rec.games.trading-cards.jyhad

Who's the Greatest?

8 messages from 7 participants · 20 November 2002 – 22 November 2002
original thread on Google Groups

L. Scott Johnson

OK. Here's a parallel example so that hopefully people can begin to view this thing objectively: Anne plays 10,000 games of foo in the red region. Barbara plays 10,000 games of foo in the green region. Anne's record is 10,000 wins and no losses. Barbara's record is no wins and 10,000 losses. Now, from this data alone, does it follow that Anne is a better player than Barbara? No. For instance, pehaps the game of foo is chess and the green region is composed of exactly two players, Barbara aka Bobby Fischer and Deep Blue, while the red region is composed of exactly two players, Anne aka Roger Fox and a hamster. That is *not* to say that the above is the case. Only that the "proof" is flawed. It could very well be that the conclusion (Anne is better than Barbara) is true - what is disputed it that the conclusion follows inescapably from the evidence. It does not. If, however, it is reasonable to *assume* that the overall quality of players is statistically the same (by whatever measure is appropriate) in the two regions, /then/ it is reasonable to *assume* that Anne is a better player than Barbara. (Still not proven, of course). That, I think, pretty much covers the meaningful parts of both sides of the debate. -- L. Scott Johnson (sjoh...@math.sc.edu) http://www.math.sc.edu/~sjohnson

Frederick Scott

[ quoted text not captured ] It's a good reduction of part of the argument. Another part of it is just Legbiter and I butting heads on logical proofs, understanding of proper statistical treatment of a subjects, and so forth. I do sense a true difference of opinion about whether different events at different places can actually vary significantly in terms of quality of competition. Legbiter doesn't seem to believe that can be but if he'd experienced what I've experienced as I've played in various places, I strongly believe he wouldn't doubt it. Whether those experiences are applicable to comparisons of top players - for instance, explaining Rob Treasure's record in England in comparison to Ben Peal's in the U.S. is clearly indeterminate. I guess I just sort of resent people waving their hand over that lack of data and somehow managing to conclude that, "We have no way of knowing so I must be right." Lovely. Fred

Jeff Kuta

Frederick Scott <freds64_at_...@removethis.com> wrote in message news:<3DDBF3BB...@removethis.com>... > "L. Scott Johnson" wrote: > > If, however, it is reasonable to *assume* that the overall quality of > > players is statistically the same (by whatever measure is appropriate) > > in the two regions, /then/ it is reasonable to *assume* that Anne is > > a better player than Barbara. (Still not proven, of course). > > > > That, I think, pretty much covers the meaningful parts of both sides > > of the debate. > > It's a good reduction of part of the argument. Another part of it is > just Legbiter and I butting heads on logical proofs, understanding > of proper statistical treatment of a subjects, and so forth. > > I do sense a true difference of opinion about whether different events > at different places can actually vary significantly in terms of quality > of competition. Legbiter doesn't seem to believe that can be but if he'd > experienced what I've experienced as I've played in various places, I > strongly believe he wouldn't doubt it. Just to toss another wrench into the works-- Whenever professionals do random statistical sampling, the magic number is 51 to allow them to consider the results "statistically significant." I'm not quite sure if this applies (or has even already been discussed at length), but I would venture you can't even MAKE a statement comparing two "independent" groups of players unless the sample size is 51 or greater. :)

The Horror

"Jeff Kuta" <jeff...@hotmail.com> wrote in message news:621dd332.02112...@posting.google.com... > Whenever professionals do random statistical sampling, the magic > number is 51 to allow them to consider the results "statistically > significant." I'm not quite sure if this applies (or has even already > been discussed at length), but I would venture you can't even MAKE a > statement comparing two "independent" groups of players unless the > sample size is 51 or greater. :) WTF? Firstly, the less the number of people in a sample group, the more you have to use statistics to compare the results. There is no magic number for doing statistics. You do statistics with what you have. If your sample group is large enough, you don't even have to use statistics because the sample mean approaches the group mean, and statistical errors decrease with increasing sample size. Secondly, I don't know what you are saying about 51 being the magic number for significance. When performing significance tests it is common to use 95% as the significance level, often even more like 99% or 99.99%. I'm not going to launch into an explanation of what this actually means in real terms. It just occured to me that it would require way too much effort to do so, and that anyone who knows anything about statistics would be bored silly by it anyhow. Thalles

Kamel SENNI

If i play 10 000 games with my grand mother, my little sister, may cat, and my fridge (if they have a vekn numlber, of course) and i make 10 000 TW, i will be the greatest player in the world, probably forever. Is this can be real ? Not really... But let's imagine a group of, let's say, twelve players, with one of them, far beyond the others (by the numbers of cards owned, by experience, or just internet access;-))) winning all tournaments (in fact, not all games, just sanctionned tournaments) : this player, with a perfect honesty, will be the best player of the world.... Even, if he haven't played just a game with any others players around the world... In United States, the storyline reports said : "we were ten, we were twelve"... In europe, storyline reports said : "we are 59"... The ratings system doesn't take in account the numbers of players in a tournament, and doesn't include a system for final ropund. In fact a player can win a tournament, and see his ratings falls... In France, the ratings system is seem so ridiculous, unfair, meaningless, that players simply doesn't care. Competitive Players only wants to win tournaments, and the rating race is forgotten.

Jeff Kuta

"The Horror" <tdem...@anhb.uwa.edu.au> wrote in message news:<arhtol$j19$1...@enyo.uwa.edu.au>... > "Jeff Kuta" <jeff...@hotmail.com> wrote in message > news:621dd332.02112...@posting.google.com... > > > Whenever professionals do random statistical sampling, the magic > > number is 51 to allow them to consider the results "statistically > > significant." I'm not quite sure if this applies (or has even already > > been discussed at length), but I would venture you can't even MAKE a > > statement comparing two "independent" groups of players unless the > > sample size is 51 or greater. :) > > WTF? Firstly, the less the number of people in a sample group, the more you > have to use statistics to compare the results. There is no magic number for > doing statistics. You do statistics with what you have. Surely you jest. If gather information on a pool of 10 people representing a certain population, then gather information on 100 people representing an independent yet similarly sized population, you know a lot more about the group from which you sampled 100. > If your sample > group is large enough, you don't even have to use statistics because the > sample mean approaches the group mean, and statistical errors decrease with > increasing sample size. Yes, and your sample group should be 51 to be considered "large enough." Anything beyond that number adds relatively little significant information since you've surpassed the 95% threshhold. > Secondly, I don't know what you are saying about 51 being the magic number > for significance. When performing significance tests it is common to use > 95% as the significance level, often even more like 99% or 99.99%. Exactly. Having a random sample of 51 from a given population gives you the 95% significance level so you can be confident in your numbers. > I'm not going to launch into an explanation of what this actually means in > real terms. It just occured to me that it would require way too much effort > to do so, and that anyone who knows anything about statistics would be bored > silly by it anyhow. Apparently there are lots of people who lack this knowledge on r.g.t-c.j. Jeff

XZealot

> Not really... But let's imagine a group of, let's say, twelve players, > with one of them, far beyond the others (by the numbers of cards > owned, by experience, or just internet access;-))) winning all > tournaments (in fact, not all games, just sanctionned tournaments) : > this player, with a perfect honesty, will be the best player of the > world.... Even, if he haven't played just a game with any others > players around the world... I agree. > In United States, the storyline reports said : "we were ten, we were > twelve"... In europe, storyline reports said : "we are 59"... The > ratings system doesn't take in account the numbers of players in a > tournament, and doesn't include a system for final ropund. > In fact a player can win a tournament, and see his ratings falls... Well let's also consider that the population density of Europe is vastly greater than that of the United States, so it is not unreasonable that you would have a high number of players at your qualifiers. There are only two regions of the U.S. that even come close to the population density of Europe. One is the Northeast where you have such thriving tournament scenes as Seekonk and Boston, and the other is California where you have the L.A. group. In fact, I just drove 800 miles to get to a tournament this past weekend in Atlanta, GA. That is roughly 1440 km, which is like driving from Lodz, Poland to Paris, France. That was for a 10 player tournament. > In France, the ratings system is seem so ridiculous, unfair, > meaningless, that players simply doesn't care. Competitive Players > only wants to win tournaments, and the rating race is forgotten. Well I wouldn't care about the rating system either if I had a regular 59 player turnout for my tournments. I think you have the better end of the deal to be honest with you. -- Comments Welcome, Norman S. Brown, Jr. XZealot Archon of the Swamp

Sten During

XZealot wrote: > Well let's also consider that the population density of Europe is vastly > greater than that of the United States, so it is not unreasonable that you > would have a high number of players at your qualifiers. There are only two > regions of the U.S. that even come close to the population density of > Europe. One is the Northeast where you have such thriving tournament scenes > as Seekonk and Boston, and the other is California where you have the L.A. > group. Can't be that simple. Sweden has a grand total of nine million. The westcoastregion where I play has some two million inhabitants. Gothenburg numbers 450 000. In Gothenburg we're 34 registered players and play casually twice a week gathering aprox ten players both days out of an active group of 15. Monthly tournaments have gone back to ten players, but we'll find a way to up that number. Spring EC-qualifier at Gothcon attracted 35 players. I don't know about the rest of Europe, but Sweden has a tradition of non-profit work. Gothcon (yearly gaming convention in Gothenburg) totals around 1100 unique individuals - no salaries involved at all. If people expect to put in unpaid work - and still get the response as if paid, ie get shouted at sometimes, I believe it's easier to create an active group doing whatever. Especially as the only thing unforgivable is inactivity - punishment is meeted out by means of someone else steppping in and taking care of the acitivty :) > > In fact, I just drove 800 miles to get to a tournament this past weekend in > Atlanta, GA. That is roughly 1440 km, which is like driving from Lodz, > Poland to Paris, France. That was for a 10 player tournament. Ouch! Sten During