Hrqls: These are the GnuBg rollouts. According to these, making the 2-point is a doubtful move and the standard drop-and-split is the best for the non-gammon situation and second best even in gammon-save! But in the gammon-go situation the 2-point comes to the fore.

Double match point. Gammons don't matter to either side.

#. Move _________ Win __ W.G __ W.Bg __ Lose __ L.G ___ L.Bg ___ Equity
1. 24/18 13/9 ____ 50.9% _ 20.6% _ 2.7% ... 49.1% _ 19.7% _ 2.8% ... +0.018
2. 24/14 _________ 50.5% _ 17.1% _ 2.5% ... 49.5% _ 17.4% _ 1.9% ... +0.010 (-0.008)
3. 8/2 6/2 _______ 49.4% _ 22.5% _ 3.6% ... 50.6% _ 18.9% _ 2.0% ... -0.012 (-0.030)
4. 24/20 24/18 ___ 49.1% _ 17.5% _ 2.0% ... 50.9% _ 18.8% _ 2.3% ... -0.019 (-0.037)
5. 24/20 13/7 ____ 48.9% _ 19.0% _ 2.4% ... 51.1% _ 20.1% _ 3.3% ... -0.022 (-0.040)

Error: ___________ 0.3% __ 0.6% _ 0.3% ..._ 0.3% __ 0.6% _ 0.4% ..._ 0.006


1-away, 5-away. Gammon-Save for the leader.

#. Move _________ Win __ W.G __ W.Bg __ Lose __ L.G ___ L.Bg ___ Equity
1. 24/14 _________ 50.1% _ 19.5% _ 2.9% ... 49.9% _ 13.1% _ 0.3% ... -0.202
2. 24/18 13/9 ____ 49.7% _ 21.0% _ 3.3% ... 50.3% _ 13.7% _ 0.4% ... -0.222 (-0.019)
3. 8/2 6/2 _______ 49.3% _ 22.4% _ 4.0% ... 50.7% _ 14.0% _ 0.5% ... -0.235 (-0.032)
4. 24/20 13/7 ____ 47.9% _ 19.4% _ 2.6% ... 52.1% _ 13.0% _ 0.4% ... -0.249 (-0.046)
5. 24/20 24/18 ___ 48.6% _ 18.2% _ 2.9% ... 51.4% _ 14.1% _ 0.5% ... -0.253 (-0.050)

Error: ____________ 0.2% __ 0.6% _ 0.3% ..._ 0.2% __ 0.2% _ 0.1% ..._ 0.006


5-away, 1-away. Gammon-Go for the trailer.

#. Move _________ Win __ W.G __ W.Bg __ Lose __ L.G ___ L.Bg ___ Equity
1. 8/2 6/2 ______ 50.3% _ 15.6% _ 0.4% ... 49.7% _ 18.3% _ 2.5% ... +0.245
2. 24/18 13/9 ___ 51.2% _ 13.0% _ 0.4% ... 48.8% _ 20.6% _ 2.3% ... +0.231 (-0.013)
3. 24/14 ________ 51.1% _ 11.4% _ 0.4% ... 48.9% _ 19.5% _ 2.1% ... +0.207 (-0.037)
4. 24/20 13/7 ___ 49.5% _ 13.1% _ 0.5% ... 50.5% _ 21.3% _ 3.8% ... +0.198 (-0.047)
5. 13/9 13/7 ____ 48.5% _ 14.1% _ 0.4% ... 51.5% _ 23.2% _ 4.9% ... +0.190 (-0.055)

Error: ___________ 0.2% __ 0.2% _ 0.1% ..._ 0.2% __ 0.6% _ 0.4% ..._ 0.006


Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen.
Seed 985902322 and quasi-random dice.
Play: 0-ply cubeful prune [expert]
Cube: 0-ply cubeful prune [expert]

(I made a mistake when doing the gammon-go rollouts in that I forgot to set the Crawford-game flag. It's thus calculated as if the cube were live for the trailer. But that's fair enough if you think of it as a 7-pointer and the trailer has won the Crawford game.)

Pythagoras: No, I didn't correct my "mistake", lolol.

I anticipated (nay, hoped for, lol) a wee bit of discussion so I plugged in a qualifier.

I wouldn't say that playing for gammon means a significant difference strategy (though, of course it depends on what makes for significance).

I guess we differ if what we see as significantly different. I picked the game up again a couple of years ago at VogClub. I must admit that gammons were part of the game right from the start - whether played without the cube (first 12 months or so) or with (thereafter) - so my play is more gammon-centred than it would be if I'd been playing here. I do often play for gammons even when not necessary because that win is a thrill and so worth the risk of losing the point. (That's obviously not a good tactic for serious tournament play, so it's a good job that I don't play serious tournaments!)

But when I do play without needing a gammon then it's not a significant difference in game play. There are some differences in the opening moves (though 6-4: I'd make the point in both gammon-go and gammon-save) but the majority of the game is the same for gammon-go and gammon-save. You might have "Gammon..Gammon..Gammon" running in your mind the whole time but it may not be obvious on the board. Closing the home board is always a priority and needs those builders, double-taps are a joy in many occasions (though sometimes a foolhardy one, lol). To me it's more a toning down of aggressiveness and less risk taking. A shift in emphasis towards the race and away from the battle. But, like I said, my base level is perhaps already more gammonish than others.

We're talking about the same things here yet while Alan says "significant difference" and you say "very different", I say "tone down" and "shift in emphasis".

alanback: It should never appear routine.

grenv: Heh heh. My best is only viewable in the fellowships.

alanback: (Obvious line.. ) I've already gathered that impression. Boom boom.

alanback: I think of gammon and the cube as part of the package.

Gammons are part of chequer play while I'm used to having the cube as a separate deal. I wouldn't say that playing for gammon means a significant difference strategy (though, of course it depends on what makes for significance). I think it would be an interesting enhancement to non-cube matches without being alien territory to non-cubists.

alanback: Lol. I was having second thoughts even as I posted, so I just did. My reply's been deleted.

ZEROZERO: Curiosity, not suspicion.

I can understand that frustration as I get periods of play like that against a robot that I play at a different site. I'm glad to hear that nobody25 spotted the "madness" and you had a chat.

investigate what other players are doing here ...

Aye, there are a number of players who do that. It's not investigation as such because usually it's stumbled upon and then brought to the board (specifically as suspected cheating not, like in this case, as an interesting happening). That can only be a good thing because there are cheats who have been exposed by the "investigators".

...But oh well...everybody has the right to do here what he(she) thinks is fun.

Looking at the graphs, rating lists, reading other's profiles, going through games. Oh yes, that's definitely part of the fun!

LOL...didn't know we were that interesting

Thanks for enlightening us, it was interesting! And you write with good humour.

I hope your illness isn't too much hassle and you get better soon. Sounds like you'll be able to make a bit of holiday out of it. Have fun, too!

Pbarb2: It's a pity that you were sickened, Barb, as I had certainly had no thoughts that there was cheating going on. If I discover cheating then I'm "happy" to shout it out.

Pythagoras: Yes, still over 200 mtches going strong. Maybe they were/are experimenting...

Pythagoras: No suspicion, just curiosity. It's a long-standing playing partnership and maybe there was a falling out that day, for instance. It just struck me as an interesting snippet in the Backgammon soap opera.

One goes up.
One goes down.

The matches.

Curious... I wonder what happened that day.

Walter: Good questions. I think some jolly discussion could be had about how the dice distribution should work. :-)

Backgammon Galore assumes a literal domino set and assigns the 21 dominos in a fixed pattern. Each player plays only their own dominos. When the dominos are exhausted the players swap sets to ensure fairness.

Pythagoras envisages 36 dominos to correspond with the 36 dice rolls and each player picks one from what's left in the set. Whether that's one set per player or a single set between them isn't perfectly clear, though I think he's saying one set each.

This program that I've just played gives each player twelve completely random dominoes. Eg, the computer's selection for the current game includes two 4-4s and four 6-3s, while my set has two 6-3s of its own.

With the 36 domino set per player, one round of dominos would be sufficient for quite a few matches and I imagine that games between good and experienced players would become quite predictable. The program's small random set of dominos makes for more variety of games.

Nightstorm: And then the prediction loses pace again... I downloaded that program and played a match. I didn't even have to look at the computer's dominos in order to thrash it. Nor did I particularly plan my own domino usage much. Someone using it to make their moves may go to the top of the rankings, like so many players do with their first few matches, but they'll have a hard job staying there.

Nightstorm: LOL. Well I never! So the prediction gets more likely. Now we just need someone who would buy and use the program in order it to become "champion". Perhaps R**too would be interested.

The game as described by Pythagoras.

alanback: I predict that someone here would have a computer program that would play perfectly every time, would shoot to the top of the ratings and stay there.

I'd think there are a few wee problems with that prediction. The first is that there probably is no such program. Domino-backgammon is rather an obscure variant and so someone would have to be a real enthusiast to write such a program. The individual moves could be done using the GnuBg engine, of course, but there would still a fair bit of programming to handle the mechanics of the game and the interface, etc. That's within the capabilities of a decent programmer. However, playing perfectly would mean choosing the correct sequence of dominos for both the player and the opponent. Given that the branching factor is so high, finding the solution would be best served by having dedicated neural net, which is highly specialised programming. Then that program would have to be in the possesion of someone playing here. And finally that person would have to be comfortable with using a computer to make their moves.

Hrqls: I'd welcome a reduced-doubles variant but also think that no blocks is a detriment.

In the existing Hypergammon I'd welcome the Jacoby rule: Gammons don't count unless the cube has been turned.

Hrqls: Two loose blots and a resonably open enemy table.. I'd have played on for the SIZZLE. But then I do like gammon.

... GnuBg thinks ...

GnuBg 3-ply says it's right on the border between double and too-good-to-double. 1296-trial rollout says it's a definite double and pass, not good enough for gammon.

Czuch: Yep, I won't play singles when they are so unfairly rated. That's one part of the solution. Another part is to make fewer mistakes.

Czuch: It's hard to know if the free fall is due to the runs of luck that Pythagoras was referring to or whether it's a change in your playing. Certainly stress will make me play badly (and it even seems to affect my luck somehow). Sometimes these things happen because you get a set of games which are exploring new territory so you go down because you're making mistakes, but you learn from them and are better in the future. Have a look at the graphs of other players and you'll see similar patterns. Alanback, for instance. His skill cannot be doubted but he has the ocassional bad patch followed by a climb back to glory..

Czuch: "What happened?????"

You played the last month in a dream and didn't notice all those "Your opponent has moved all their pieces out" messages.

alanback: LOL. That's funny!

alanback:

And there's very possibly a correlation in that idea.

Pythagorus: As BBW said: "They can continue to write and think they are writing to you, but you will no longer have to read it.". I tested this by blocking a friend in a game just now. All the messages - his and mine - disappeared and then reappeared when I unblocked him. Presumably, then, the offending diatribe plus any additionas will be in the game record for such time as the writer is unblocked.

Czuch: I'm slow today but now I gotcha. Don't mind me, I'm from sleep deprivation, lol.

And it's a 9100 to 1 happening, so pretty wierd indeed. ;-)

Czuch: Then I'm not sure what you mean by 50 different games. You've only got 51 games and 8 of them aren't Backgammon and now it turns out you haven't moved in a number of others, so... I'm lost. What are you saying happened?

Czuch: Eh? The first link is a pass for your opponent, janka/Amálka, and a 4-4 for you (move #21). The second is a pass for CindyTN and a 4-4 for you (move #9)! Have another beer, mate.

Pythagoras: LOL

Matarilevich: She be past the age of majority plus a little bit more.

Czuch:


http://brainking.com/en/ShowGame?g=1256089
http://brainking.com/en/ShowGame?g=1131653
http://brainking.com/en/ShowGame?g=1313325
http://brainking.com/en/ShowGame?g=1290552

I stopped looking after finding these 4. Your last move in each is a double.

Pythagoras: Yep, it's a one in 9100 chance - an everyday kind of occurence.

Pedro: You checked tham all?

Matarilevich; Yep, and I believe her name's Babygirlle.

alanback: Yep!

alanback: That's much what GoldToken said. There's no need to protect the player from a catastrophic blunder, in fact slightly unfair to the opponent. But they disable the cube when it's dead for both players which, in my mind, contradicts that argument. Even if it isn't an established rule I'd still implement dead cube = no doubling.

Hrqls:

Hrqls: For sure, it's crazy to double a dead cube. The question is whether it's against the rules to allow that option or is the cube disabled only because it's logical, and convenient for the player?

Is it against the rules to double when the cube is dead? I know that Brainking, Vog, IYT and TrueMoneyGames disable the cube when it's dead. So does GnuBg and, I suspect, Snowie. But is that a matter of being sensible or is it a requirement? I've just discovered at GoldToken that at 1-away I'm still be asked whether I want to double or roll the dice.

Hrqls: I don't believe that's the case. There's one formula for the first 25 matches and another for match 26 onwards. Thus at 26, 100 or 10,000 it should make no difference. Achieving stability is a different issue. 25 matches in a luck-based game will often be insufficient to bring a player to their "true" rating but with the current formula that's not accounted for, given that it assumes that there's no luck. The proper Backgammon formula has a diminishing "bonus" factor which lasts until the 399th experience point.

But that's only my understanding from having read (over-complicated) explanations of the Chess formula and from a limited set of observations. The code actually in use at BrainKing is a mystery. Do you have any further data about the points awarded to matched players at different experience levels?

grenv: Er, I'm lost. Could you explain the bit about fewer games? And, er, the bit about new players and the low-initial-rating compensation?

pentejr: It depends on who the winner is. The higher rated winner gets less and loses more. Vice versa for the lower rated player. But the two are not balanced and that makes your point substantially correct.

For instance with me, at 2430, against a player at 2150.
I'd go up 2 for a win or drop 13 for a loss. He would lose 3 or gain 23.
Thus the ratings pool loses 1 if I win and gains 10 if I lose.

Given that I'm never going to win 10 : 1 against any player, the rating pool will gradually increase in value from me playing this opponent. The same will happen with other pairings. This supports Alan's theory that the rise in Backgammon and Hypergammon is a result of the number of games played.

Pythagoras:



Q: Do you have a pencil?
A: No, I've got three.

Q: Do you have a pencil?
A: Yes, I've got three. Which colour would you like?

It's like you said - open to interpretation.

alanback: I've never been to the former Dailygammon so I'm referring to the procedure that BBW was talking about.

BBW: It doesn't protect people who don't realise that they don't need to lose those 6 points. Just like Dailygammon doesn't protect those who don't realise that they don't need to accept only a single.

Yet, I'll say it again. The Dailygammon procedure is open to "abuse" by a subset of players, those who would take advantage of that ignorance. The procedure here always "takes advantage" of the player's ignorance. (And given that cubes and gammons are so new here and resignations in Backgammon have only had consequences for that game and not the match, that ignorance is rife!)

Pythagoras: LOL. I have the same response to maths as I do poetry and song lyrics - my eyes jump right to the end! I have to force them to go back in and engage. Gimme logic and prose!

.. So the equivalence of the 2 ways is obvious. The "problem" with the second is that we don't have immediatelly the several discrete cases, although the procedure of the second way is good for programming as we gain some time by not re-checking cases....

Grenv and I are both into programming. And I dare say we both like efficient code, too.

grenv: Yes, I like that wording. It's longer than most implementations would use but I like it, especially the embolding.

Pythagorus, Czuch, BBW.. ah, what the heck, anyone: That's a good point (P & C), I'll happily offer a cube early than I should to a player who is liable to be scared by it and drop the game. That's taking advantage of a weakness in their understanding of cube use and may be worth 1/4 a point, for example.

It's only a small step to offering a resignation of a lower value in the hope that the opponent will misread the position and accept. If accepted that's a gain of a full point (most usually).

Earlier I called such a resignation action "unscrupluous" and have reconsidered it in the light of the above. But - and maybe because it's too new a change of perspective, maybe because the magnitude of the gain is so much greater, I don't know - somehow it still feels wrong.

grenv: Thank you grenv, I considered "at least" but decided that it was redundant and I also didn't repeat conditions that were carried forward by the "else"'.

Pythagorus: Yet, for absolute, no question of doubt, clarity, I would use your form with the redundancies.

BIG BAD WOLF: I applaud your concern for players who lose points that are "rightfully" theirs. And, indeed, it's the less advanced players who suffer.

The reason I dislike this implementation of resignation is precisely for that reason. Unwary players are losing points that they shouldn't. I've had people throw away entire matches because they resigned a Backgammon when a single was appropriate. That's why I used the word "harsh". Backgammons are extremely rare yet they are awarded routinely by the resignation logic.

When I've sent a message giving much the same info as the post below, informing them of how resignation works and stressing cautious use of it, I've got back "Thanks, I didn't know" responses.

I dislike it that in the standard situation a player can be duped by the occasional unscrupulous player. I deplore an implementation that systematically does so!

Kipling: This is the result of Fencer's implementation of resignation, (which is incorrect when compared to any other house of Backgammon in the Universe).

When you resign you should be asked whether you want to resign a single-game's worth (1 point), a gammon (2 points) or a backgammon (3 points). This is then multiplied by the cube and gives the number of points that you'll lose. Your opponent should then be asked whether they accept your offer of resignation (because they may want to continue play in order to win a gammon, whereas you might be offering just a single-game loss).

Fencer hasn't implemented this dialogue and thus had to make different arrangements. His answer is to examine the position and award points on that basis. This is quite wrong as far as I'm concerned but the rules are that 1) if you have taken a man off the board then you'll lose a single, else 2) if you have any men in your opponent's home table or on the bar then you'll lose a backgammon, else 3) you'll lose a gammon.

You must have had men in your opponent's table or on the bar and hence were set to lose a backgammon. Multiplied by the cube this would have been 6 points.

When your recent opponent resigned they would have had already taken a man off the board.

The important aspect of this ruling is that You must never resign a game until you have made every effort to take a man off the board, because while your opponent can make a judgement about whether a backgammon or gammon is feasible, the automated logic is rigid and harsh in its interpretation.

alanback: I did a test for someone on the Computers board to see how much bandwidth I used. It was a consistent 100k per minute just by making moves. I have 100 messages per discussion board page so if I'd added that to my activity it would perhaps have doubled.
I'm currently using a program to download match pages and create a .mat flie for GnuBg analysis. Because Fencer doesn't show the dice in the moves list I have to visit every page when one of the dice can't be determined from the move, so it can be 40K+ just to get a dice roll! This is easily the most bandwidth-rich gaming site I've seen.

alanback: LOL. That site annoys me. Good for you!

Hrqls: If you're familiar with Html then it's pretty easy. Fetch the page (using a browser control), find the table and extract the records (standard but possibly messy string handling). Now you've got the data and you're in familiar territory...