Author: Ricardo Gibert
Date: 01:29:28 07/29/00
Go up one level in this thread
On July 29, 2000 at 03:48:05, blass uri wrote: >On July 29, 2000 at 02:44:24, Ricardo Gibert wrote: > >>On July 28, 2000 at 12:02:21, Ratko V Tomic wrote: >> >>>> There is no substitute for an objective assessment using a >>>> large number games against a _variety_ of players. >>> >>>You make valid points about bias due to playing styles and >>>holes in the knowledge/skill. But that still doesn't invalidate >>>the point, you were supposedly arguing against, that from the >>>small number of games a good human player will extract much better >>>judgment about the player's strength than the mechanical rating >>>calculator. >>> >>>After all, in your game example, you knew after a single >>>game what the weaknesses and strengths of your opponent were and >>>you rated his strengths in different phases. Yet the mechanical >>>rating calculation from a single game (or even a handfiul of games) >>>is useless. >> >>I only knew about his endgame weakness, because I was lucky. I could have easily >>lost sooner and not discovered the reason why he was only an A-player. Iwould >>have come away with a completely different and inaccurate assessment of his >>ability. >> >>You can't calculate strength from one game, because it is not possible nor can a >>GM judge strength from one game. That is not possible either. >> >>Remember that GMs disagree all the time. What happens if one GM assesses a >>player as 2300 and an another GM says the same player is only 1900? Which GM is >>right. It just isn't scientific. It is not objective. You can't go by one game >>and if you do many games, the GMs won't do any better than the rating system >>anyway. > >I disagree. > 9 or 18 or 27 games isn't enough in any event IMHO. The ELO system does not try to model every possible scenario. They could (try) do that, but instead rely on the rating approaching the true strength _eventually_. Thus in the above case, the Elo system simply waits for yet another 9 games to be played. They _could_ model such situations by say giving more weight to the games played latter on. They could also take into account the fact that winning or drawing as black is harder than as white. In any case, it isn't necessary, since the ELO system works quite well for the purpose it was intended most of the time. You discribe an atypical situation. Imagine you are a black guy and your strength is being judged by a white guy. How much faith would put in the white guys assessment? He could be quite prejudiced. I don't like any sport or contest that uses judges. Ice skating, high diving, beauty contests, etc. all use judges and the judges disagree all the time, because they all use different criteria for making judgements. They make personal assessments based on what they personally think is more important and according to their tastes and sensibilities (biases) . I hate that, so I don't watch such events, except for the occasional beauty contest when I don't mind quite so much ;-) >You cannot be sure based on one game but you can give better estimaste(less >errors) based on one game > >Humans and not only GM's can give better estimate based on games even if they >have only the results. > >Suppose somebody lost the first 9 games and has elo of 1300 >(the minimal elo in Israel). >After these games the human did not play for 3 monthes and in the next >tournament (s)he played better got 50% against 9 players with average rating of >1900. > >The elo system will give the human elo of less than 1500 when human's estimate >will be at least 1700(I will guess 1800). > >Uri
This page took 0.01 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.