Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Here are the results. Any explanations??
49 views
Skip to first unread message
mu...@compuplus.net
May 24, 2012, 6:55:54 AM5/24/12
to
After I gave up on any hopes in getting any money out of the evasive mother fucking faggots in this group, I decided to just give it away and let them make what they will out of it...
I set out to play 100 matches of 25-points but after 60 matches, my win ratio was so ridiculous and things became so boringly predictable that I decided to experiment with shorter series of shorter matches, to give my inferior opponent a better chance...
So, I played 50 matches each of 15-points and 5-points.
The results are posted at http://www.backgammon.comuv.com
I won 74% of the 25-point matches (and 70% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
I won 62% of the 15-point matches (and 58% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
I won 58% of the 5-point matches (and 54% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
All of the matches are available in 3 zip files, (except 2 or 3 that I forgot to save but logged in the spreadsheets), so that you can download them and analyse manually yourself or using bots.
If you take a little time to actually look at enough of the matches, you will start to see some patterns.
And, then see that some of my strategies, like early doubling after my opponent's rolling an initial 63, don't work as well in short matches as in long matches.
Quite interestingly, I did proportionately worse in shorter matches... God forbid, does this mean that I was the stronger player...? ;) :)
I took my precious time and made the effort. Unless you are a world-class sick gambler anonymous, or a mediocre math phd with his head up his ass, or a bot developer/seller making false claims, you may want to look at those matches and relate to us all what you see...
I'll make more detailed comments later also and please feel free to ask clarifications about my statistics.
MK
I set out to play 100 matches of 25-points but after 60 matches, my win ratio was so ridiculous and things became so boringly predictable that I decided to experiment with shorter series of shorter matches, to give my inferior opponent a better chance...
So, I played 50 matches each of 15-points and 5-points.
The results are posted at http://www.backgammon.comuv.com
I won 74% of the 25-point matches (and 70% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
I won 62% of the 15-point matches (and 58% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
I won 58% of the 5-point matches (and 54% of all games during those matches, so it wasn't just a matter of "jacking up the cube").
All of the matches are available in 3 zip files, (except 2 or 3 that I forgot to save but logged in the spreadsheets), so that you can download them and analyse manually yourself or using bots.
If you take a little time to actually look at enough of the matches, you will start to see some patterns.
And, then see that some of my strategies, like early doubling after my opponent's rolling an initial 63, don't work as well in short matches as in long matches.
Quite interestingly, I did proportionately worse in shorter matches... God forbid, does this mean that I was the stronger player...? ;) :)
I took my precious time and made the effort. Unless you are a world-class sick gambler anonymous, or a mediocre math phd with his head up his ass, or a bot developer/seller making false claims, you may want to look at those matches and relate to us all what you see...
I'll make more detailed comments later also and please feel free to ask clarifications about my statistics.
MK
Tim Chow
May 25, 2012, 8:29:33 PM5/25/12
to
On May 24, 6:55 am, mu...@compuplus.net wrote:
> I'll make more detailed comments later also and please feel free to ask clarifications
> about my statistics.
The normal way to generate statistics of this kind is to decide ahead
> I'll make more detailed comments later also and please feel free to ask clarifications
> about my statistics.
of time, before playing any matches, exactly how many matches will be
played (in your case, 50 each of three different types), then to play
exactly that many matches, and to record *all* the results.
Did you do this? The fact that you report 61 matches of 25 points is
worrisome. Did you decide ahead of time to play exactly 61 matches?
Why did you pick 61? That seems to be a strange number. I'm guessing
that this is a sign that you picked 61 matches out of a larger set to
make your results look good, or at least that you didn't decide ahead
of time exactly how many matches you would play but just kept playing
until you got ahead, and then quit.
Your statistics won't be meaningful if you decide how many matches to
report, or *which* matches to report, *after* playing the matches.
Similarly, the statistics won't be meaningful if you abandon some
matches before they're finished, or fail to record their results for
some reason.
---
Tim Chow
mu...@compuplus.net
May 25, 2012, 10:33:24 PM5/25/12
to
On Friday, May 25, 2012 6:29:33 PM UTC-6, Tim Chow wrote:
> The normal way to generate statistics of this kind is to decide ahead
> of time, before playing any matches, exactly how many matches will be
> played (in your case, 50 each of three different types), then to play
> exactly that many matches, and to record *all* the results.
>
> Did you do this? The fact that you report 61 matches of 25 points is
> worrisome. Did you decide ahead of time to play exactly 61 matches?
> Why did you pick 61? That seems to be a strange number. I'm guessing
> that this is a sign that you picked 61 matches out of a larger set to
> make your results look good, or at least that you didn't decide ahead
> of time exactly how many matches you would play but just kept playing
> until you got ahead, and then quit.
I had decided to play 100 matches but I stopped at 61 because it seemed unreal in light of all the praise for Gnubg here.
> The normal way to generate statistics of this kind is to decide ahead
> of time, before playing any matches, exactly how many matches will be
> played (in your case, 50 each of three different types), then to play
> exactly that many matches, and to record *all* the results.
>
> Did you do this? The fact that you report 61 matches of 25 points is
> worrisome. Did you decide ahead of time to play exactly 61 matches?
> Why did you pick 61? That seems to be a strange number. I'm guessing
> that this is a sign that you picked 61 matches out of a larger set to
> make your results look good, or at least that you didn't decide ahead
> of time exactly how many matches you would play but just kept playing
> until you got ahead, and then quit.
I did not pick 61 out of a larger set. All of the 5, 15, and 25 point matches I posted at my site were played using Mersenne Twister with the value seed set at "match length + match number", like 25001, 25002, ..., 25061 or 5001, 5002, ..., 5050 sequentially without skipping any numbers in between.
So, actually I may decide to continue playing 25 point matches starting at 62 and go up to 100, in the same manner.
Saved matches are named using the seed as the prefix, followed by the score, like 25001-mk27-gb00.sgf, 25061-mk04-gb25.sgf, etc.
Now do you think it was a fair enough test?
Even though 161 mixed length matches may not be decisive for most of you, I find it telling enough, especially in light of literally thousands of matches I played against Gnubg over many years.
On the other hand, and by the same token, if these results were contradicting my previous performance against Gnubg, then I would be the one to say that a larger sample would be needed... :))
MK
mu...@compuplus.net
May 25, 2012, 11:21:27 PM5/25/12
to
Additional comments about my stats.
- Points per game: is simply the total number of points won by both players divided by the total number of games played in a match. I'm not sure if this is any useful info at all but I put it there just in case you may want to compare them what may be typical for you all in general.
- Points per win: is the winner's points divided by the number of games won by the winner in a match. This info can be meaningful to show the magnifying effect of the cube and maybe more importantly whether a player has exploited the "jacking up the cube", i.e. winning more points in fewer games played. You can see that in 61 matches of 25 points for example, there were only 2 ties and in only 3 matches the winner of the match actually won fewer games than the loser, (once MK, twice GB!)
- Points per game at the bottom of the page: is the average for MK (3.7) and for GB (2.6), as compared to the overall average (3.2)
- Points per match at the bottom of the page: shows that while GB won only 26% of the matches, it collected 35% of total points. I guess all it shows is that, unlike in money sessions, in matches a lot of the loser's points are just wasted.
There have bees a few "interesting" matches like GB backgammoning me in a 5 point match with the cube at 2 and winning 6-0 in a single game. Or the 9th and 10th 25 point matches that MK won 50-16 and 64-0. I think they are perfect examples of GB's "cube skill" (NOT!)... :))
I'm curious to hear how others would interpret all this. 161 matches are too many to look at but you may want to just look at a few that may catch your eye by the unlikely score or what not...
MK
- Points per game: is simply the total number of points won by both players divided by the total number of games played in a match. I'm not sure if this is any useful info at all but I put it there just in case you may want to compare them what may be typical for you all in general.
- Points per win: is the winner's points divided by the number of games won by the winner in a match. This info can be meaningful to show the magnifying effect of the cube and maybe more importantly whether a player has exploited the "jacking up the cube", i.e. winning more points in fewer games played. You can see that in 61 matches of 25 points for example, there were only 2 ties and in only 3 matches the winner of the match actually won fewer games than the loser, (once MK, twice GB!)
- Points per game at the bottom of the page: is the average for MK (3.7) and for GB (2.6), as compared to the overall average (3.2)
- Points per match at the bottom of the page: shows that while GB won only 26% of the matches, it collected 35% of total points. I guess all it shows is that, unlike in money sessions, in matches a lot of the loser's points are just wasted.
There have bees a few "interesting" matches like GB backgammoning me in a 5 point match with the cube at 2 and winning 6-0 in a single game. Or the 9th and 10th 25 point matches that MK won 50-16 and 64-0. I think they are perfect examples of GB's "cube skill" (NOT!)... :))
I'm curious to hear how others would interpret all this. 161 matches are too many to look at but you may want to just look at a few that may catch your eye by the unlikely score or what not...
MK
Michael Petch
May 25, 2012, 11:47:52 PM5/25/12
to
You may have legitimate results, or you may not. You have a history, and
a persona that precedes you on this forum, and has been discussed on
another site, the biggest concern is that you may have cherry picked or
even manipulate the result.
The fact you use the same seeds, could also be used to your advantage.
You could possibly have dumped out all the rolls for a seed and then
played accordingly knowing what the rolls would be etc. Or if you lost a
match, you could replay it and make opening moves different so that the
dice would play out differently.
I'm not saying you did any of this, but it is the perception of what
people think you are capable of, and as they say on the other site where
your matches were briefly discussed, they don't know. Your reputation is
not good on R.G.B for what you have said and done for the past 13+
years. I think people have a right to be skeptical.
A better test would be for this to be controlled. In that you play
against GNUBG remotely, and where the seeds are unknown to you. The
remote end could log all your sessions so that one would be able to see
whether things were legitimate or not.
Would you be willing to do your experiment in a more controlled
environment? I could probably rig something up using GNUBG's external
interface. You would play from your home in GNUBG but the opponent
(GNUBG on World Class settings) and the dice would be served from a
remote machine (mine at home) as an example.
If you are willing to consider such an arrangement I would be more than
willing to try and set this up for you.
a persona that precedes you on this forum, and has been discussed on
another site, the biggest concern is that you may have cherry picked or
even manipulate the result.
The fact you use the same seeds, could also be used to your advantage.
You could possibly have dumped out all the rolls for a seed and then
played accordingly knowing what the rolls would be etc. Or if you lost a
match, you could replay it and make opening moves different so that the
dice would play out differently.
I'm not saying you did any of this, but it is the perception of what
people think you are capable of, and as they say on the other site where
your matches were briefly discussed, they don't know. Your reputation is
not good on R.G.B for what you have said and done for the past 13+
years. I think people have a right to be skeptical.
A better test would be for this to be controlled. In that you play
against GNUBG remotely, and where the seeds are unknown to you. The
remote end could log all your sessions so that one would be able to see
whether things were legitimate or not.
Would you be willing to do your experiment in a more controlled
environment? I could probably rig something up using GNUBG's external
interface. You would play from your home in GNUBG but the opponent
(GNUBG on World Class settings) and the dice would be served from a
remote machine (mine at home) as an example.
If you are willing to consider such an arrangement I would be more than
willing to try and set this up for you.
mu...@compuplus.net
May 27, 2012, 4:54:28 AM5/27/12
to
On Friday, May 25, 2012 9:47:52 PM UTC-6, Michael Petch wrote:
> You may have legitimate results, or you may not. You have a
> history, and a persona that precedes you on this forum,
I have quite a likable persona and I am proud of what I have posted in RGB fo the past 16 years..
> You may have legitimate results, or you may not. You have a
> history, and a persona that precedes you on this forum,
> and has been discussed on another site,
> the biggest concern is that you may have cherry picked or
> even manipulate the result.
I am not a mentally ill gambler. I have never used or required my opponent to use a cup to roll precision dice. In fact I believe I still have a pair of handmaid small dice made out of bone, which we rolled by hand, often with pretentious gestures and snap of the wrist, as though being able to roll the dice one needed was part of the skill of playing backgammon...
I am not promoting by book, or my blog, or my bot, etc. here. Do you think I would spend all that time just to get a little attention from a few half-brained cocksucking faggots here...? :)
No, indeed not! My goal is to accomplish much more than that. I want to be the person who has done away with the world-class bullshit about "cube skill" and the uselessness of bot roll-outs because of the inherent bias that you all have already admitted to various degrees...
> The fact you use the same seeds, could also be used to your advantage.
If you don't trust me, there is no need to waste another minute of your time talking to me.
> You could possibly have dumped out all the rolls for a seed and then
> played accordingly knowing what the rolls would be etc. Or if you lost a
> match, you could replay it and make opening moves different so that the
> dice would play out differently.
As of yet, nobody has been able to take the challenge...!!!
> I'm not saying you did any of this, but it is the perception of what
> people think you are capable of,
If that's not what you meant, why don's you concoct 61 matches of 25 points against gnubg using the same sequential seeds that I used, in a way that you will win about 75%? Let's set the clock and see how long it will take you to do that. Start! Tick, tick, tick....
> and as they say on the other site where your matches were briefly
> discussed, they don't know.
So, what is it that makes you all suspicious? That I haven't made the same move in the same positions consistently? What?
Even though I am not a robot and should not be held to the same mechanical consistency, I am sure that you can see enough patterns and consistency in my plays that there is no cheating. In fact, most likely, you will see that when I deviate from my own pattern, I end up losing...
Surely the larger than life, world class, teflon coated assholes can detect what needs to be suspected and underline them for the of the rest of you...??
> Your reputation is not good on R.G.B for what you have said and done for the past 13+ years. I think people have a right to be skeptical.
Aside from that, my reputation in RGB goes further back than 13 years and is actually quite good with a lot of readers.
I have continuously fought and encouraged others to stand up against the patronizing that was going on in this group by the mentally ill gamblers, mediocre math phd's, etc. on the newcomers...
I would like to take at least partial credit in making the kitchen too hot for those scums that they had to leave this forum and start their own private discussion blogs or whatever.
> A better test would be for this to be controlled. In that you play
> against GNUBG remotely, and where the seeds are unknown to you. The
> remote end could log all your sessions so that one would be able to see
> whether things were legitimate or not.
Any such test would be the end of commercial products like ExtremeGammon and Snowie, and also expose what kind of scum are behind even the free gnubg.
> Would you be willing to do your experiment in a more controlled
> environment?
> I could probably rig something up using GNUBG's external interface.
> You would play from your home in GNUBG but the opponent (GNUBG on
> World Class settings) and the dice would be served from a remote
> machine (mine at home) as an example.
> If you are willing to consider such an arrangement I would be more than
> willing to try and set this up for you.
Since nobody here had the confidence to bet money on gnubg against me, could we at least conduct this experiment for non-monetary but otherwise tangible stakes?
What I mean, is that if I am able to demonstrate that I can beat gnubg at its highest strength by defying all of the bullshit teachings by all of the con artists in the "small incestuous bg circles", will you all agree to recall the books that you have published praising bot-like play, declare all current bg bots as useless, worthless and obsolete, close down web sites, blogs, etc. dedicated to bots, cube skill, rollouts and all, and forever cease and desist promoting such bullshit...???
If you can say "yes", I am willing to dedicate all the time and effort necessary on my part, to conduct the experiment you proposed above!
Otherwise, I have no time to waste in waging a battle against teflon coated assholes, who will keep arguing that they are right regardless of the facts...
MK
Tim Chow
May 27, 2012, 1:43:40 PM5/27/12
to
I hope that you and Michael Petch can work something out. The results
should be interesting.
On May 27, 4:54 am, mu...@compuplus.net wrote:
> What site is that? Can I read what has been discussed about myself??
http://www.bgonline.org/forums/webbbs_config.pl?read=122579
Be sure to scroll down and click on the followups to see other
people's followup comments.
> > the biggest concern is that you may have cherry picked or
> > even manipulate the result.
>
> Why would I do such a thing? Any speculations in that direction?
I was told by someone who used to read this newsgroup that you
admitted to discarding matches in which side was "ridiculously" lucky.
Also, you have just now admitted that you quit while you were ahead
(61 matches) instead of playing out the predetermined number of
matches.
These may seem harmless actions to you, and I believe that you did
them without malicious intent, but they nevertheless skew the
statistics and make the results scientifically invalid.
Even professional scientists sometimes unwittingly skew their results
by doing things that they think are harmless, so pointing out that you
might make this kind of mistake is not an accusation that you're
intentionally falsifying data.
---
Tim Chow
should be interesting.
On May 27, 4:54 am, mu...@compuplus.net wrote:
> What site is that? Can I read what has been discussed about myself??
Be sure to scroll down and click on the followups to see other
people's followup comments.
> > the biggest concern is that you may have cherry picked or
> > even manipulate the result.
>
> Why would I do such a thing? Any speculations in that direction?
admitted to discarding matches in which side was "ridiculously" lucky.
Also, you have just now admitted that you quit while you were ahead
(61 matches) instead of playing out the predetermined number of
matches.
These may seem harmless actions to you, and I believe that you did
them without malicious intent, but they nevertheless skew the
statistics and make the results scientifically invalid.
Even professional scientists sometimes unwittingly skew their results
by doing things that they think are harmless, so pointing out that you
might make this kind of mistake is not an accusation that you're
intentionally falsifying data.
---
Tim Chow
mu...@compuplus.net
May 31, 2012, 2:48:40 AM5/31/12
to
On Sunday, May 27, 2012 11:43:40 AM UTC-6, Tim Chow wrote:
> I hope that you and Michael Petch can work something out. The results
> should be interesting.
Yes, indeed. Especially if the experts here are willing the state what is expected to happen and what will the result will mean.
> I hope that you and Michael Petch can work something out. The results
> should be interesting.
> http://www.bgonline.org/forums/webbbs_config.pl?read=122579
Thanks for the link. I will respond to those comments in a different thread.
>> Why would I do such a thing? Any speculations in that direction?
>
> I was told by someone who used to read this newsgroup that you
> admitted to discarding matches in which side was "ridiculously" lucky.
> Also, you have just now admitted that you quit while you were ahead
> (61 matches) instead of playing out the predetermined number of
> matches.
I played 17 more during the past couple of days and won 14 of them. I understand your reactions. If someone else had told me, I would have believed it either.
> These may seem harmless actions to you, and I believe that you did
> them without malicious intent, but they nevertheless skew the
> statistics and make the results scientifically invalid.
"Easy come, easy go" attitude helps. Especially if I win the first few matches. After that I'm not tense about losing and actually play bolder and probably win even more because of that. Under the pressure of having to perform, I may not do as well.
Kind of like enjoying masturbation in the comfort of your own house vs. trying to ejaculate into a tube at the doctor's office... ;)
In order to "make a statement", I don't need to duplicate results winning 75% of the time against gnubg. Which is so ridiculous to me also and that's why I joked about whether I was playing against a cracked version of gnugb in a different article a few weeks ago. Even a lesser success will still be enough to make my point.
Now I'm at a point trying to analyse and define what I may be doing right..? I wonder if anybody has looked at my matches yet and offer any observations?
MK
Tim Chow
May 31, 2012, 10:45:45 AM5/31/12
to
On May 31, 2:48 am, mu...@compuplus.net wrote:
> I explained why I stopped. Since the they were played using 61 consecutive seed values, it wasn't cherry picking...
It's not cherry-picking in the sense of discarding intermediate
> I explained why I stopped. Since the they were played using 61 consecutive seed values, it wasn't cherry picking...
values. However, it is still statistically invalid.
If you still don't see this, try the following experiment. Take a
coin from your pocket. Flip it 50 times and record the results. Now
keep flipping it, and stop as soon as you get, say, at least 55%
heads. Now report your results and tell everyone that the coin is
biased in favor of heads, and ask for explanations.
---
Tim Chow
mu...@compuplus.net
Jun 1, 2012, 6:37:04 AM6/1/12
to
On Thursday, May 31, 2012 8:45:45 AM UTC-6, Tim Chow wrote:
> On May 31, 2:48 am, mu...@compuplus.net wrote:
>> I explained why I stopped. Since the they were played using 61 consecutive seed values, it wasn't cherry picking...
>
> It's not cherry-picking in the sense of discarding intermediate
> values. However, it is still statistically invalid.
No, it's not invalid at all. You may argue whether 50, 61, 100, 1000 or 1000 samples are statistically significant but you can't argue that they are "statistically invalid".
> On May 31, 2:48 am, mu...@compuplus.net wrote:
>> I explained why I stopped. Since the they were played using 61 consecutive seed values, it wasn't cherry picking...
>
> It's not cherry-picking in the sense of discarding intermediate
> values. However, it is still statistically invalid.
In fact, there is no such thing as "statistically invalid". That is some bullshit that you must have just invented...
After 61 matches, I got tired of my own success and lost interest.
Then wanted to see what would happen in 15 and 5 point matches. And with those, I set out to play only 50 matches of each.
So, an other way you can look at 61 matches, is 11 more than 50 instead or 39 less than 100...
But don't worry, I will oblige to play those following 30 matches of 25 points just to shut up you and your ilks, so that you poor souls will not have to resort to such bullshit as "statistically invalid"...
> If you still don't see this, try the following experiment. Take a
> coin from your pocket. Flip it 50 times and record the results.
What if I don't like the results after the originally intended 100 flips and want to go on to flip 200 times...?
Show the small change up your phd ass Chow, and make a real effort to understand the subject and make a meaningful contribution to the discussion...?
MK
0 new messages