(However, all this is now in the past. The last time Steven accepted to leave unrecorded the first and last names of an opponent was on October 29, 1995).
After each session of go-playing, Steven records this raw information in two places: in a yearly agenda book and on the index cards of his opponents. The agenda book helps Steven to keep track of his games in chronological order and it is the all-important source of his raw data. The index cards were useful at first as a means for allowing Steven to measure his progress vis-à-vis his opponents. This last function has long since lost its importance, but Steven has never abandoned the practice of recording his games on these cards.
(Note, however, that the use of these cards is reserved for those players whom Steven feels he will play with again on an ongoing basis because they live in the same city as he does. Steven does not use these cards for those players with whom he plays on the Internet).
In order to determine which one of the many games that Steven plays may be included in his statistics, Steven eastablished the four ctriteria listed below.
Tied games are treated as losses. In the early part of his go career, Steven treated those games in which his opponent and he had not settled on the value of the komi and which ended in a tie, as losses. (There's a certain logic at work here: afterall, if a game is not won, then it must be lost). Not long afterwards, Steven stopped the practice of recording the result of such games as a tie. A new practice evolved in which, if, prior to the beginning of a game, Steven and his opponent had not settled on the value of the komi for their game, and, also supposing that their game ended in a tie, then, in such cases, rare though they be, theses games are counted and recorded as losses. Steven has no idea whatsoever of the number of games he counted as losses even though they are in fact ties (in other words, Steven did not lose these games, he just did'nt win them).
After reading these criteria carefully, it's obvious that the nature of the games that would survive the application of these criteria would easily be considered by many players as being normal games (normal in the sense of being typical). However, these selfsame criteria would also be responsible for producing another category of games whose nature, unlike those in the category just mentioned, would not be viewed by many players as normal. What are these games? These would be the games played against computers.
Over the years, Steven has been plagued by a nagging dilemma: Should he or should he not not count the games he plays against computers? So far, Steven has succedded in putting off the need of making any decision on this matter by simply abstaining from playing any 19 x 19 games against these machines.
(However, to be perfectly honest, at some point in his carrer, Steven has probably played a game or two against computers, out of curiosity (no, these games were not counted). This would have been at a time when he would not have been fully concious of the wider implications of such an actfrom the point of view of the integrity of his statisticsof playing against a machine.)
Why the reluctance to deal with this issue? At the heart of the matter lies the issue of a missing standard in the ranking of computers. When Steven plays against a human, he knows that he is playing against someone who is ranked at a certain level, and that the hancicap, if any, that is given to Black will produce a game whose outcome will be fair to both players, more or less. But against machines, there is no such standard. The ranking of these machines varies, and it varies not only over time but also from machine to machine. This can produce lopsided results. In such a situation, for example, how would players evalute the significance of a statistical analysis that would be made of, say, victories?
Another factor that militates against the inclusion of games played against computers, a factor just as important as the one on the lack of a standard of ranking, is the fact that computers don't have names. One of the items of information that is recorded on each game is the name of the opponent.