return to indexE-mail me at kwoolsey@netcom.com
Thanks for putting together the wealth of information and insight in gammonline.com. This is a true treasure trove for any backgammon aficionado.
A question on the Quiz Solutions point system: I understand that people could just run the quiz problems by a bot and get an answer, but could you explain why the point system is based on a democratic vote among experts? The premise of the book "New Ideas in Backgammon" is that we can learn from neural net analysis to evolve the overall level of playing (at least in situations where you feel that the bot mightactually be on the right track, which is not all the time). For example,in problem 4 of the Demo issue, my internal neural net agrees strongly with Snowie's best move, yet no one else did.
It would be interesting to see not only Snowie's "best" move, but also the rollout results for all candidate moves (if Snowie is not lost in theweeds). In any event, I realize that any point system will be somewhat artificial, and that the real learning comes from analyzing the explanations attached to the moves.
Thanks again for the contribution of your time and energy to the world of backgammon.
Ron Dumont
If I scored the quiz by rollout results or bot opinions, everybody whoowns Snowie or Jellyfish could perform the rollouts and get a perfectscore. What fun would that be? Since we won't know for sure what the"right" move is anyway, I think it is better to have the score be decidedby a vote of experts. Their vote may not always come up with the bestmove, but the results will be reasonable most of the time. I should pointout that panelists are specifically requested not to use a bot to assistthem in their choices. Of course readers may use any means they haveto get their solutions, but there is no guarantee that a computer rolloutwill generate the top score. Also, it will be more instructive for readersto try to determine the best plays without use of the bots and then comparetheir reasoning to that of the panel.
As for including rollout results, I don't believe this would be beneficial.The actual results are not what is important; what matters is the line ofthinking which leads to a choice of play. Also rollouts can lead towrong results, even when made by the strongest programs. This occurs whenthe program is thematically misplaying the position. The number of trialswon't make any difference because the program will continue to make thesame mistake. It is true that I make use of rollouts in "New Ideas inBackgammon", but as a tool not a crutch. The positions in the book werechosen because rollout results convinced me to change my mind about theposition, and the narrative reflects my new thoughts. There were manypotential candidate problems which I rejected simply because I didn'tbelieve the rollout results.
Kit Woolsey
When playing against Snowie or JF having the pip count at your disposal is agreat help. Before the arrival of such programs the overall importance ofthe race was underestimated. It is clear now that knowing the pip count atall times would lead to more efficient cube play. But is this possible? Arethere any players that keep a running pip count while playing. I suppose itis possible but is it practical? I know it would be very difficult doing sowithout serious interference in your game plan. If it is possible to achivea level whereby you are able to do this pip count almost subconciously,without interfering in your checker or cube play then will it have beenworth the effort? I'm not sure but I do believe it would require a greatamount of practice if it is at all possible to do this and I think you mightbe better off spending this time developing other areas of your game.
Some of the very best players today don't make too many checker play errors.In order to play this efficiently it is necessary to know the pip count inmany positions where it is not the primary factor but it is a factor. Thismeans that unless you were keeping a running pip count you would have to domany pip counts each game which would slow play down or you would have to bevery good at judging positions using various techniques.What would your advice be in order to play more efficiently without keepinga running pip count?
Tony Moutzouris
It is possible to keep a running pip count as the game progresses. Keepingthe actual pip count of both players would be very difficult, but just keepingthe difference in pip count is quite feasible. Simply add or subtract thedice rolls, and include the pips lost when a blot is hit. For example:
Blue opens with 6-3, playing 24/18, 13/10. Blue is up 9 pips.
White rolls 6-1, making the bar point. White gains 7 pips from the rolland 7 pips from the hit, so White is up 5 pips.
It is as simple as that. I'm sure there are players who can learn to dothis almost subconciously. Personally I find the effort takes away from myoverall thinking about the game, so it isn't worth it.
It is not necessary to take a pip count on every play, or even close. Usuallythe only time a pip count is necessary is on a cube decision in a race, orpossibly on a play decision where you have to decide whether or not to breakcontact. When these positions occur, take a pip count. For other positions,while the race is important, getting an exact pip count is not. Simply havinga feel for who is ahead and by how much is quite sufficient to guide you inyour choice. of plays. When I am playing on FIBS or GamesGrid, I will oftenlook at a position and estimate the relative pip count -- then I will havethe computer tell me what the real pip count is. I find that I can getwithin 5 pips (the difference in pips, not the actual pip count) over 75%of the time simply by eyballing the position. This is all I need for mostplay decisions, since these decisions are based largely on a feel of theposition rather than any actual counting. In a normal 11 point match, Idoubt if I ever do a pip count more than 3 or 4 times.
Kit Woolsey
Thanks very much for putting together GammOnline. I'm having lots of fun &hopefully learning something too.
I have a suggestion for the ongoing match format. I've only been joining infor a few days and already I find that I forget what I voted for. Over theweekend I'll be even worse. Would it be possible for the vote results paneto show what my selection was, e.g.
Moves VotesTake 29Pass 21 ian, you voted for PASSorian, you did not vote.Ian Shaw
An excellent idea. I have done this (not quite in the format yousuggest -- your vote appears on a new page), but you still get the infoyou are after. A couple of the vote files from the first couple of gameshave been lost, but most of them are intact and I will save all the votefiles in the future.
Incidentally,there is an easy way for you to hang onto a web page such as the page which acknoledges your vote. Click on "file" in the upper left corner of your web browser, choose save, and save the pagesomeplace on your hard drive where you can refer to it later. Thisway you can do your own archiving.
Kit Woolsey
Here's a dumb question. Is there any point to the Jacoby rule? Thereason usually given is that playing for a gammon with the cube in themiddle is "a waste of time", a strange argument coming from anyone whospends his or her spare time playing backgammon. If this argument makessense, why is a player owning a 2-cube allowed to "waste time" playingfor a gammon without turning the cube to 4?
A better justification might be that it adds an extra dimension ofskill to the game, although a very small and arbitrary one, in judgingwhich positions are good enough to double only because the Jacoby ruleis in effect, a question unanswered in backgammon literature, as far asI know. Perhaps this could be the subject of a future GammOnLinearticle.
Roger Gabrielson
Time is money when you are playing a fish for $, and we are all egotisticalenough to think that the other guy is the fish. We don't want to waste thattime watching our opponent take all year playing on for a gammon when heforgot to double several rolls ago. It is more profitable to give up thepoint and get on with the next game so we can make some money. So that isthe justification for the Jacoby rule. Once the cube is turned, however,we are talking about real money, so allowing a play-on is quite reasonable.
As to when a position is a double only because of the Jacoby rule, I rememberseeing an article discussing positions for which the proper cube actionfor money is double and beaver. This can occur when the doubler is anunderdog in the game but will probably win a gammon if he wins. He needsthe cube jacked up to get his points for the gammon, but if his opponent isstill the money favorite it is a beaver.
Kit Woolsey