Author: fca
Date: 16:53:46 08/14/98
Go up one level in this thread
This thread has contained many correct statements and a few less correct ones. A treatise on probability and inferential techniques is not needed from me, and I'm too pooped anyway, so I'll make but one comment. An important matter in SSS is granularity where the results are non-continuous (e.g., 3.2-0.8 ain't going to happen). Of course Dan's computations implicitly recognise this, but it may be worth spelling this out. The effect of granularity is obviously more important as sample size decreases. Take a 3-gamer: the only possible results for A - B are 3-0, 2.5-0.5, 2-1, 1.5-1.5, 1-2, 0.5-2.5 and 0-3. Referring only to results and not to game-content, I've heard 2-1 referred to as "crushing". Silly when the smallest perturbation in favour of B would have drawn matters. Ditto with the 1.5-0.5 40/2 result from Ischia... ;-) Kind regards fca "The generation of random numbers is far too important to be left to chance" <i wish I had thought of that one, so I'll give you one of mine: "The difference between mathematicians and physicists is best left to the statisticians">
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.