backgammon: lack of fishes?

[mattw]

i dont know how to link the original post from the beginners forum but the post is titled "i want to learn backgammon" with is dated within the last several days. here is my cross post:

please explain why there are more sharks than fish in backgammon. if the games are only pro's vs pro's, wouldnt the game eventually become extinct? perhaps that last sentence in not exactly true but try to see my point. if poker only involved sharks against sharks, where would the fresh money come from? there would not be many games. why butt heads against equal, possible better, oppenents? if BG is not +EV, then i am not interested in learning/studying/playing it. i take note of the ratings system is somewhat corrupt, just like handicaps in golf can be munipulated. but if a player is willing to play without a spot, game on.

i will cross post this in other gambling due to i am very curious and want the fullest response.

thanks in advance.

[mattw]

oops. the original post is the books/pulications forum.

[pzhon]

[ QUOTE ]
please explain why there are more sharks than fish in backgammon.

[/ QUOTE ]
It looks that way if you are a fish.

There are many casual backgammon players who expect to lose money, but enjoy playing. Most are even happier to take your money if you are worse.

[Uglyowl]

If ESPN or travel channel can find a way to make backgammon look cool and easy to the dummies out there then things may change.

[Nate tha' Great]

The main thing is that I think the skill differential manifests itself a little bit more quickly in backgammon than in poker. If I'm an intermediate level backgammon player and I play an hour's worth of matches against an expert, I'd expect to emerge as the loser probably in the neighborhood of 75 or 80% of the time. However, if I'm an intermediate level limit hold 'em player and I sit down at the same table as an expert for an hour (not playing heads up, but at a full table), the expert's edge is quite a bit smaller, and he'd probably only emerge with a bigger stack say about 55-60% of the time. It's the fact that the long run *does* take so long in poker that keeps fishes coming back.

Also, backgammon has not been able to benefit at all from online money play because there are some widely available bots/programs that play very close to an expert level.

[pzhon]

[ QUOTE ]
The main thing is that I think the skill differential manifests itself a little bit more quickly in backgammon than in poker. ... It's the fact that the long run *does* take so long in poker that keeps fishes coming back.

[/ QUOTE ]
I disagree. In fact, I think the opposite is true.

A winning backgammon player may win about 1 point +- 10 points in an hour. A winning LHE player may win 1 BB +- 12 BB in an hour. That difference is negligible.

What does it look like for the fish, though? A losing backgammon player may lose 1 point +- 10 points against the backgammon expert. A losing poker player may be supporting multiple winning poker players plus the rake. The losing poker player is losing a lot more than 1 BB, maybe -5 BB +- 20 BB. I think a losing backgammon player has a greater chance of having a winning 2-hour session than a losing poker player.

One difference is that poker is a multi-player game. Losing players can see better players at the table, yet still convince themselves that they are better than average.

At the moment, I think the main difference is that poker seems like more fun.

[ QUOTE ]

Also, backgammon has not been able to benefit at all from online money play because there are some widely available bots/programs that play very close to an expert level.

[/ QUOTE ]
That's not true. Bots play very well, but there are several backgammon servers with money play. Most online money players do not play nearly as well as bots do, yet I have found little evidence of cheating. It is very easy to test whether your opponents are playing better or more like a bot than they should.

[stigmata]

Nice post. Would you mind elaborating on this, thanks:

[ QUOTE ]
It is very easy to test whether your opponents are playing better or more like a bot than they should.

[/ QUOTE ]

[Robertie]

OK, here's the idea. If you're serious about playing online backgammon for money, you buy a program called Snowie and install it on your desktop. Now you get on one of the sites and play a session against somebody. The first indication that he might be using a bot to help him is that his play will be unnaturally slow, especially on moves that are obvious and forced. That's because he has to stop and keep updating the position on his bot before making it on the board.

If the speed of his play makes you suspicious, then at the end of the session you move the file where the match has been saved (generally located in "\Program Files\Site Name\SavedGames" on your computer) to Snowie's "Imported Files" directory, convert it to a Snowie file, and tell Snowie to analyze it. Snowie will come back with an annotated copy of the whole session and an "average error rate" for both you and your opponent. Your rate is what it is. But if your opponent's is too low (under 1 millipoint per move is suspicious), and his play was too slow, then you have someone to be seriously suspicious about.

In general, this almost never happens. I caught one cheater on GamesGrid several years ago, who wasn't even playing for money! He just wanted to get a high rating, and after his rating got high enough, people got suspicious and he was caught and booted off the site. I play for money occasionally on various sites, and haven't run into a cheater in years.

[pzhon]

[ QUOTE ]
Nice post. Would you mind elaborating on this, thanks:

[ QUOTE ]
It is very easy to test whether your opponents are playing better or more like a bot than they should.

[/ QUOTE ]

[/ QUOTE ]
All of the serious backgammon servers let you record the match or money session. Afterwards, you can use a backgammon bot such as Snowie or gnu to analyze this record and summarize how much each players' decisions disagree with the bot's evaluations. Many backgammon players analyze just about every match they play online.

Top human players have an "error rate" of under 4 millipoints per move by Snowie's method of accounting. (One-point matches should be excluded from this, as it is much easier to have a low error rate when there is no cube and you can ignore gammons.) Bots will almost always rate the play of another bot as much more accurate, usually with under 1 millipoint per move of disagreement. Top human players occasionally play so much like a bot that they have an error rate under 1, but no human does this on average. In fact, bots might say perfect play has an error rate of about 2.

If a player with a low rating has an error rate under 4, the low rating is not accurate. If a player maintains an error rate under 2 over several matches, then this is either one of the best human players playing well, or else you should suspect cheating.

After thousands of money games and matches for money, I have only suspected cheating a few times. I have found many more cases of grossly misrated players.

[TheShootah]

How strong would you have to be to profit from playing online. I only have GNU, so how much would I have to be beating GNU, on average? This is probably a tough question to answer. I am not very good so far, so it could be a while before I even contemplate playing for cash online.