grenv:

Czuch: Holy crap! You're not kidding. It's why I didn't join it when it was announced.

Carl, AlliumCepa: That's wierd. I copied the links directly from the tournaments page. I've changed them from absolute urls to relative ones. It should be fine now. Thanks, guys. :-)

A monthly series of fast-paced Hypergammon tourneys.


I like it fast, 5-point Hyper, 1st Jan 2010

I like it fast, 5-point Hyper, 1st Feb 2010

Checking another two 21-pointers yielded a more reasonable number of equivalent rolls.

Game 1   35, 44
Game 2   52, 52
Game 3   54, 55
Game 4   53, 55
Game 5   64, 53
Game 6   32, 53
Game 7   51, 54
Game 8   23, 62
Game 9   42, 43
Game 10   41, 43
Game 11   31, 32
Game 12   61, 11
Game 13   46, 52
Game 14   62, 62
Game 15   63, 65
Game 16   13, 31
Game 17   65, 66
Game 18   65, 63
Game 19   35, 63

Game 1   34, 35
Game 2   54, 65
Game 3   45, 64
Game 4   51, 52
Game 5   14, 51
Game 6   14, 42
Game 7   61, 41
Game 8   16, 26
Game 9   23, 23
Game 10   13, 41
Game 11   35, 63
Game 12   52, 54
Game 13   42, 43
Game 14   25, 62
Game 15   65, 66
Game 16   63, 65
Game 17   65, 61
Game 18   32, 43
Game 19   34, 63

So the coincidences can be readily not found too.

pedestrian: I also did a very quick check using alanback's 21-pointer against furbster. It was the first and only match that I looked at.

Of the 12 games in the match, 5 started with the same roll for each player, although the order of the numbers was different in 3.

Game 1   61, 46
Game 2   23, 42
Game 3   12, 31
Game 4   14, 51
Game 5   13, 31
Game 6   45, 54
Game 7   15, 51
Game 8   41, 42
Game 9   41, 41
Game 10   13, 41
Game 11   65, 63
Game 12   31, 31

If it's coincidence then it certainly seems to be one that can be found without much effort.

ps. Abigail's right about the odds. It doesn't matter what the first roll is, the chances of the second matching it are dependant on the 36 possibilities for that second roll.

spirit_66: Ah, so turn-based backgammon just doesn't suit you. That's fair enough.

spirit_66: I'm still thinking more and more if BK is a good place for playing BG

BrainKing is great for a whole range of games but it's not the Home of Backgammon. If you're not already at DailyGammon you ought to check it out. See my profile for more details.

Heh heh, one of the attractions that I list is Pit yourself against a world class robot!

wetware: spirit_66 has said it too: Playing against bots I can do this on my PC with GNU, Jellyfish whatever for free.

The reason I asked you what you meant is that I think that argument rather misses the point. The point is that someone is masquerading bot as a person not that they are providing a bot to play against.

With regard to that latter point, whenever a site provides a bot to play against it is well attended! You may well have GnuBg at home but you cannot play turn-based against it with anything like the ease that you can at a site designed for the purpose. gb001, one of Dailygammon's GnuBg-based bots is very popular and not just with those who don't have GnuBg themselves. ;-)

wetware: If I wanted to play against gnubg, I could do that without coming here to BK.

I've heard a number of people say something like this and, of course, I know what it literally means, but what is your point?

grenv: I'd be more interested in if the player's moves were identical to an easily utilized program...

Erm, more interested than in what? We're already talking about play that matches a bot's.

since even computers make minor mistakes, the odds of the player making exactly the same mistakes over an extended time is... ?

Well the odds of the same position coming up again is pretty low so that's not the best thing to look for and such a search would involve looking at the moves made. However, the odds of playing a zero error rate in match after match is nil and at or below a "World Class" error rate is very low, whereas the ease of detection is high. All you need are the error totals for the matches. When nabla and I investigated we found a zero error rate and no other possible conclusion than that a bot was being used for the moves.

spirit_66: Do you mean game or match?

I play many games flawlessly or with a couple of minor mistakes but very rarely do I play a flawless or near perfect multi-point match. The reason is that some games consist solely of positions and moves that occur so often that they become standard. Such a game will be played perfectly. Most matches take a player into unfamiliar territory and that requires thinking and judgment. Such games generate mistakes.

When checking out someone that you suspect of cheating it's best to look at enough matches that you are feel both convinced and that you could convince others if required to do so.

spirit_66: What would anybody think if BG games are analysed by GNU and one can see that ones opponent didn't make any mistake and what if this happens again and again???

If it happens again and again then the player rises to the top of the ratings, round about where sergey82 is.

Not that I'm saying that sergey82 is a cheat, of course.

rod03801: Lol. It was a good thought. Granted, not a correct one, but good nonetheless.

rod03801: I doubt that it's that kind of isue. The problem with the images occurs after the missing move. Up until then everything's fine. That suggests that something on the server is getting confused. They have tooltips on the points to show how many pieces there are. A point with "b2" would have two black pieces. The tooltips for these of the dodgy images are things like "612", ie. no colour and way too many pieces. Also, the dodgy images are on points which don't have pieces, for instance the bearoff tray when the game is far from the bearoff stage.

Pedro Martínez: That game goes wierd at this point, unless it's just for me.

http://goldtoken.com/games/play?g=306620;v=1762

Clicking on the > arrow gives a page with bar and bearoff images that don't show. This happens just after a missing move in the game record at 882. After that there are more and more missing images. Do you see the game properly?

Pedro Martínez: Now that's the one that made me laugh!

But, in fairness to the others, maybe they paved the way.

gogul: I haven't a clue what you're talking about.

gogul: Why not using online computing power capacity?

What for?

gogul: I'm glad that mine was a hint. Yours arouses curiosity. It's more of a riddle!

gogul: Did you know that the recycling of tetrapacks to roof tiles is a Brazil innovation?

That would be have been good protection against the droppings of ptetrasaurs, which inhabited Brazil up to about 70 million years ago.

AlliumCepa: A millisecond is a long time for a modern computer. My 3.0 GHz AMD can produce over 250,000 dice rolls per second using a very basic formula (although it doesn't do anything further with the dice).

But for online backgammon you need to take into acount the context in which the dice roll occurs. There's an html page to be generated for each dice roll. Fencer fills each game page with loads of information, some of which requires access to the database. I think that 1 page per millisecond would be impressive even for a server written in Java.

CryingLoser: if Fencer implemented his RNG according to the first way, then
sooner or later somebody will count the cycle of the numbers...

Er, how "later" can your scale go?!

Czuch: Whenever someone says "it seems more often than random that ..." about a random process it seems more often than not that it's their brain seeming things and not the random process.

Without knowing how BrainKing's dice are generated I can say nothing with certainty about the rolls, however, as I see absolutely no benefit to Fencer or the players in having every dice generated according to the clock it would surprise me greatly if it were the case.

But it's moot anyway because there's no way for us to determine the timing of dice events.

playBunny: Further to my "not sure there's a hard way", there are several steps.

** First get a lot of matches. That's a fair challenge in itself.

** Unless Fencer has fixed the .mat export bug which makes multi-game match files useless, you'd need to run each match file through a script that would fix the error.

** Run the matches through another program that will extract the dice rolls.

** Do your analyses. Here it's very much a question of how complex you want it - whether you want to look at dice independantly of the games, ie. just streams of rolls, or whether you want to know about dice in relation to the players and positions. The latter would be a considerable adventure in programming.

Thad: Hmmm, you say "easy"... ... I'm not even sure that there's a hard way!

coan.net: Fencer might redo all the ratings with a better rating system and fix everything up to this point

That would be very interesting and I hope he does it that way. It would also be nice to have a preserved copy of the ranking tables just before the conversion. It would be fascinating to study how some players will have gone up in the rankings and others down because of the mix of match lengths and opponents that they play.

alanback: Heh heh, fortunately that's as much a product of attitude as it is of the bodily resources. Keep it up, old timer, enlightenment is yours for the taking.

alanback: Alan, those are fancy words but, if it's to be viewed from that perspective, it is an absolute truth that nothingis important, including, but not limited to, our quality of life and whether we live or die.

You mention ego and sanity. Although clearly insane back in 2006, you were thevertheless being much more human, if somewhat boastful, than when you take the lofty view! Backgammon (alanback, 2006-10-18 20:20:52)

And now you are out of the top 5 as a result of a handful of cheats. It's a good job that you no longer care about your position.

I think that even with a flawed rating system it is worth protecting the validity of the ranking charts. If people are put off playing because of that lack, especially if they're put off paying to play, then it's a loss. Of course changing the BKR to use the standard backgammon formula would help prevent the cheating and fix the rating formula problem in one go.

lukulus: I think problem with rating in BG games is not so big. Most of the cheaters are pawns so they will disappear in one month w/o finished game.

Conversely, as long as they continue to play, their rating will persist. The rankings list will be spoiled for as long as this game amuses them. And it does amuse them; they have been at it for many, many months, as you would see if you cared to look.

And if someone will earn high rating due to his skill and luck, than he wont be able to defend it in longer period.

Of course he will. Having gained a high rating by cheating, do you think that honest play will ensue? Not at all. The cheating game continues.

alanback: The main reason it's not so important is, well ...

spirit_66: I guess there's a lot of cheating around. That's why I decided not to prolonge my membership here.

It is important, Alan, and it doesn't help for you to state otherwise as a fact simply because it's not important to you. Would you play at a backgammon club which had a reputation for cheating? If so, would you debase other's concerns by declaring it a non-issue?

lukulus: I think problem with rating in BG games is not so big. Most of the cheaters are pawns so they will disapper in one month w/o finished game.

The people in question have been doing this for months, for example, one of the cheat accounts was created back in November. Pedro Martinez posted about it http://web.brainking.com/en/Board?bc=26&plla=942630 but they are still here and their game has got bigger with more accounts involved.

They've even, it appears, brought in established players, AcunaMattata, who are doing themselves a disservice by associating with these accounts, whether wittingly or not.

Constellation36: If all these happen then why the site owner does not do something to clear his site from all these morons?


Without changing the rating system, and thus eliminating a core resource of the cheats, it does need someone to do the detective work. Fencer doesn't have the time to do that himself but he will act on a case by case basis if he's given evidence.

grenv: Perhaps we need to set up a team of "agents" to root out the cheats. :)

One of the players is someone who I pointed out to Fencer a long while ago. Filip took action against him but he's back to his games and now seems to be part of a team or interacting with them. Paully has done some cheat busting but has stopped playing and thus stopped detecting.

I think the most effective way to hinder these people is not to allow new accounts to go to 2600 on the first win. If players start at some lower value and work their way upwards through good play then there's no way for new accounts to be so easily misused.

Artful Dodger: Given the luck factor, how does one "cheat" at the game?

Actually I wasn't quite accurate. One of the cheats isn't a rating cheat per se, he's just a player who uses a backgammon engine for his moves. Plays most of his games in private so he can't be checked. (One reason why I think that private moves shouldn't be an option)

The others, as Rod says, are true rating manipulators. There's a pattern of starting new players and winning to get the exorbitant starting rating that the BK formula awards. These then lose to other accounts. There are also mid-range accounts which play and win against other mid-range players to pump up their rating before losing to a top-rated player to boost them, then going back to leech off some more mid-rangers.

I count 6 in the top 10!

AbigailII: Well, I imagine a programmer could implement autoplay in such away that if players A and B are playing, both have one man out, and A likes to mindlessly push buttons while B is a real player, one each turn, A pushes a button, B move is played automatically, and it's A's turn again. Repeat until B has two men out.

That would be pointless. There are unlikely to be enough player A's to make it worth the effort enabling play for mindless moves and some of the player B's, being real players, would want to make their own moves and wouldn't care for automation.

Or maybe I'm misunderstanding your judgementalism?

For those that want to thoughtfully make the moves that are forced, the thinking being in terms of the game not the move, something which you don't seem to appreciate, it is of course the case that autoplay options would be per player. That's exactly how I implemented autoplay in my own version of Ludo.

Of course, in the case of Ludo, it may be that the computer plays an entire tournament, never encountering a play where a player actually has a choice.

It would be rare for a single game to be completely forced but it's certainly possible. An entirely forced tournament, lol, that would be something!

AbigailII: Yep, autoplay of forced moves would speed the play up a lot. I use it when I play real-time Ludo as I can see the pieces move but I might not use it in turn-based play because the game would be too disjoint. Witnessing the chase is part of the fun for me even if the decision making is absent but I imagine that plenty of people wouldn't miss that. Certainly hexkid's Ludo Auto-player for BrainKing was popular when it was available.

pgt: Ludo's fun. You might not think there's any skill in it but it's there and it's a challenge to tease it out.

Pedro Martínez: OK, it must be my bad memory. ... My bad. Sorry.

Lol. Such humility when you think you are correct.

You are, but here's a way to interpret pgt's words that makes him right too.

pgt: Two years ago I gave up my paid membership and promised to renew it when there was a decent autopass facility. Despite a lot of great suggestions (at the time) as to how this should be done, there has been absolutely no action

It hinges on how strictly you want to interpret "decent". I take it that Phil's view is that we've seen action towards a half-assed implementation but that doesn't go any way towards decent.

My own view is that an autopass that's disabled because of the ignorance or whim of one of players isn't autopass in any sense of the word but I have enjoyed being able to play several moves in a row when my opponent (a fan of autopass) was on the bar and I had the cube.

However, what I'm more concerned about - and mystified by - is the complete lack of autopass in Ludo, the game that would most benefit from it!

grenv: Nice try, but you know what an outlier is

Nice try? What's being tried? Oh, you mean trying to get you to expand on your use of the word? Yeah, I do that when I'm not sure exactly what someone means. It doesn't tend to work when that someone assumes that I should know what they mean and must therefore be asking for nefarious reasons.

And if that's the attitude then ... yeah, whatever you say. You're right. Great idea, grenv, perfect choice of metric, it should work wonderfully. No more exploration is necessary.

grenv: Okay, but I'm still not clear why "too many games" is relevant. I'm also still wondering about where the speed-of-play information is to be displayed and how and when is it to be used? Oh, and what's an outlier? ;-)

grenv: I don't think that doubling the number of games will double the number of moves, but if it does then clearly that player isn't playing too many games.

What's the meaning pf "too many" and how is it relevant?

AbigailII: Such a single datum isn't really an improvement over an average. Consider our chess player with a 30-days chess game but who loves a quick game of backgammon.

I do think a time per match metric is better than a time per move one. Those 5 anti-backgammons with the friend would only be 5 matches rather than 500 moves and would therefore distort the average much less.

All: One think I'm not clear on, where is the information to be displayed and how and when is it to be used?

Some considerations:

Whatever metric is chosen, there ought to be one for each type of game, or the games should be broken into categories. Ludo can be played in an instant but chess can take a while to ponder. A moves per day metric could penalise someone who likes to bash out a game or two of Ludo if they ponder long and hard in their chess games.

The metric also needs to be computationally cheap. If, like a moving average, it requires a fair bit of data manipulation per move then it's probably not a good idea. If, like a graph, it requires a lot of storage, ie. values for each move, then it's also very expensive. Bear in mind that the ratings graphs are just a couple of data per match. The utility/cost value for the speed graph would be very low, for all that it's an interesting graph at times.

pgt: What? Give my carrots to those who are the cause of so much suffering and pain?

grenv: Hmmm ....... let's see now .......... oh, I've got just the thing ...

A gibbet!

grenv: At GoldToken there is a clock value shown for each player which gives the amount of time that the game was waiting for that player. Although there's no average available, examination of a few games can show you whether a player is a cheetah or a sloth.

Here, for instance, is a perhaps an extreme example, but one that would pale against some of the BrainKing players, - a player who has taken 105 days versus his opponent who has taken only 16.
http://www.goldtoken.com/games/play?g=4665387

Pedro Martínez: What's "Dumb" in Estonian?

grenv: Then I think that your analogy is both too strong and too weak. It's too strong because Chess without the king is not Chess whereas any gammon game without the cube is still the same gammon game. The scoring changes, as well as a touch of tactics, but not the game itself.

On the other hand, the cube/king analogy is too weak. If you said. referring to the game of Grasshopper itself, "Shall we play Chess where all the pieces are the same, get placed on the board one after the starting with an empty board, and where the pieces must start in the centre and each piece must touch another, and whereby placing any piece which forms lines of opponent pieces which are "bracketed" by that piece and another of the player's get turned into pieces for that player?" then you'd be about right in matching Grasshopper to Backgammon.

Or putting it a different way, if the board was unrolled so that it was flat and went in only one direction and the quadrant boundaries were drawn as hedges and the peices were turned into frogs then Grasshopper could be called Froghopper and nobody would think that it was derived from Backgammon any more than Reversi is derived from Chess.

grenv: Shall we play chess without a king next?

Are you referring to Grasshopper being without the cube?