The scientific approach to the game of backgammon. Part6.

The next step, about which I can not begin to say that. Still, there are situations when it is necessary to be optimistic in the game, is rare, but do happen. And, perhaps, not very often. What it is? This is where we are losing badly. That is, when the position is certainly not in our favor, we still lose. There is no option to lose too much. (A little distracted, in backgammon, there are different options for the loss, but as a rule, this speech does not go into the specific position.) When it comes to how to lose either one point or maybe still win, we it is necessary to build an optimistic risk function that takes into account the probability of getting us a good indication of cubes. Then these probabilities should be greatly increased. Why? Russian proverb is - drowning man clutches at straws. AG This is a strategy game? BM Oh sure. But how do we know how we are in the game - good or bad? For example, in the same static position estimate, secondly, by dynamic evaluation position in the game, taken with some simple function of risk. All of this leads to a dynamic selection of the risk function. About choosing how we are, win, loss, or in the middle, we thereby dynamically build the risk function. For example, when a priori position approximately equal to, the risk that the function is really a little bit decreases. That is, it looks like a linear function, a constant which is between minus one and plus units lose anything, starting with the unit value to about one second. About a function of risk, decreasing, a little pessimistic, consistent with what - will go rain or rain will not go - we'll take an umbrella.If we deliberately lose, risk function is greatly increased. If we win very much, we must be sverhpessimistami and a very poor prognosis in the game suggests is far more likely. And the risk of function will become sverhpessimista function. And depending on this preliminary calculation, preliminary generation, we build a dynamic risk function.The fact that I have described examples is specific implementations, statistical estimates of positions, when it should give an advantage, should give the pros. But how often these examples are shown in the game - I find it difficult to say .A.G. You do win in their program? BM Play. This, incidentally, is an interesting question, well, he arose, it would be bad if he did not appear. I'm in the game of backgammon expert, but, of course, relatively speaking, not GM. Although, maybe my skills in backgammon and higher than was my expertise in chess, when I was still playing - a candidate master. That's an interesting point here - why do I lose? I'm still a man, and sometimes give in to passion, although, of course, do not go to the casino. And here in the game depends on me, on my intelligence.