Can 1 elo program beat 1000000000 elo program?

[JVMerlino]

Oh yeah. Myrddin, still rated under 2000, is one lucky program. It has won in tournament games when solid and much stronger programs like Cippolino and Tornado have crashed. It got draws when DanaSah and N2 both made horrible endgame blunders. These happened when Myrddin was probably rated around 1600.

Recently, the ancient and theoretically "100% stable" GnuChess crashed when it had a Mate in 3, and Faile threw away a draw in a Rook vs. Bishop endgame -- with no pawns.

Anything can happen....

jm

[michiguel]

It is an interesting question, but one which is unfortunately impossible to answer. The problem is that the Elo formula doesn't distinguish between wins and draws. So if a player gets 1% of the points, it can because he won 1% of the time, drew 2%, or anywhere in between.

So perhaps the answer is "yes, it's possible", but I'm pretty sure you can't prove that based on the Elo theory (H.G. and others are free to correct me!).

It is not taken into account for the ELO theory. The basic assumption is that a gauss curve will approximate well. That is a reasonable assumption when players are up to 2 or may be 3 standard deviation. Beyond that, I doubt about the accuracy, and with such a big difference, the deviation between what it could happen and the Gauss curve could be huge. It is like applying Newtonian physics when objects travel close to the speed of light.

In fact, I suspect that the Gauss curve overestimate the chance of the strong player when they are not that close in strength. I other words, I suspect that upsets are more likely than the ELO predicts, when difference is say > 500. In own my rating program, I do not use a Gauss curve. I use a sort of Boltzmann, in which the "tails" do not fall so rapidly. I am treating it like Free Energy differences between molecules and the chance they have to react. When the difference is < 300, they are almost identical. Beyond that, nobody knows because it will take ages to measure it accurately. Well, maybe it could be done with engines... but nobody really cared anyway.

Miguel

[S.Taylor]

Isn't it a bit like crossing over the road?
We assume that everyone takes precautions other than _knowing_ how to cross the road. But if they fall behind in doing so, things can go wrong even for a genius.

If the biggest expert in any field says to a beginner, "don't even question any word I say", it is true tyat he ought to be able to say so with confidence. But he might make a big fool of himself one day! That's why all humans must be humble.

Once when I was much less good in chess, a certain master NEVER let me even get as far as ever calling check, or almost anything in the world. He played with one minute on the clock aginst my 5.
But another master gave me only one game, and suddenly exchanged his queen for my knight, and resigned next move. I would have thought that was quite elementary, even for someone with 1200 elo, that would NEVER happen.
Chances of a 1200 elo player doing that are about 999 to 1. And the same goes for a 2300 player, perhaps.

[Tommy]

İ asked this question to you before:


İf A can beating B in sometimes...

İf B can beating C sometimes.

"A" must can beating "C" sometimes?İs it true according to you?

İf you are saying "yes",look at this.

1 elo program can beating 2 elo program sometimes.

2 can beating 3......

1 can beating 3 sometimes?

And

1 can beating 10000000000000(forever zero) sometimes?

Best regards and peace.

That depends. Does the 1 Elo program own a hand gun?

regards,
Tom.

[Bo Persson]

the 600 higher rated player is going to win unless he just simply blunders .

The point is finding what is the chance of that player blundering. At these very low ratings the chance is very high. I once beat an opponent 800 rating points stronger than me as he overlooked a mate in one.

But when the ratings get to a certain point, chance of blundering approach 0, and these ratings are several orders of magnitude less than 1000000000.

But, as you have shown, there is a slight difference between always winning and almost always winning.

A program that has only a one in a trillion chance of winning, might just take that chance in the first game. And retire undefeated. .-)