Incredible miss and backgames against bots

[Yvon]

Thanks for the laugh, Yvon.

I won't post on MSN, but tell me this:  What are the odds of encountering a 1-in-5000 sequence in 1000 games?

You can't even set your question right and you are telling me you want an answer? Laugh as much as you like my dear, perhaps that would help others start laughing at you.

Btw if the link doesn't work try this one

http://groups.msn.com/IsFIBSinterferingingames

[Patti]

wow this comes to 1:5598.
The question is: have I played so many games to expect that to occur once?The answer is NO I only played a litlle more that 1000

OK, what is the likelihood of a 1:5598 sequence occurring in 1000 games?'

Or do you find that question inadequate as well?  I understand that it's impossible to calculate, because you don't know how many rolls occur in 1000 games, but you can probably come up with a back-of-the-envelope estimate to work with.  My guess would be that an average game goes about 30 rolls.

[Yvon]

OK, what is the likelihood of a 1:5598 sequence occurring in 1000 games?'

Or do you find that question inadequate as well?  I understand that it's impossible to calculate, because you don't know how many rolls occur in 1000 games, but you can probably come up with a back-of-the-envelope estimate to work with.  My guess would be that an average game goes about 30 rolls.

First of all you should have known that a sequence of any 4 non doubles has a chance much less to occur than 1:5000 in fact it is approx 1:100,000. This alone should have raised you the suspicion that your question is fallacious.
Here we are talking for a sequence of 4 worst rolls within 8, whose chance to occur has already been calculated and it was 1:5598. The sequence decided the outcome of 1 game under a specific setup (actually it costed me 4 out of my  1000 games). The calculation showed it should occur once every 5598 times. Times here means GAMES not number of arbitrary appearances.
How many games like that would you like to see to be convinced Patti?


Anyway end of discussion and I apologize to PersianLord for derailing his topic. If you want further discussion you know where you will find me.

I also apologize on your behalf to the owners of this forum for your copying from the other forum and pasting here.

[Patti]

My question was really "Do you have a basic understanding of probability and statistics?"  Thanks for the answer.

[playBunny]

If we only just include my rolling of the worst possible rolls and opponents rolling of best possible rolls in a sequence we get
3/36*5/36*5/36*4/36 wow this comes to 1:5598.

If we do that, then it may well be 1:5598 but it's not that meaningful a number. Here's an example to illustrate.

Let's say you spin a coin 6 times and get a mixture of 4 heads and 2 tails.. If you then ignore the tails and multiply the probabilities of just the heads you get 1/16. But what does it mean? What about the tails in there? 1/16 is the probability of getting 4 heads in 4 spins but it's not the probability of doing it in 6 spins.

This investigation can't work if you do probabilities of sequences which omit items in the sequence that don't fit, they must be included to get a meaningful value. In other words it has to be about collections - the probabilities of combinations and permutations : 4 heads out of 6 spins or, in your case, 4 him-best-me-worst in 6 rolls.

Check that this stuff makes sense and then look at the problem again because these are the formulas that apply here.*   

http://www.mathsisfun.com/combinatorics/combinations-permutations.html

(*) except that you are also using the fuzzy concept of "best" and "worst" which complicates things rather a lot. Ideally your calculations should factor in how much these rolls are the best or worst

[playBunny]

Just out of interest, the average number of moves in a game is 20 per player but 25 per player if there's no cube.

[webrunner]

Just one question: why open up a seperate forum?
This is EXACTLY the place to discuss this, isn't it?
Patti is here, the members are here. I don't understand your hostile approach as your first post.

[PersianLord]

Just one question: why open up a seperate forum?
This is EXACTLY the place to discuss this, isn't it?
Patti is here, the members are here. I don't understand your hostile approach as your first post.

How good of you to say that. I really want to open a 'whiner hall' here. People then can post positions when their opponents, bots or humans, got a very lucky roll and then whine about the damned FIBS dice 

Then each week we will choose the 'whiner of the week' title. I guess me, vic and ZW... have a great chance to be successful here 

Regards

[socksey]

We had a whiner champ once in the body of Anakin!  May he rest in peace.  I'm sure many here miss him a great deal as I do. 

Your guess now is pretty good!  (vic for sure! Love u sweetie!)   

socksey



When two egotists meet, it's an I for an I. -  Ann Gardner

[dorbel]

Copy of post to Yvon's "forum".

Yvon, what you are demonstrating here is that you don't understand the game of backgammon! In your post you have tried to show that you lost the game after a sequence with a possibility of 1 in 5598. Actually, several of the asumptions on which you base this figure are false, but let that pass. What you have ignored are all the other possible courses that this game could have taken, many of which would have also led to your defeat.
In the original position that you doubled for example, you are about 43% to win the game from there and of course the opponent's 57% wins will include some gammons, probably about 14. In other words, your double is awful and in money play should be met with a beaver! On the bar after being hit, your opponent is still about 20% to win before rolling. You then think that his best rolls are 1-1 or 3-1, but in fact 6-1 is clearly superior to either. Need I go on?
Take it on the chin buddy. Your game is still at the novice level and your allegations about fibs manipulation of dice is patently absurd. Take some friendly advice. Close this forum and go off quietly and learn to play. I suggest downloading GnuBg. No, it doesn't cheat. Enjoy the game!

[Tomawaky]

Dorbel Thanks for this final post.
Webby Thanks to host it.
Patti Thank for your patience.
Yvon Thanks for this funny post. It's now time to study the game.

[inim]

Just found this incredible thread, I'm in tears. Would it be possbile to tell JokeBot about it, so we can enjoy it for years to come, quote by quote?

[blitzxz]

This topic is now old topic but propably somebody will bring it up time after time so let's think about why. Actually let's think about following set up. You play backgammon with your good friend. But there's a little twist. Your friend will roll dice for both players and he will do it behind his back! After he has rolled he will tell you what numbers he did roll but not show the dice. If you're not thinking that your friend might be cheating you are absolutely naive. And still this is essentially what bots and backgammon sites are doing. You don't know how the dice are generated inside the program you just see the outcome. How can you trust?   

[Hardy_whv]

How can you trust?   

Very easily: For GNU Backgammon the whole source code is available. You can download the source code, analyse it's dice algorithm (and anything else) and then compile the code. It would be so easy to prove any cheating. However, there is none.

Hardy 

[blitzxz]

Very easily: For GNU Backgammon the whole source code is available. You can download the source code, analyse it's dice algorithm (and anything else) and then compile the code. It would be so easy to prove any cheating. However, there is none.

Hardy 

That's true but only for gnu and only if you can understand programming. I sure can't so I just have to trust them.

[playBunny]

If you're not thinking that your friend might be cheating you are absolutely naive.

I would hope it that depends on the friendship. Rather than naive it could be paranoid to think that they are cheating.

And still this is essentially what bots and backgammon sites are doing. You don't know how the dice are generated inside the program you just see the outcome. How can you trust?   

The issue is very different from the example given anyway. A more realistic analogy would be someone that you both hired through an agency to do the dice rolling for you. This person is in another room and calls out the dice through a loudspeaker. They never see you, they don't know your names, they know nothiing about either of you and have been hired by a third party on both your  behalf.

Now you can ask the question about trust and the answer is the same in both cases. There needs to be an incentive for the dice provider in order for selectively biased dice to be given. The dice provider has to have some means of choosing who is to benefit and there has to be a gain, one that's worth it even if the biased dice system is exposed. The likelihood or rather the unlikelihood of that occuring would mitigate that aspect.

For a games server people suggest the incentive of better dice for the better players in order to keep them playing, other suggest that it would be better dice for the weaker players in order to keep them playing. Hmmm.

There's the possibility of the Special Lucky $50 dice for those who value money less than their integrity. But that method requires advertising in order to attract people willing to pay and sufficient people willing to pay. Needless to say the risk of exposure is immense.

There's the possibility that the dice provider does actually know one of the participants and wants to play favourites. If that's done sporadically then nobody else would know, not even the beneciary, but it's hard to imagine much satisfaction for the dice provider. If it's done with greater regularity such that it materially affects the beneficiary's rating and winning percentages then they are likely to start wondering how they can be so lucky at this site but so unlucky elsewhere. Eventually that wondering could lead to discovery of the subterfuge.

It could be done in collusion with the dice provider, of course. In that case let's hope that the dice provider remains friends with the beneficiary lest they get grassed up (http://www.urbandictionary.com/define.php?term=Grassed+Up)

It's difficult to come up with a suitably worthwhile incentive for a dice provider. more ideas are required......

[blitzxz]

I would hope it that depends on the friendship. Rather than naive it could be paranoid to think that they are cheating.

Yes, this would be very interesting phychological test. How many people would cheat if they are rolling behind back? No money just for fun. Maybe I'm paranoid but I'm sure that almost every one would cheat time to time in long session even against a friend. And you could do it with the server analogy as well. Third person calling rolls. I'm sure that many people would cheat there too even if they don't get anything from it. But luckily computers are much more trustworthy then humans.

[playBunny]

Maybe I'm paranoid but I'm sure that almost every one would cheat  time to time in long session even against a friend. And you could do it with the server analogy as well. Third person calling rolls. I'm sure that many people would cheat there too even if they don't get anything from it. But luckily computers are much more trustworthy then humans.

[Bunny's coloration]


Oh my, I see it totally the other way. For me it's most people would not cheat, indeed would be proud to not cheat. And as the anonymous third party dice rollers i think that very, very few would cheat.

It's one of those things that says more about the person than about other people. Not that there's an implcation of it being self-referring, ie. to think that most others would cheat doesn't mean that the person is a cheat themselves, but there's a difference between people who see human nature as basically self-serving, and cheating is one such action, and human nature as basically good. The majority of people actually are basically good but not everyone sees that. The combination of someone thinking that most people are cheats and also that most people deserve to be cheated against .. that's definitely not a good one!!

[lewscannon]

Not for nothing, but I would love to be able to roll manual dice and then tell the bot what it had rolled. Would I cheat, lie, do whatever I could to humiliate the bot? I'd like to think that I'm above that kind of thing.

[Mookie]

 

Not for nothing, but I would love to be able to roll manual dice and then tell the bot what it had rolled. Would I cheat, lie, do whatever I could to humiliate the bot? I'd like to think that I'm above that kind of thing.

I don't know, lews.  If you'll remember, when I met you in that bar in New York, we did a similar experience.  You mixed drinks behind your back and then told me what they were, but I had no way of knowing what you had mixed.  The funny thing was, regardless of what you made me ("Mookie, here's a screwdriver," "Mookie, here's a Maker's and water,"), it all tasted like squid oil.  Now, granted, I'm not a bot, but I also feel quite confident that you weren't "above that kind of thing."

MOOKIE