Author: David Dory
Date: 00:18:27 08/20/02
Go up one level in this thread
On August 19, 2002 at 05:56:42, Russell Reagan wrote:
>On August 19, 2002 at 05:34:22, David Dory wrote:
>
>I once saw a guy in a game programming forum making fun of newbies who posted
>with lots of exclaimation marks, and just stated things they retrieved from
>their ass. His post was very funny, but I never personally came across one of
>these people, until now...
>
>>DB2 beat the human world champ - and * please * don't even THINK about
>>"statistical validity". It's a MATCH, not a statistical study!!
>>
>>In the Olympics, you run your marathon race ONE time, just one. That decides who
>>wins, who loses. Easy as pie!
>
>
>>If you run the 100 meter dash in 8.0 seconds, you
>>set the world record!
>
>You also get your gold medal taken away for cheating :)
>
>>You don't have to run it 100 times for "validity" or to be "deemed" best.
>
>
>>You do
>>it once, and you're the world record holder.
>
>Again, this is clearly false. You have to run MANY races just to even get to the
>olympics. Even _IF_ you make it there, you still have to run several races to
>win the gold medal.
>
>>DB2 won the match and I'm not "deeming" anything. I'm saying this is something
>>no other program/system has done. That's why DB2 is tops, and every other
>>program is a wannabe/wishIwas.
>
>100% WRONG...you don't win a single 6-game match and become the best. If you
>don't defend your title, you lose it, plain and simple. If you believe that you
>can win it once and keep it, then Fischer is still the world champion and Deep
>Blue didn't beat the world champion. You can't have it both ways.
>
>>Until they can, they're not the champion in man vs. machine chess. They may very
>>well be the WCC or WCCC champs, though, and that's great. And as soon as they
>>beat the human chess champion (which I believe is much more difficult than
>>beating a bunch of other chess programs), in a match, they'll be as good (or
>>better) than DB2. *
>
>It is debatable whether Kasparov was even world champion during this match. If
>you want to say Deep Blue was the human-computer world champion, great for Deep
>Blue. Just make up whatever title you'd like and slap it on Deep Blue.
>
>"TSCP is the Russell's Backyard Human-Computer Champion!" Who cares about a
>title I make up? Who cares about a title you make up? Answer: no one (except you
>apparently).
>
>>Until then, nyet! (though Vincent posts ad nauseum to the contrary). That's OK,
>>I just LOVE reading fiction. :)
>
>You apparently love writing it as well.
>
>>And Uri: Deep Thought and Deep Blue (one), have NOTHING to do with this topic.
>>How long Kaspy has played chess, how well esteemed he is, etc., are irrelevant.
>
>Ummm...if Kasparov's accomplishments don't matter, then I guess it doesn't
>matter whether or not he was the world champion either then. Therefore Deep Blue
>might as well have beaten you or me at chess, and it's accomplishment is nothing
>special.
>
>"Simple as that."
>"Easy as pie!"
>"Until then, nyet!"
>
>LOL...
>
>Russell
If you run the 100 meters in 8.0 seconds in a qualifying heat of the Olympics,
you STILL have a world record (as you would at any certified meet). Whether you
win the Gold medal or not depends on your success in the finals.
It is VERY meaningful despite the fact that it isn't repeated (with all the same
runners in the race), several times for "statistical validity".
A match is not designed to be a "statistically valid" event, obviously.
People who think that need their bolts tightened/loosened just a bit.
The winner of the human champion vs. computer is apparently of great interest!
They're paying a lot of money to make that match in Bahrain as well as the one
in Israel, happen.
The ONLY thing that matters in Kasparov's "esteem" in the DB2 match, is this:
He was (probably), the strongest human player on the planet at that time.
And DB2 beat him.
I can't imagine why some guys have such a hard time giving props to DB2. Maybe
it's just rusting out somewhere in a basement in pieces - but once it was the
best - and it still is the only computer system to beat the world champion.
Please - don't bring up cheating. Kaspy found his conditions for the match did
favor DB2, and he lost. So his feelings are oh so h-u-r-t. So naturally he says
"they c-h-e-a-t-e-d". You could hear the same thing around any grade school.
He could have used a good lawyer and chess second to look over those conditions
before the match. But he did agree to them.
Complaining afterwards, charging they cheated, etc., is just poor sportsmanship.
Russell posted:
>It does determine who wins and who loses, but it does _NOT_ determine who the
>best is, because like you said, it's just a single competition, which in the >big picture doesn't mean anything.
No tournament is designed to statistically determine who (or which program) is
the best. Why should this match have been any different?
and again:
>100% WRONG...you don't win a single 6-game match and become the best. If you
>don't defend your title, you lose it, plain and simple. If you believe that you
>can win it once and keep it, then Fischer is still the world champion and Deep
>Blue didn't beat the world champion. You can't have it both ways.
You misunderstand me. I'm saying DB2 made a first of it's kind, landmark
victory, which no other program/system has yet achieved against the human chess
champ. Like Bannister when he broke the 4 minute mile. Dr. Bannister may retire,
but until someone came along to run a 3:59.9, clearly, no one was faster/better.
I never said DB2 was the current chess program champ versus other top programs
-(such comparisons are pointless) just that it deserves a lot of respect for
it's considerable, unique, accomplishments. It does NOT deserve to have Vincent
drone on about how Diep would beat it up so bad in short order.
That's just bs.
and again, Russell:
>If you only run it in 8.0 seconds once, you'll never even make it
>to the qualifier for the olympics. Put a little thought into what you say >before you type please.
Wrong again, Russell (don't run much?)
If you ran one 8.0 second 100 meter, you'd be a sensation, and (assuming the
wind and other technical factors were within accepted parameters), a world
record holder.
You could run all your other heats in 8.7 seconds, and still win the gold!
So it's VERY doable, notwithstanding the unlikely source of my writing
inspiration. :)
Yes, I should think more before posting. You have somehow misunderstood the
heart of my post. Or my *ss is "thinking" better than your brain, you choose.
Instead of decrying my alleged lack of thought before posting, why don't you
really READ and take a little time to understand what I'm posting, Russell?
My points are this:
1) DB2 reached a victory no other program/system has yet matched.
2) Posting belittling remarks about DB2 is really unsportsmanlike. It smacks of
"I can't beat oh-so famous DB2, (because it's retired), so I'll just say MY
program can beat it easily."
Which is really crap, and frankly, just unsportsmanlike and embarrassing to CC.
Didn't figure Vincent would figure this out, ('cause he's such a courteous
gentleman, and all) but I thought you surely would, Russell.
3) Match victories are NOT and never were intended to be "statistically valid"
exercises. I am NOT talking about DB2's ELO or possible SSDF rating. Just ONE
very important match againt the world champ. (OK, "probably" the world champ at
that time.)
4) I'm NOT claiming any title for DB2 to ever lose or defend. I'm saying just
that it rose to a top and unique spot among chess programs by beating Kaspy.
That's NOT enough games, for a statistically valid outcome, clearly, and I'm not
looking for or expressing an opinion about such a SV outcome with DB2.
I think you will agree that every program would love to beat one of the super
"K's" in a match.
THEN, IMO, that program will be on a par with DB2 (since no actual games can be
played directly with DB2). Until then, no!
David
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.