Author: James Robertson
Date: 17:08:34 04/25/00
Go up one level in this thread
On April 25, 2000 at 16:18:47, Jeroen van Dorp wrote: >>Absolutely biased ""reporting"". > >That is not *biased*, that is a conclusion you can agree or disagree with. >There is _no_ evidence throughout the article that he is biased. He didn't know >how Crafty would compete, so he tried it and he didn't like some features he >saw. If you are biased, you should already *have* an opninion beforehand. >Coming to a conclusion based on observation and data is *not biased*. Lets step through this. You are saying he had no preconceived opinions about Crafty's strength before the match, and that his conclusion is based on the results the evidence he presented in his article. Ok, that is fair. I will prove my point by contradiction. Let's assume he is unbiased. Assume someone has a random number generator that generates random numbers between 1 and 10. That someone does not know if it is biased toward one end of the scale or the other. He runs the generator twice and gets the numbers 2 and 10 (one high and one low). He then concludes that his rng is strongly biased on the low side, and that the 2 proves this; the 10 is just a fluke, an anomaly. Is this statistically sound? It is obvious that it isn't. The only reasonable conclusion that can be drawn from our rng example is that we don't have enough data. Two numbers is insufficient to prove a pattern. Now we go back to the tournaments. He ran two tournaments, one in which Crafty placed 2nd of 12, and in the other it placed 11th of 11. In other words, two conflicting pieces of data, exactly like our rng analogy. In the same way that it is impossible to come to a conclusion in our rng analogy, it is IMPOSSIBLE to come to a sound decision on Crafty's true strength from these two isolated pieces of data alone. I believe this is sound proof that he made his decision based on preconceived ideas instead of his data. Maybe he doesn't like Crafty's understanding or misunderstanding of mobility; that is irrelevent, and his own opinion. What is important is that he said "Crafty is a weak program compared with the best commercial chess programs". It is not sound to come to this conclusion based on the evidence from his two tournaments. Even bad understanding of statistics cannot make this conclusion. The only explanation for his statement is preconceived conclusions, or - in other words - bias. Heck, he even presents a third piece of data, this one also flying in the face of his conclusion: that of Crafty's spectacular rating on the net. :) James
This page took 0 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.