining with effort, reddened eyes
>bulging, sweat- and grime- encrusted hair sticking out in all directions, a
>string of drool dripping down onto his greasy shirt.
>He missed.
Nice story Bruce !! Very poetic story !!
>I said, "Jesus Christ! Are you nuts?", and walked away before he could find
>another garbage can.
!!! I see you speaking this. Yeah ! I really could imagine you telling
this !!
>Since these two situations share some similarities, I believe that I will walk
>away from this one as well.
>Oh, I forgot.
>Jesus Christ! Are you nuts?
>bruce
Have a nice day anyway !!
Stefan Baeurle
Rolf Tueschen wrote:
>
> Bill Newton <Bi...@notwen.demon.co.uk> wrote:
>
> >I do not condone the comments attributed to Rolf T in his postings
> >to Ed. But then again I do not condone Ed's action in taking matters
> >to court.
>
> >Why?
>
> >Because Ed is one of the 'top guns' in this group. He is highly
> >regarded and has nothing to prove or dis-prove.
>
>
> Second Schroder is sometimes like a little girl just crying.
> (Remember his leaving the group like a sissy.)
>
> Third in case of the famous Czub nazi-like propagations.
> In this case Ed is naive like a little boy. He simply wants to tell the
> world that I had not the right to criticise Czub for this mean posting.
> Furthermore this Ed Schroder insulted me reacting on my critic against
> Czub. Now he wants to construct a case as if I had called him a nazi or
> something equal. Also here Schroder completely lost his head. Because I
> simply didn't do this. I never wrote something like this...
>
> So in my view Schroder should definetely go for a trial. That I could
> prove his many lies in the end. Don't help him out of this, please.
>
> >It maybe significant that a number of other highly regarded
> >contributors to this group have, in the past, suggested that
> >Ed should simply ignore Rolf Ts postings if they upset him so much.
>
> a little bit of logic, Bill.
Interesting point! How is your logic defined?
>
> >IMO that's the only action that has ever been required.
>
> >All this court action will do is put money into Lawyer's pockets.
>
> You confuse official court actions with propaganda shows. Everybody
> could pay a lawyer to write whatever -- if he believes it...?.
>
> I already stated that all what this man wrote on behalf of Schroder is
> completely WRONG. Utter nonsense. He doesn't even know to write normal
> English. ,-)
>
> That's why Ed behaves like a little girl. She runs to mama. He has
> stolen my ball, mom. And her mommy comes down yelling, will you give
> back my Nilly her ball, you naughty boy.
>
Is this still a news group talking about chess & computers? If not, why
is this news group still named rgcc?
> You know the truth? I only kissed her on the mouth and it's true I took
> a deeper look into her pa___. Now, think of this.
>
I thought of this. But I wasn't able to find any correlation to computer
chess.
Frankly spoken, it's really getting annoying, to read all this weird
stuff not related to computer & chess somehow...
Kindly Regards,
Stefan.
mclane
kirk...@ee.utah.edu (Dan Kirkland) wrote:
>PLEASE!
>Get a CLUE!!!
>You do NOT know what you are talking about!!!
>You are HARDHEADED to the MAX!!!
>Every chess programmer in the world could tell you that asymmetry is
>NOT a problem, and you would still disagree! Well, fine!
>Feel free to write your own program...
>Otherwise, as far as this topic is concerned you should just
>SHUT THE HELL UP!!!
Thanks. Thats exactly why I wanted him to DEFINE what HE understands
under asymmetrie. I knew that he would present us a very strange point
of view.
>Asymmetry is the ANSWER, not the problem!
>Get that through your THICK skull!
>And find some other (likely useless) topic...
>dan
mclane
Stefan....@t-online.de (Stefan Baeurle) wrote:
>Frankly spoken, it's really getting annoying, to read all this weird
>stuff not related to computer & chess somehow...
>Kindly Regards,
>Stefan.
Please send this to the posters in email and don't quote their stuff
in complete way Stefan !
Hello !
Where do I get the name Stefan Baeurle from ? ComputerSchach and
Spiele ?
What are you doing in the moment ?