Author: Rolf Tueschen
Date: 15:11:16 06/06/02
Go up one level in this thread
On June 06, 2002 at 17:57:52, Dann Corbit wrote: >On June 06, 2002 at 17:37:54, Rolf Tueschen wrote: >[snip] >>You are right, I confused organism, as the English word, with species. That was >>it what I begged you always to remember - that I am not original in the English >>language. >> >>As to your question. Let me repeat what is wrong in SSDF and equally in your >>defense or explanation of the design. >> >>I take as a basis between the two of us, that you can't get results for the >>variable "chess strength" or performance _if_ you let vary (instead of keeping >>em constant) variables like hardware or the program once without learning >>function, because it was not yet invented, and with the function in modern >>progs. These adds are gaps in computerchess. (When in human chess human beings >>have differences too but _not_ in the absolute sense but relative, and these >>diffs are distributed on a normal graph.) > >Consider a population of beavers. Shall we consider only the males? Only the >females? Only the adults? Only the adolescents? Only the infants? Smile. Of course we do it this way IF we have found out (that is always the criterion) that the results are( totally different) for e.g. males and females. Didn't you know that?? > >Shall we consider only Castor canadensis or will Castor fiber also be >considered? > >If we do a study of a population, it will be impossible to control every aspect, >but you are right -- we must be very careful about what we control and what we >do not control. But I gave you two most important points, where is your answer? > >Fortunately, all of our SSDF experiments have not only separate measurements, >but they are also carefully labeled. Hence, we can separate out those very >factors for which you have concerns. For instance, in the SQL queries I >provided, you can see that I can easily study *only* those tests performed on an >Athlon 1200 machine as a class. No, you don't. We have still not the same basis. You would take results from matches between A1200 and progs without learning and on P200. That makes no sense. > >>Now the worst mistake is your permanent presention of numbers and functions, but >>you always forget to discuss the important question in the design of the >>statistical method. If we begin to talk about calculations we are lost. At first >>we must talk about the variables and their control. Then we talk about the best >>tests. And then on the end we talk about test results and their interpretation. >> >>I wouldn't say we have a bad discussion. I think we have a good occasion to >>explain the possible fallacies of stats. > >What I was asking earlier was: >Exactly what is wrong with the SSDF measurements? In other words -- what would >you do differently and why? I already gave you some answers weeks ago. But before I could make a serious proposal I would like to see the exact details of the actual practice. Bertil: This is not for the public, this is in private for the programmers, this is what I kept in my mind. And therefore I find your question a bit unfair. Why do you write with two "n" in Dann? Rolf Tueschen
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.