Playing to an opponents weaknesses
Michael Petch
I've decided to put this question in a new thread, but its related to the
concept of playing BG taking into account an opponents weaknesses. This
concept is a sound one if you really know your opponent.
In the Monty vs Mpetch series Monty has suggested he is the superior
opponent. He states his errors were to trap me into difficult situations
that he had hoped I would play wrong. Lets exclude the cubes in this
discussion. I think the statistics are clear that Monty likely didn't
succeed in making more/larger cube errors compared to himself.
So lets assume that Monty is a 1950 rated player (this would likely be in
Advanced/Expert territory) and seems to be in line with what Monty thinks he
plays at. This discussion would still apply if he was 1925. Monty has stated
he made specific errors to trap me into playing my known weaknesses.
Over the series of 7 matches I ended up with a lower checker play error rate
than he did. Assume I had average ~1900 based on the matches (The average
was slightly higher)
With Monty at 1950 and myself at 1900 he would be the better player. So lets
assume this for the discussion.
My view is this. If you make a checker play error (doubtful/bad/verybad -
doesn't matter) in the hopes that your opponent gets trapped into a hard
situation (Thinking they will make the wrong play), one would expect that
the opponents errors in such difficult situations would be larger than the
initial error. Simply put, if you are going to make an error and play to an
opponents weakness, the opponent on average is going to have to make errors
that are worse - So the NET equity loss remains in the stronger players
favor.
So lets see what happened. Monty apparently played to my weaknesses, and
believed he succeeded but ended up with an overall checker error rate worse
than the weaker player (me). In essence it appears that based on the final
skill level that he gave up more in checker play errors anticipating what I
would do, compared to the errors I actually made.
One might say, if Monty was favored on checker play error rate (because of
his superior abilities) at the outset - than my checker play error rate
should have dropped by at least a proportional amount (and then some!).
So is it possible that Monty went into this thinking he knew how I played,
but in the end his "tricks" to put me in difficult situations actually
backfired to the point that I now became the better player? This is
possible, so lets assume it is true for the sake of argument!
If a near world class player goes into a series of matches, and
underestimates his opponents ability to play tricky situations and ends up
with a higher error rate - wouldn't one conclude that the near world class
player may not really know his opponents strengths and weaknesses to begin
with? If you can't estimate what your opponent will do, you will be throwing
away valuable equity inherent in being the better opponent (to begin with).
Maybe a naïve intermediate player such as myself can't understand how a net
deficit on error rates at the end of the match (compared to the perceived
strengths of each player at the start) can benefit the stronger opponent.
If a near world class player can not estimate an opponents weaknesses before
playing into them, then I do not think they are near world class at all.
Playing to weaknesses that either are not very weak, and giving up equity in
the process makes no logical sense.
So maybe the question to Monty is - if you're near world class in actual
skill, how do you rate yourself in knowing the opponent?
Someone asked this in another thread, but can you point out the weaknesses
in my playing ability, and the places where you took advantage of it? To
know there is a weakness is to know the opponent has made the errors before.
So you'll need multiple examples of similar situations that lead you to
conclude I would make the error you expect (I have never played Monty until
this series so there is no past history but these matches to go by). Next
you'll have to show that in these given difficult spots that I actually gave
up more equity than the initial wrong play.
My conclusion is this. Either a) Monty is near world class, but doesn't have
a clue what his opponents weaknesses are but uses it as an excuse for his
errors, or b) Monty is an Intermediate Advanced player like myself who play
at nearly the same level of play.
If I were to guess. It isn't A. So I will go with B. But I am open to Monty
showing all these weaknesses of mine over time that were a NET benefit to
him because they were exploitable.
For the record, I played into a weakness of Monty. On cube plays I often
cubed when it was too early, knowing that there was a good chance he would
drop. I can especially show this pattern in his first few matches, BUT I am
also aware that if he perceived me as the weaker player that he would drop
them (Basically, I was willing to give up equity on cubes hoping Monty
practiced what he preached on this forum). I also believed he felt I was a
WEAK opponent before the first roll.
I believe that since my error rates were within my expected norms, that I
likely was a NET beneficiary of cube errors compared with a supposed
"stronger" opponent. In the overall series stats its interesting to note
that Monty had 16 bad passes (Compared with 8 of mine), and 2 bad takes (0)
for me. ON Takes/Passes alone Monty gave up a huge amount of equity
(Comapred to my errors). Unfortunately it wasn't enough to overcome the luck
- and that's the nature of backgammon.
Mike
Michael Petch
On 5/11/08 2:31 PM, in article C44CB3CB.58D83%mpe...@capp-sysware.com,
"Michael Petch" <mpe...@capp-sysware.com> wrote:
> Monty likely didn't
> succeed in making more/larger cube errors compared to himself.
Should have read:
"Monty likely didn't
succeed in TRICKING me into making more/larger cube errors compared to
himself"
mont...@lycos.com
creating complicated positions, or making the position simpler for
myself. I'd like to continue to play (9 pt. match at TMG/XG for a
Euro is the only format that will hold my attention, unless you want
to offer me the same deal as before), and things will clarify. I
won't offer an opinion on what your weaknesses are, because that would
obviously defeat the purpose of the whole thing.
Rich
You have put considerable time and energy in this response. I truly
hope you are not expecting a coherent logical response from Monty's
keyboard.
Firstly, it is very unlikely that Monty will answer your question of
what about your style of play, that he can exploit. He will weave and
dodge to avoid answering, most likely explaining that he will give you
an advantage by telling you, if you play a series of matches again.
Secondly, Monty's response to your post is likely going to boil down
to the fact that he won the match 5-2 and results are what count. He
will insist that he will consistently be ahead after a similar series
of matches in the future although in one of his responses, he suggests
that he does not want to do it again. He only wants to play nine point
matches for one Euro on TMG/XG. Monty will not be able to provide any
objective evidence to support his assertion that he can consistently
come out ahead in future matches with you. The two of you are so
closely matched that luck is going to be the deciding factor in any
matches you play IMO. But the last thing that Monty will do is admit
that he won the match because he got lucky. He is far too narcissistic
to do that.
Rich
>
>Mike
>
>
>
>
>
>
>
>
>
Michael Petch
On 5/11/08 2:50 PM, in article
7586ed9d-7d5e-4d6e...@l64g2000hse.googlegroups.com,
"mont...@lycos.com" <mont...@lycos.com> wrote:
> It's not just tricking someone into making inferior moves, but simply
> creating complicated positions, or making the position simpler for
> myself.
Are you saying against a weaker opponent you would make a less complicated
move for yourself if it means the weaker player gains in the process?
> I'd like to continue to play (9 pt. match at TMG/XG for a
> Euro is the only format that will hold my attention, unless you want
> to offer me the same deal as before), and things will clarify. I
> won't offer an opinion on what your weaknesses are, because that would
> obviously defeat the purpose of the whole thing.
I am asking you to show me the weaknesses you already said you took
advantage of in the first 7 matches. You have already aid your error rate
was higher to get me into harder positions. I am asking for you to point in
the 7 matches we played examples where you knew from experience that I would
play a certain way, and that you were able to take advantage of it.
I've already described how it is I played into your major weakness. Since
you told me you did the same, I'd be more than happy to see all of these
weaknesses that demonstrate your superiority in knowing what I'll do.
Since you said you HAD already taken advantage of my weaknesses, I am asking
you to point them out. No need to play more matches to answer the question.
Remember, you brought it up, not I.
mont...@lycos.com
blunders, which I do when I play online, because I just play too
quickly. However, I do think that if we played a large number of 9
pt. matches, assuming there was no huge difference in luck (and also
if you continued to play on the same level, obviously, rather than
improving significantly), I would be at least a bit ahead. If we
played best of 101 nine pt. matches, for example, I think I would have
at least 52 wins. Only one way to find out, and that is to play.
Talk about "objective evidence" makes no sense in this context,
because this is an opinion based upon human intuition. There is no
software presently available that can assess how one style of human
play will fare against another human's style of play, so how could
there be "objective evidence." If you want to continue to play,
that's fine, but at present, my opinion is what it is, based upon my
BG experiences. Luck is clearly a huge factor, so the minimum would
be something like best of 101 nine point matches.
Michael Petch
On 5/11/08 3:11 PM, in article C44CBD34.58D92%mpe...@capp-sysware.com,
"Michael Petch" <mpe...@capp-sysware.com> wrote:
> and that you were able to take advantage of it.
Should have read:
"and that you were able to take consistently advantage of it."
Basically to know I will play a certain way or am weak a certain way means I
had to have demonstrated it already. So first show me where you learned of
the deficiency, then specific examples of how you took advantage of it later
on, and how it substantially improved your overall ability to win.
The thing here is that I have weaknesses. You need to prove you triked me
into them, or made things complex for me, and I did the wrong thing
repeatedly in similar circumstances.
If you wish I can start by showing you specifically where consistently
cubing you early worked to my advantage to make you drop clear takes. I
mean, When I cube you and you have race chances and potential to maybe hit,
why drop because its easier to play from scratch? Why give me free equity
for not having to continue the current game ;-). I can also demonstrate how
I knew you would make such bad drops. The key to knowing you would drop
clear takes in race positions is directly related to your belief that I am
clearly the weaker opponent (in your eyes).
Mike
Rich
<mpe...@capp-sysware.com> wrote:
>Since you said you HAD already taken advantage of my weaknesses, I am asking
>you to point them out. No need to play more matches to answer the question.
>Remember, you brought it up, not I.
Is it not obvious that Monty said this, only to give a plausible
explanation for making inferior moves? He won't be able to point them
out because they don't exist. Not only would he have to point out the
moves that were errors but also that you misplayed the subsequent
moves to a greater degree than him. Since he knows he cannot do it, he
will completely avoid it.
Monty may be a pathologic narcissist but he aint stupid. He got you to
give him free money. And all you got for the trouble is more of
Monty's tortured logic. Now I see you want to give him more money. You
sure are a glutton for punishment. Monty will likely take your money
and give you more of the same.
And if you happen to beat him in the next round, he will whine about
his bad luck.
If your purpose is to prove that Monty is not as good as he claims
that he is, I would suggest you save your money since all you would be
doing is belaboring the obvious.
Rich
mont...@lycos.com
it worked, as it usually does against the players I've encountered up
to now. Your claim that I dropped clear takes is only relevant, IMO,
against those playing at a very high level (such as GNUBG). Thus,
there is an impasse, as I what I think were appropriate plays (against
you) you think are "mistakes." Again, only one way to clear this up.
Rich
>I think my error rate was higher mostly because I made some "goofball"
>blunders,
First Monty says his error rate was higher because he deliberately
made plays to exploit weaknesses in his opponent. Then when challenged
about this, he changes it to being because he plays too quickly. But
Monty is not making excuses:-) Excuses are like assholes; everyone has
one and they all stink.
> which I do when I play online, because I just play too
>quickly.
No one forces you to play quickly. You choose to play quickly. And now
you are using it as an excuse for your higher error rate.
> However, I do think that if we played a large number of 9
>pt. matches, assuming there was no huge difference in luck (and also
>if you continued to play on the same level, obviously, rather than
>improving significantly), I would be at least a bit ahead.
That would only be the case if you were luckier than Michael. You
would not have to be that much luckier than him to achieve this but
based upon your error rates, you clearly would need to have some luck
on your side. If there was zero luck, then Michael would be slight
favorite to be ahead after a large series of matches.
> If we
>played best of 101 nine pt. matches, for example, I think I would have
>at least 52 wins.
At least 52 wins based upon what objective evidence?
> Only one way to find out, and that is to play.
How about doing that and if Michael comes out ahead, you pay him back
the money that he gave you. He gave you free money. If you are so sure
that you would come out ahead, you have nothing to lose.
>Talk about "objective evidence" makes no sense in this context,
>because this is an opinion based upon human intuition.
In other words it is just you blowing hot air out of your ass.
> There is no
>software presently available that can assess how one style of human
>play will fare against another human's style of play, so how could
>there be "objective evidence."
So what makes you so sure of your opinion?
> If you want to continue to play,
>that's fine, but at present, my opinion is what it is, based upon my
>BG experiences. Luck is clearly a huge factor, so the minimum would
>be something like best of 101 nine point matches.
And if one person wins 52 and the other 49 that would prove nothing
even if it was you who was ahead.
Rich
Rich
>In the last match, I decided to use all my techniques against you, and
>it worked, as it usually does against the players I've encountered up
>to now.
Until now you said that your opponents were clearly inferior to you
and that your techniques would not work against an equally matched
opponent. Are you suggesting that Michael is a very weak player
compared to you?
> Your claim that I dropped clear takes is only relevant, IMO,
>against those playing at a very high level (such as GNUBG).
Not true. It is also relevant when you play equally ranked opponents
or opponents who are closely ranked to you. Why give up equity to an
opponents who plays at about your same level?
> Thus,
>there is an impasse, as I what I think were appropriate plays (against
>you) you think are "mistakes." Again, only one way to clear this up.
I think it has already been cleared up. No need to belabor the
obvious.
Rich
mont...@lycos.com
even if it was you who was ahead. "
Then what will, Rich? You don't seem to understand that results are
the only measure that is relevant in this context, due to the nature
of the claim ("opinion"). And you never state exactly what kind of
real results (which to me are "objective") would compel you to
acknowledge that I'm correct. If you don't make such a clear
statement, then all you are really doing is blowing around that "hot
air" that you despise so much.
gammon_...@comcast.net
This is laughable. After posturing for several months, Monty swindles
one poor buffoon into paying him to show up and play so that both can
demonstrate how truly inept they are. At least Petch made no claims to
near-world class status, but Monty demonstrates what an incredibly
poor winner he is.
The one salient fact is that Monty won five of seven long matches> I
think this is pretty significant and I'd guess if they played again,
Monty would prevail. At the end of the day, however, all this really
proved is that one no-talent hack can beat another. The fact that
Monty got paid to demonstrate that he's basically an intermediate must
have him laughing all the way to the bank.
Bob
Michael Petch
On 5/11/08 3:21 PM, in article
255741dd-6dc6-413a...@26g2000hsk.googlegroups.com,
"mont...@lycos.com" <mont...@lycos.com> wrote:
> I think my error rate was higher mostly because I made some "goofball"
> blunders, which I do when I play online, because I just play too
> quickly.
You made goofball blunders. Finally an admission that makes sense. I did too
(And I gave that reason as well), and I have already discussed many of them
here when asked why I played certain moves a particular way (See the posts
between Bob and I regarding errors I made).
Since I play at your level (Or *slightly*) better can I declare myself "Near
world class" because I too am an intermediate/advanced player who made
goofball blunders?
> However, I do think that if we played a large number of 9
> pt. matches, assuming there was no huge difference in luck (and also
> if you continued to play on the same level, obviously, rather than
> improving significantly).
Before all this started I basically said I play in a range between 1875 and
1925. If you look at Kees post (And add 50 to each of our Fibs based Ratings
because he based perfect play at a 2000 level instead of 2050, you will see
that you played at 1875 level, and I played at 1900 (Right smack in the
middle of my average range for long matches).
I didn't improve my game, in fact it is right where I told you it was. And I
have played at this level for some time now (You wouldn't know because its
likely you had never seen me play).
> I would be at least a bit ahead
I assume this is gut feeling? (Do you have any evidence that your feeling
is correct - if so please indulge me). Did you happen to look at the Luck
adjusted results of our 7 matches. Gnubg will try to determine / estimate a
players strength when the luck is factored in. Would it surprise you to know
that once luck is factored out (as well as my skill level compared to yours)
that I was the favorite in every match - even the ones you won.
1 +55.23% Favorite
2 +54.23% Favorite
3 +51.73% Favorite
4 +54.72% Favorite
5 +58.92% Favorite
6 +60.62% Favorite
7 +52.13% Favorite
The best articles about "Luck Adjusted Winning chances" can be found in may
articles by Douglas Zare (If you don't know much about it):
http://www.bkgm.com/articles/Zare/HedgingTowardSkill.html
http://www.bkgm.com/articles/Zare/AMeasureOfLuck.html
This link discusses it withing Gnubg:
http://www.gnu.org.ua/software/gnubg/manual/html_node/Output-of-an-analysis.
html
Based on an independent Bot, your assertion that you would be ahead may be
premature. There is some evidence that statement may not be entirely
accurate.
> Luck is clearly a huge factor, so the minimum would
> be something like best of 101 nine point matches.
It took you 7 matches to determine you were consistently "better" than I
was. Why would it take 101 - 9pt matches to determine how things would look
with more even luck? Most evidence right based on the 7 matches (Which you
have drawn a conclusion about regarding me), run contrary to what the bot
says about luck in those same seven matches.
In the series you have declared yourself the outright superior players based
on a 5-2 record and you can consistently beat me in that format. I'll use
the stats that suit me now since 7 matches WAS a good sample set for when it
worked in your favor. And I think the telling stats are the "estimated" Luck
adjusted Values that made me match favorite in all seven matches, the fact
that I had a lower total checker play error rate, and a lower total cube
error rate across all the matches. Basically, I outplayed you, and when luck
is factored out the bot thinks I should have been favored in every match.
Mike
Michael Petch
On 5/11/08 3:52 PM, in article C44CC6AE.58D9B%mpe...@capp-sysware.com,
"Michael Petch" <mpe...@capp-sysware.com> wrote:
> Gnubg will try to determine / estimate a
> players strength when the luck is factored in
Should have been:
"Gnubg will try to determine / estimate a
players strength when the luck is factored OUT"
My fingers are going numb lol
Rich
wrote:
But it is worse than that. Petch is offering to pay him even more. I
realize that Usenet attracts very dysfunctional individuals but this
takes the cake. Petch even sent money to someone other than Monty
because Monty did not want to reveal his true identity. I think Monty
and Petch deserve each other. They sure make for great entertainment.
I will be sure to watch their next match. It will be a grudge match
for sure given all the smack talk.
Petch essentially gave Monty, for free, more money than he has likely
made in a year playing on line.
Rich
Michael Petch
On 5/11/08 3:49 PM, in article
9be286a1-69de-415b...@y38g2000hsy.googlegroups.com,
"gammon_...@comcast.net" <gammon_...@comcast.net> wrote:
> The one salient fact is that Monty won five of seven long matches> I
> think this is pretty significant and I'd guess if they played again,
> Monty would prevail.
An acceptable opinion. There will be those who agree and disagree.
> At the end of the day, however, all this really
> proved is that one no-talent hack can beat another. The fact that
I can't argue with that. And in fact its bang on!
> Monty got paid to demonstrate that he's basically an intermediate must
> have him laughing all the way to the bank.
Oh come on, more to life than money. There is the entertainment value you
didn't account for :)
Mike
Michael Petch
On 5/11/08 3:56 PM, in article 6lqe24dfqtd2sh685...@4ax.com,
"Rich" <ri...@NOTyahoo.com> wrote:
> They sure make for great entertainment.
> I will be sure to watch their next match. It will be a grudge match
> for sure given all the smack talk.
>
See my other post about entertainment value. No argument there ;-)
> Petch essentially gave Monty, for free, more money than he has likely
> made in a year playing on line.
Chuckle, you are probably right. I was hoping he might be able to use the
money to fund the purchase of a better computer (seriously).
Mike
mont...@lycos.com
basis, though as I said, we'd need something like 101 nine point
matches for it to show for sure (since luck is obviously a huge
factor). You could get lucky against my techniques, of course. I
guess you just don't understand my position, which is that GNUBG
assessments are not all that relevant in human vs. human play, unless
those humans are playing very similarly to GNUBG's style. As to Bob's
claim of a "swindle," I never came up with this idea. I'm willing to
play anyone in at least one 9 pt. match for a Euro (if I think you are
cheating in that first match, though, I might not play you ever
again).
Rich
<mpe...@capp-sysware.com> wrote:
> Basically, I outplayed you, and when luck
>is factored out the bot thinks I should have been favored in every match.
Cue Monty to say something along the lines of "What do the bots know?"
He will ask for evidence of the validity of the error rate and luck
scale of gnu. He will question whether we should use gnu as the
arbitrater of who is the better player.
There is only one person who can tell you who the better player is,
and that is almost World Class "Monty". He is almost World class, even
though by his own admission, he cannot maintain concentration longer
than an hour. Imagine how difficult it would be for Monty to play in
real life where he had to count pips. Cue Monty to say he is a world
class pip counter.
And he is almost world class even though he plays too fast and makes
blunders, as a result.
Rich
mont...@lycos.com
and instead have spent a huge amount of time on your long posts, which
often contain all kinds of irrelevant claims and notions. You strike
me as someone who is not afraid of losing a few Euros, but as someone
who has some sort of severe inferiority complex - your grandiose and
obnoxious posts are a very good indicator of this kind of problem. By
contrast, I'm just putting forth my opinion, and I'm willing to
actually play others to demonstrate my claims or hunches (since there
is no other way of demonstrating that human players can outplay other
human players by playing in a manner that is "wrong" according to
software like GNUBG, unless the person is just about as "good" as
GNUBG).
gammon_...@comcast.net
Let me try to clarify my SWINDLE comment so that Monty can understand
it. Over the past several months Monty has made such an insufferable
flaming, gaping, red asshole of himself that sooner or later someone
was going to bite, even if he had to pay Monty an appearance fee.
It's true that Monty didn't originate the proposition, but with all
the childish egos at play here it was just a matter of time before
some brainless jagoff jumped at the opportunity to humiliate both of
them in front of world-wide audiences.
Let's make the rematch a pay-per-view on HBO. Does anyone have Don
King's phone number ?
Bob
PersianLord
The only way one can support these "tricks" is to disqualify the
software's capabilties to be the judge. In other words, one may argue
that he doesn't accept that a certain move is wrong, though it has
been flagged as blunder by bots, cuz the bots are weak in certain
areas.
Up until recently, the bots were REALLY weak in some situations,
especially defending against backgames, doing recirculation and also
evaluating the rolling outfield primes.(There are some useful articles
on these issues on snowie page) But there has been much improvement in
their understanding and now few may claim that there is still room for
the claim that bots are not capable judges to spot the mistakes and
analysing the overal performance of players. Thus, IMHO, we should
trust the bots now, at least most of the times. So when one's err rate
is lower according to gnubg's analysis, then we may conclude that he
actually played BETTER.
Just my 2 cents
Regards
Rich
> I
>guess you just don't understand my position, which is that GNUBG
>assessments are not all that relevant in human vs. human play, unless
>those humans are playing very similarly to GNUBG's style.
If gnubg assessments are not all that relevant in human vs human play
then why did you post, after playing a match with Progressive, that
your error rate was 6, and Progressive's was 7 to justify your
frustration that you lost the match due to bad luck. I watched that
match and thought that Progressive exploited your weaknesses to
explain his higher than normal error rate. And it worked. He kicked
your ass.
Funny how you think highly of gnu's assessment when it benefits your
position but dismiss it summarily when it does not.
You also said that gnubg was a great learning tool. Why would it be a
great learning tool if you never play anyone who plays very similarly
to gnu's style?
Monty, it is my considered opinion that you are a pathologic
narcissist who cannot admit that he is not as good as he thinks, and
Petch is a fool to pay you to belabor that obvious reality.
Rich
mont...@lycos.com
So how many matches do I have to win before this "trust" is broken?
That is the key question that nobody will answer. If it was answered
directly, it could then be refuted by actual play, and nobody seems to
want to allow that to happen. Some of you seem to be worshipers of
GNUBG (and similar software), at least on a subconscious level.
Rich
>Rich, you've consistently refused to play against me over the years,
You bet your ass I have. And with good reason. Why should I play a
narcissistic blowhard for pennies when I can make much more money
playing for bigger stakes?
>and instead have spent a huge amount of time on your long posts, which
>often contain all kinds of irrelevant claims and notions.
Name a single irrelevant claim/notion I have made.
>You strike
>me as someone who is not afraid of losing a few Euros, but as someone
>who has some sort of severe inferiority complex - your grandiose and
>obnoxious posts
"Grandiose and obnoxious posts"? Project much? I am not the guy
claiming to be almost world class backgammon player. I am not the guy
who claims he could consistently beat someone when analysis by gnu
indicates that at best you are equal to Petch in playing ability.
> are a very good indicator of this kind of problem. By
>contrast, I'm just putting forth my opinion, and I'm willing to
>actually play others to demonstrate my claims or hunches
Yeah, you got some fool to pay you to play him. Kudos to you for
that.
> (since there
>is no other way of demonstrating that human players can outplay other
>human players by playing in a manner that is "wrong" according to
>software like GNUBG, unless the person is just about as "good" as
>GNUBG).
Your statement above makes no sense. I think that gnu is a perfectly
good way to evaluate luck adjusted skill levels. And apparently so do
you when it works in your favor (eg. match with Progressive where your
error level was a point below his and you lost).
Rich
PersianLord
lol, there is a way for that. Go play 50 9-pt matches against GNUBG at
home, using your ultimate tricks, and then come here report your
overal results and match logs. Then we will see whether human players
still can outplay bots or not.
And yes, at least I'm a GNUBG worshiper.
Regards
Michael Petch
On 5/11/08 4:56 PM, in article
bfdf5446-feb9-4e7e...@a23g2000hsc.googlegroups.com,
"PersianLord" <sharif...@gmail.com> wrote:
> lol, there is a way for that. Go play 50 9-pt matches against GNUBG at
> home, using your ultimate tricks, and then come here report your
> overal results and match logs. Then we will see whether human players
> still can outplay bots or not.
I believe I hear footfalls approaching... I think its Murat approaching ;-)
Stick
I worship GNU - You beat me over 50% of the time in the best of 100 ++
+ number of matches you'll break that trust. I won't play you 1 Euro
matches, complete waste of time. Why don't you put up the money you
just got paid for doing basically nothing by Michael and I'll put up
twice as much, the winner at the end of however many matches you state
initially gets the goods. I will play you on TMG and those matches
can also be for any amount of money you like. Assuming 9pt match as
you've said, and we'd put the side bet money in someone else's hands
we both trusted. (some sort of escrow) Someone like Michael, who I
don't know but has earned this sort of trust, would be fine with me.
Others on this board who occassionally post are trustworthy also, but
I don't know if you'd trust them. [I've met them]
Stick
http://www.bgonline.org/forums/
paulde...@att.net
>
> - Show quoted text -
Hi GammonGang,
Just wondering what the reason is for your anonymity?
Are you really a world-class player? Adam Tansley seemed to imply
that you are an extremely strong player (perhaps world-class) and I
certainly consider Adam to be a reliable source.
If you are a world-class player, how about offering some world-class
analysis?
I'm not a particularly good player myself. I estimate that I play at
about the same standard as Michael Petch. However, if I beat Michael
Petch 5 to 2 or even 6-1 or 7-0 and if my error-rate was not better
than his, I would be more than happy to accept that my victory was due
to luck. Many many times, I have beaten players far stronger than me
with better dice.
Your comment about the "salience" of a 5-2 victory seems bizarre.
It's a rather small sample size after all.
Paul Epstein.
Michael Petch
On 5/11/08 8:02 PM, in article
64233c4a-9d53-48a3...@q24g2000prf.googlegroups.com,
"paulde...@att.net" <paulde...@att.net> wrote:
> the "salience" of a 5-2 victory seems bizarre.
> It's a rather small sample size after all.
>
If its the same person they have a rating on Fibs of 2200+
EthicalBackgammon
Interesting perspective. I am pretty much convinced that you can make
less than optimal plays to take advantage of an opponent's weaknesses
ONLY when you are very sure of those weaknesses. If you are not very
sure, it can backfire, as it did in the matches you experienced with
Monty where he ended up a higher error rate.
I use this tactic often, but generally only when I am very sure I know
my opponent's game and weaknesses, and only if I am confident that my
opponent is a much lower level player. Even then, MOST of these
changes in my game revolve around the cube...cubing early if I think
there's a good chance of a drop, or cubing when I'm too good if I feel
there's a good chance of a take.
I have often heard that if you are a much better player than your
opponent, it is better to maneuver into more complicated positions and
stay away from simple games and pure running games, but how to
actually do this over the board and not give your opponent a big
positional advantage can be very tricky. When you say a player makes
checker plays to take advantage of a player's checker playing
weaknesses, I'd love to see some specific examples of that, because I
think that would be extremely difficult to do over the board without
making some intentionally-large errors. I'd love to hear exactly what
kind of specific checker error deficiencies one might think his
opponent might have and how to take advantage of them over the
board.
So yes, I do adjust my game if the player is extremely worse or
better, and I do adjust my cube play if I am confident I know my
opponent's weaknesses there, but as for attacking specific checker
play weaknesses....I would have to see a lot of evidence to be
convinced this is a valid, or even possible and practical tactic.
paulde...@att.net
>
> "pauldepst...@att.net" <pauldepst...@att.net> wrote:
> > the "salience" of a 5-2 victory seems bizarre.
> > It's a rather small sample size after all.
>
> If its the same person they have a rating on Fibs of 2200+
I can't figure out gammongang because there's evidence that he's an
incredibly strong player but his analysis is consistently non-existent
(on this thread for example) or expressed in extremely vague non-
technical language. There's a thread where he says something like
"This is a pass because I wouldn't want to go into a battle with 10
soldiers against 13 soldiers". (Turns out he was wrong and it was a
clear take.) Anyone can be wrong but he seems to have a profoundly
non-technical and simplistic way of looking at the game.
Can't figure him out -- perhaps he's a great intuitive player with
poor analytical skills.
Paul Epstein
EthicalBackgammon
I truly believe that the GNUBG and Snowie assessment are of great
value when evaluating human vs. human, but what needs to be considered
is not error rates, but the comparative error rates. If I double too
early, I lose points, even though I might have gotten a drop from my
opponent. But overall, I gained in the comparison, so that makes my
early double correct when playing this particular human in this
particular match, and for this analysis, the bots are most helpful.
paulde...@att.net
>
Robertie's Advanced Backgammon has an example where he recommends a
suboptimal checker play in a money game. The reason is that, after
any reasonable checker play, the opponent should hold. However,
there's a checker play which is likely to have the effect of luring
your opponent into a poor double. This is the anti-theoretical
checker play that Robertie recommends. (Sorry, I don't have the text
and therefore can't reproduce the example.)
Paul Epstein
Rich
Best to remain anonymous on the internet. Especially true for Bob
Ebeler. Oops. There goes his anonymity. Sorry bout that.
>Are you really a world-class player?
No. But he tries to pretend that he is. His name is Bob Ebeler and his
nym on FIBS is NIHILIST. He gets off on manipulating his rating so
that he is at the top of the leader board. He and BIGGLES are cut from
the same cloth.
Some think of him as a world class asshole though.
> Adam Tansley seemed to imply
>that you are an extremely strong player (perhaps world-class)
He is as world class as Red Top aka Hank Youngerman.
>
>I'm not a particularly good player myself. I estimate that I play at
>about the same standard as Michael Petch. However, if I beat Michael
>Petch 5 to 2 or even 6-1 or 7-0 and if my error-rate was not better
>than his, I would be more than happy to accept that my victory was due
>to luck. Many many times, I have beaten players far stronger than me
>with better dice.
>
>Your comment about the "salience" of a 5-2 victory seems bizarre.
Actually the comment is par for the course for Bob. I think he has an
axe to grind with Michael Petch.
>It's a rather small sample size after all.
Rich
paulde...@att.net
Hi Rich
Thanks for explaining it.
Paul
mont...@lycos.com
home, using your ultimate tricks, and then come here report your
overal results and match logs. Then we will see whether human players
still can outplay bots or not."
I wouldn't use "trick" against GNUBG - that's the point of my argument
- you only do these things against humans who don't play like GNUBG!
Now let's take a look at Stick's statement:
"I worship GNU - You beat me over 50% of the time in the best of 100 +
+
+ number of matches you'll break that trust."
What does this mean? Do you think that you play as "well" as GNUBG?
Or are you going to use GNUBG during the matches? And what is the
point anyway, assuming you play so much better (otherwise, your
statement can't make any sense)? My strategy is for players who are
weaker or about the same in strength.
As to playing for higher stakes, it's not going to happen. For me,
that ruins the purity of the game, and brings in a factor that may
hamper the strategy. That is, if a person doesn't want to lose a
certain amount of money, he/she will likely play more cautiously, and
I can't say that I wouldn't (even if it's not entirely conscious).
For a Euro, I don't mind implementing my strategy and taking risks
(according to GNUBG) with this small sum of money. As I've said,
backgammon needs to be more "professional," with top players being
paid to attend the "big" tournaments. As things stand, OTB BG is
ridiculous - 17 point match elimination format and all those expenses
at the "world championship." What a joke.
Now, what exactly do Rich, Stick, and the rest of the "boys" want to
claim about BG more generally? If they are correct about me, then the
game is not worth playing - too much luck (and as I've said, poker
then becomes a much better "deal" in every way). In the last match
with MP, I doubled early (or perhaps right on time - didn't look at
the analysis), and I refused supposed "takes," and I won with that
strategy (along with doing other things I won't mention). The last
double (that he took and redoubled) was a move too late, meaning that
I tried to double the previous move but somehow accidentally rolled
the dice, as I mentioned in the comment box while I was playing.
Again, I still don't know what was "correct," but I went all out for
the win, because I wanted a 5-2 victory rather than a 4-3 one. I'd
like MP or someone who is actually willing to play state exactly how
much I'd have to win against him/her (9 pt. matches) before he/she
conceded that I am a better player, regardless of the GUNBG ratings/
assessments. Until then, there is only "GNUBG worship" opposed to
actual results, and for now, under the circumstances that the match
was played (in other words, if the same format against MP was played
again), I side with actual results.
Stick
good. lol@using GNU during the matches btw. I'd use Snowie to cheat
you of course, it's quicker to set up positions with duh.
The point of playing you was to shut you up, you constantly make
remarks about the rubber room boys who won't play you for pennies -
but in this thread it actually seems, unless I'm misinterpreting what
you said, that you're suggesting I may indeed be a stronger player
than you .. in which case I'll let everything die.
I'd 86 the poker references and how poker involves more skill than bg
if we were to swap shoes. Poker compared to bg & the luck factor are
close enough to equal as far as I'm concerned.
Stick
http://www.bgonline.org/forums/
Rich
>"Go play 50 9-pt matches against GNUBG at
>home, using your ultimate tricks, and then come here report your
>overal results and match logs. Then we will see whether human players
>still can outplay bots or not."
>
>I wouldn't use "trick" against GNUBG - that's the point of my argument
>- you only do these things against humans who don't play like GNUBG!
>
>Now let's take a look at Stick's statement:
>
>"I worship GNU - You beat me over 50% of the time in the best of 100 +
>+
>+ number of matches you'll break that trust."
>
>What does this mean? Do you think that you play as "well" as GNUBG?
No. But he thinks he plays a lot better than you.
>Or are you going to use GNUBG during the matches?
Monty paranoia.
> And what is the
>point anyway, assuming you play so much better (otherwise, your
>statement can't make any sense)?
The point is for him to win money. He is giving you two to one odds.
If you are as good as you think, you should jump at it. After all you
are not risking any money at all. Michael Petch gave you a gift. And
you could double the gift by taking up Stick's offer.
> My strategy is for players who are
>weaker or about the same in strength.
Your strategy is for players who are the *same* in strength? How does
that work? If you are the same in strength then you will split games
over time.
>
>As to playing for higher stakes, it's not going to happen. For me,
>that ruins the purity of the game, and brings in a factor that may
>hamper the strategy. That is, if a person doesn't want to lose a
>certain amount of money, he/she will likely play more cautiously, and
>I can't say that I wouldn't (even if it's not entirely conscious).
But if you take up Stick on his offer you will play *very* carefully
and slowly because you have incentive to win the money. With Petch's
offer, you had no incentive to play well. You get the money win or
lose. If the only way for you to win money would be to win the match,
you would be more careful with your play.
>Now, what exactly do Rich, Stick, and the rest of the "boys" want to
>claim about BG more generally? If they are correct about me, then the
>game is not worth playing - too much luck (and as I've said, poker
>then becomes a much better "deal" in every way).
If you are playing to maximize profits then you need to be selective
in picking your opponents. If you play opponents who are equal or
better than you, then you will lose money in the long run. Of course
if you get fools to pay you just for playing then you cannot lose.
Sounds like you found a Sugar Daddy in Petch. He wants to pay you even
more.
> In the last match
>with MP, I doubled early (or perhaps right on time - didn't look at
>the analysis), and I refused supposed "takes," and I won with that
>strategy (along with doing other things I won't mention). The last
>double (that he took and redoubled) was a move too late, meaning that
>I tried to double the previous move but somehow accidentally rolled
>the dice, as I mentioned in the comment box while I was playing.
>Again, I still don't know what was "correct," but I went all out for
>the win, because I wanted a 5-2 victory rather than a 4-3 one.
The fact that you won the match does not mean that your strategy was
the reason you won. You could have won in spite of your strategy.
There are beginners who play every move safely. They sometimes win
when they roll a number of doubles. Does this mean that their strategy
was correct?
> I'd
>like MP or someone who is actually willing to play state exactly how
>much I'd have to win against him/her (9 pt. matches) before he/she
>conceded that I am a better player, regardless of the GUNBG ratings/
>assessments. Until then, there is only "GNUBG worship" opposed to
>actual results, and for now, under the circumstances that the match
>was played (in other words, if the same format against MP was played
>again), I side with actual results.
I wonder if Monty would side with the actual results if it was 2-5
against him. Apparently the luck adjusted rolls indicated that Petch
should have won all the matches. But because Monty got lucky, he won 5
of them. That is the nature of backgammon. In the next series it could
go the other way. Then of course Monty's tune will change. It is so
predictable.
Rich
Michael Petch
On 5/11/08 9:40 PM, in article
45462dcf-4e51-44ac...@w7g2000hsa.googlegroups.com,
"mont...@lycos.com" <mont...@lycos.com> wrote:
> The last
> double (that he took and redoubled) was a move too late,
> meaning that
> I tried to double the previous move but somehow accidentally rolled
> the dice, as I mentioned in the comment box while I was playing.
I remember this quite vividly and is exactly as you described.
> Again, I still don't know what was "correct,"
I knew what was correct. In fact in my annotation of the match said this:
"
Here is an interesting situation when you take into account Match score.
Under many circumstances this would be a cube and drop. Unfortunately
dropping would lead to a 6away-1away crawford situation. I didn't think
about this play, it was an automatic take and recube to 4. I just didn't get
the luck I needed to pull it off.
"
The position is here:
http://www.capp-sysware.com/analysis/mpetch_vs_Monty-match7_013.html
Move #24
Monty it was an error (but not a huge blunder) to cube me on Move #24.
On Move #23 when you wanted to cube the erorr would have been a massive
blunder at -.333 (EMG)
3. Double, take +0.353 -0.333
And the roll before that would have been even worse.
This was an easy take. For a number of moves I had hoped you would cube me
giving me equity. I wish you had cubed me a roll earlier!
> but I went all out for
> the win, because I wanted a 5-2 victory rather than a 4-3 one.
This is a very perplexing statement. You made an error on cube to win 5-2.
But I correctly took your cube and increased my own chances of winning (Your
MWC-Not EMG was 88.61% before and 88.15% after. How does that help you -
unless I totally screwed up the final moves and gave away all the equity you
handed to me (Which didn't happen). At one point you dropped to only a 68.3%
MWC favorite (for game and match) but got a lucky double 1 on Move #34:
Alert: lucky roll! ( +0.493)
I can sum up Monty's rationale. If he makes an incorrect play (and a big
whopper of an error had he cubes when he wanted to) and wins - it is skill.
But if I had won the game, it would not have been for being skilled enough
to take and accept the equity he handed me but because I was just lucky.
Sure I needed to be lucky, but after the cube I didn't need to be AS lucky!
> I'd
> like MP or someone who is actually willing to play state exactly how
> much I'd have to win against him/her (9 pt. matches) before he/she
> conceded that I am a better player, regardless of the GUNBG ratings/
> assessments.
I don't know. Since I am the more skilled cube and checker player but less
lucky - how many matches would I have to play and win before you concede I
am the better player? Your ego doesn't seem to comprehend the fact you may
have to concede that I might be the better player ;-).
Since you are the expert statistician, and have a firm grasp upon sample set
selection, and statistical interpretation - I will defer to your expertise.
How many matches do you think you need to play to prove you are the better
player AND How many matches would we have to play (And hw many wins) for you
to concede I am the stronger player?
In reality all of Montys post point to one thing. Even though he was the
luckier player, he won (In his mind) on Skill. Unfortunately he can't accept
that there if luck was factored out that I may have have won the series.
I do know this. Had things been reversed and I won the series 5-2, but I was
outplayed by Monty - he would be quoting how lucky I was, and that the skill
difference proved that over the long term he would be the superior player in
the long run.
It reminds me of the little (and humorous book) "How to lie with statistics"
by Huff.
I have offered to play you for $150/Match (25ptrs, and a limit of 9
matches). If you are the superior player than I have no chance, and it
should be seen as free money to you.
Mike
Michael Petch
On 5/11/08 10:43 PM, in article C44D26F6.58F09%mpe...@capp-sysware.com,
"Michael Petch" <mpe...@capp-sysware.com> wrote:
> Monty it was an error (but not a huge blunder) to cube me on Move #24.
Should have read:
"Monty it was an error (but not a huge ERROR) to cube me on Move #24."
Michael Petch
On 5/11/08 9:04 PM, in article
ddcbf678-6e28-4e7c...@24g2000hsh.googlegroups.com,
"EthicalBackgammon" <psim...@sbcglobal.net> wrote:
>
> So yes, I do adjust my game if the player is extremely worse or
> better, and I do adjust my cube play if I am confident I know my
> opponent's weaknesses there, but as for attacking specific checker
> play weaknesses....I would have to see a lot of evidence to be
> convinced this is a valid, or even possible and practical tactic.
Thanks for the post Phil, nice to get your input. I totally agree it is a
lot easier to take advantage of a weaker opponent on cubes, than it is to
show it on checker plays. Likely 7 matches isn't enough to determine
weaknesses, and show that they weren't one off situations etc with checker
plays. Monty helped me on knowing how he cubes just by his posts, and he was
pretty faithful to what he said (Didn't ake too many games to get him to
drop a clear take in a race situation). In our series I think the cube stats
spoke for themselves.
One of the only 2 books I own (One is more of an article than a book). The
book I have is Jacobs and Trice's "Can a Fish Taste Twice as Good?". I
bought it for the assymetric MET'd because of some experimenting I wanted to
do with GnuBG.
I find the topic of playing based on assymetric skill fascinating.
Michael Petch
On 5/11/08 9:04 PM, in article
71e39c24-5962-4074...@q24g2000prf.googlegroups.com,
"paulde...@att.net" <paulde...@att.net> wrote:
Neither can I. The posts on this forum - I agree his posts here don't seem
to match the playing ability that some people associate with him. That's why
I'm unsure about the poster myself.
Michael Petch
On 5/11/08 9:18 PM, in article fodf24lvhp427j0nc...@4ax.com,
"Rich" <ri...@NOTyahoo.com> wrote:
> Actually the comment is par for the course for Bob. I think he has an
> axe to grind with Michael Petch.
This clearly caught my attention. I know many people who have an axe to
grind with me, and many who have an axe to grind that I don't even know. I
assume I may know Bob under some other nickname? From the zone? His name
isn't familiar to me, but that doesn't mean I don't know him.
Its much easier to be a target when one isn't anonymous ;-)
Philippe Michel
> I use this tactic often, but generally only when I am very sure I know
> my opponent's game and weaknesses, and only if I am confident that my
> opponent is a much lower level player. Even then, MOST of these
> changes in my game revolve around the cube...cubing early if I think
> there's a good chance of a drop, or cubing when I'm too good if I feel
> there's a good chance of a take.
It's simply that these tactics seldom work if the opponent doesn't have to
make his bluder *immediately*, not after a dice throw.
That leaves :
- early double / drop : common, usually huge gain when it works, immediate
- late double / take : less common, huge gain possible, have to complete
the game with the randomness it implies
- checker play to stave off a double : still less common, often doesn't
win anything even if it "worked" : you're doubled similarly in or out one
or a few rolls later
Raccoon
> In the Monty vs Mpetch series Monty has suggested he is the superior
> opponent. He states his errors were to trap me into difficult situations