16 Jan 12
Originally posted by SmittyTimeThe order & times people view various pages.
Ooops, I goofed up the quote. What I meant to write was:
Please pardon my ignorance. What sort of "...activities whilst on the site..." indicate machine usage?
TBH I'm amazed that Russ doesn't have a Page View Log facility. It makes making the final decision to hit the "ban" button that little bit easier.
16 Jan 12
Originally posted by VarenkaI don't have the numbers in front of me, but I did analyze Kramnik-Kasparov and both players were below the 85% threshold for the top 3 matchups (though Kramnik barely so if I remember correctly).
Do you know how many of Kasparov's games were analysed? And what his top 20 matching games were (consecutive and meeting criteria for being eligible)?
16 Jan 12
Originally posted by no1marauderTo know that Kasparov is incapable of achieving a certain match up rate, don't we need to analyse most/all of his games and look at the figures for every possible consecutive, eligible batch of 20?
I don't have the numbers in front of me, but I did analyze Kramnik-Kasparov and both players were below the 85% threshold for the top 3 matchups (though Kramnik barely so if I remember correctly).
16 Jan 12
1 edit
Originally posted by VarenkaThe game mods never said "This user is not a cheat". If they ran a batch and it came out borederline, they kept that person in mind and moved on to another player. After some time they would go back and pull another eligible batch of 20 games and check that.
To know that Kasparov is incapable of achieving a certain match up rate, don't we need to analyse most/all of his games and look at the figures for every possible consecutive, eligible batch of 20?
Think of it as a 55 gallon drum of black and white marbles that you can't look inside of. You reach in and pull out a big handful. If about half are white and half are black you've probably got an even mix.
If you reach in and pull out all black marbles you may have a problem. You shake the drum up a bit and reach in again. If they are still all black... you can rest assured something is wrong and there are not enough white marbles in the drum.
You don't need to pull out every marble to know this... but you also have to be careful to check more than once in case you got a fluke batch.
I'm sure you could stumble on 20 games that make Kasparov's numbers seem outrageous... that is why 20 games are pulled and not 3 or 5.
P-
16 Jan 12
Originally posted by PhlabibitI'm not sure pulling out a few batches of 20 is enough. Suppose we said it wasn't humanly possible for a top GM to win 20 consecutive games against other top GMs, how many batches would you have to pull out of Fischer's hundreds of games to prove that it *is* possible?
You don't need to pull out every marble to know this... but you also have to be careful to check more than once in case you got a fluke batch.
I'm sure you could stumble on 20 games that make Kasparov's numbers seem outrageous... that is why 20 games are pulled and not 3 or 5.
Remember, we're not trying to figure out Kasparov's typical or average match up rate. We're trying to find out what is humanly possible at the extreme.
(By-the-way, I've yet to read of an RHP banning that I didn't agree with, so I'm not trying to defend anyone here.)
16 Jan 12
Originally posted by VarenkaThan yes, you need to run them all if you are trying to prove Kasparov couldn't do it. I was thinking backwards where when we ran a batch and the numbers were crazy high we ran another batch to be sure it wasn't a fluke.
I'm not sure pulling out a few batches of 20 is enough. Suppose we said it wasn't humanly possible for a top GM to win 20 consecutive games against other top GMs, how many batches would you have to pull out of Fischer's hundreds of games to prove that it *is* possible?
Remember, we're not trying to figure out Kasparov's typical or average match up rate. ...[text shortened]... d of an RHP banning that I didn't agree with, so I'm not trying to defend anyone here.)
Originally posted by VarenkaThere's no certainties in life. Saying we shouldn't do something because there is an extremely remote possibility that the action might be wrong leads to paralysis.
I'm not sure pulling out a few batches of 20 is enough. Suppose we said it wasn't humanly possible for a top GM to win 20 consecutive games against other top GMs, how many batches would you have to pull out of Fischer's hundreds of games to prove that it *is* possible?
Remember, we're not trying to figure out Kasparov's typical or average match up rate. d of an RHP banning that I didn't agree with, so I'm not trying to defend anyone here.)
It's called "beyond reasonable" doubt - not "beyond any and all, taking into account infinitesimal possibilities" doubt.
