# Is there any chance that a 1300 rated player can beat a 2700 rated player?

I think we can all agree that a 1300 cannot beat a 2700 if the 1300 is playing like a 1300 and the 2700 is playing like a 2700.

"I think we can all agree that a 1300 cannot beat a 2700 if the 1300 is playing like a 1300 and the 2700 is playing like a 2700."

After reading this thread, I'm not sure why you would think that.

In any case, define "playing like a 2700." Even genuine 2700 players don't always play moves that a 2700 would normally play.

Yes, and even genuine 1300 players don't always play moves a 1300 would normally play. I'm saying that under any kind of normal circumstances where a 1300 might encounter a 2700 in a chess game when both are in reasonable chess shape and play a game at their respective levels of play, the 2700 will win every single time.

Alright, new direction in this thought experiment. Since people seem to bring in chances, and human conditions, and what not, into the problem, let us replace the organic players with silicon ones. Lets make the problem purely of elo strength. Lets say, pitch 'Belle (2250)' against 'Houdini (3250)'.

Now, who here thinks Belle has a chance to win a game against Houdini in a long string of games? Raise their hands. How many games do they need to play for Belle to come up with a win? How about for a draw?

I expect Belle to win about 1 in 100 games and draw 3-5 in 100 games at long time controls.

There is always a chance however unlikely!

@tmb, sorry I did not see your phrasing of the question, I was just addressing the question as asked by the OP.

You are right that even with your question the answer is yes... but as I explained earlier, only in the same sense that the Red Lion Pub FC consisting of overweight middle-aged men technically has a chance of beating Barca or Man Utd. I.e. in real life it isn't going to happen.

@tmb, sorry I did not see your phrasing of the question, I was just addressing the question as asked by the OP.

You are right that even with your question the answer is yes... but as I explained earlier, only in the same sense that the Red Lion Pub FC consisting of overweight middle-aged men technically has a chance of beating Barca or Man Utd. I.e. in real life it isn't going to happen.

Yes, in real life it is NOT going to happen, EVER.

The wonderful thing about imaginary life is that all things seem possible, and every little child gets a pony.

You could consider winning the lottery as analogous to a random move generator playing 6 'perfect moves' in a game of chess, where in each position there were ~50 possible alternatives. Many people here have argued that something along those lines is impossible, but as the lottery testifies, if enough people participate, or enough time elapses, it will happen. And even then, people can just come along and win the lottery on their first go!

Of course, a 1300 isn't a random move generator, he will have to find more than 6 'perfect moves', and there aren't millions of 1300's playing 2700's every weekend. The odds are ridiculous, but finite. It could very well happen in 'real life', and if it did it would be explained it as a stroke of luck on par with winning the lottery.

Blindly ruling out events which seem close to impossible must leave you often impressed by coincidences, Estragon.

@tmb, by your strict definition of 'impossible', I don't think anything is 'impossible'.

beck15 wrote:

Alright, new direction in this thought experiment. Since people seem to bring in chances, and human conditions, and what not, into the problem, let us replace the organic players with silicon ones. Lets make the problem purely of elo strength. Lets say, pitch 'Belle (2250)' against 'Houdini (3250)'.

Now, who here thinks Belle has a chance to win a game against Houdini in a long string of games? Raise their hands. How many games do they need to play for Belle to come up with a win? How about for a draw?

Not the same. Houdini doesn't make basic oversights. Carlsen and Aronian recently missed a fairly easy tactic in their game, and they are both better than just 2700. Kramnik missed a mate in one against Deep Blue. Stripunsky hung his bishop on move 11 against Onishchuk.

That's not true, all possible outcomes of an event add up to its total probability - 1. We can therefore forcefully define impossible events using our own rules. Any outcome beyond a defined set has 0 probability to occur, i.e. rolling a 7 when throwing a dice.

In the physical universe we rely on physical understanding, theories and laws, to understand the probabilities of events. As these are forever incomplete ( http://en.wikipedia.org/wiki/Popperian ), it is a struggle to completely rule anything out as impossible formally. Nonetheless, even statistical laws such as the second law of thermodynamics can present us with events far more improbable than the 1300 winning, e.g. all of the gas molecules in your room confining themselves to 1 square centimeter.

Tmb86 wrote:

That's not true, all possible outcomes of an event add up to its total probability - 1. We can therefore forcefully define impossible events using our own rules. Any outcome beyond a defined set has 0 probability to occur, i.e. rolling a 7 when throwing a dice.

In the physical universe we rely on physical understanding, theories and laws, to understand the probabilities of events. As these are forever incomplete ( http://en.wikipedia.org/wiki/Popperian ), it is a struggle to completely rule anything out as impossible formally. Nonetheless, even statistical laws such as the second law of thermodynamics can present us with events far more improbable than the 1300 winning, e.g. all of the gas molecules in your room confining themselves to 1 square centimeter.

If you want to argue it is theoretically possible, that's one thing.  But it is practically impossible, just not gonna happen.  Suppose the 2700 blunders his Queen, that's not enough, he will still win.  So it must be presumed he will overlook a mate in one and, beyong that, that the 1300 would see it, and that both players would get to a position where the 1300 had such a possibility in the first place.

Not gonna happen.

In the immortal words of the great philosopher Yogi Berra, "In theory, theory and practice are the same.  In practice, they are different."

Yes, it quite clearly falls into the category of 'extremely unlikely events', we concur on that. But in the immortal words of Lawrence Krauss, "The universe is huge and old, rare things happen all the time."

"Not gonna happen."

Are you aware for how many more countless centuries the game of chess is going to be played? On how many occasions players rated ~2700 are going to play their 1300 friends? I propose you're not.

Also you seem to be aware of the possibility of a 2700 blundering once, and perhaps quite rightly suggest this isn't enough to lose. But I take it you completely rule out they may blunder again? And would this be enough?

I wonder what the probability of self-replicating molecules arising on Earth was?

iam abrar ali and how are u

Tmb86 wrote:

You could consider winning the lottery as analogous to a random move generator playing 6 'perfect moves' in a game of chess, where in each position there were ~50 possible alternatives. Many people here have argued that something along those lines is impossible, but as the lottery testifies, if enough people participate, or enough time elapses, it will happen. And even then, people can just come along and win the lottery on their first go!

Of course, a 1300 isn't a random move generator, he will have to find more than 6 'perfect moves', and there aren't millions of 1300's playing 2700's every weekend. The odds are ridiculous, but finite. It could very well happen in 'real life', and if it did it would be explained it as a stroke of luck on par with winning the lottery.

Blindly ruling out events which seem close to impossible must leave you often impressed by coincidences, Estragon.

Lets say, for every move to make there are a total of 40 different possible moves. So to pick that one perfect move is a chance that is seen as 1/40. Lets say that the 2700 picks up a random defense (not a string of random moves, mind you, but one of the countless opening available as either black or white, which have so far been tested as 'sound') and this defense lasts a total of 50 moves against the perfect play before the 2700 is checkmated or forced to resign. To keep things simple, lets assume that for each of these 50 moves made, there are always exactly 40 move choices for the random move generator. Now do you know what the chances are of making all 50 perfect moves in the game? The answer is - 1/40*1/40*1/40 ..... 50 times. That is, 1 in 40^50. That is, 1.267x10^80. Which happens to be equal to or greater than the estimated number of atoms in the observable universe. Let's say it takes 1 second for both 2700 and the random generator to make one move (50 seconds for the total game), it would take 2.01x10^74 years for that one perfect game to occur. For reference, the universe is only 13.75x10^9 years old. Of course you could argue that the perfect game could be the very first game, but I think we can see that the chances are very very very slim. So slim, you can practically write it off ever occuring.

Don't forget that chess is basically a draw. If the 2700 plays well, as an 2700 can, he has no reason to lose even against perfect play. There's a relatively wide margin of error in chess, which still leaves you in draw land.

2700s generally know how to stay there when necessary - which is part of their skill set (not obvious or common to someone rated in the 1300's, or even 1800's for that matter), and frequently a strong player can build a safety net to 'minimize damages' when they recognize that they've got themselves in some 'soup' against weaker opposition.

Perfect play a win does not guarantee.

By the way, this thread is fairly old, and last time I checked the score was : 2700 - 54, 1300 - 0

I'll keep you updated and we'll have multiplex with the alternate universes in our next broadcast

"Perfect play a win does not guarantee" (solskytz)

That's actually a very good point. Rather than arguing endlessly that epsilon > 0 (I think everybody understands that), I would be interested if our apprentice mathematicians come up with a model to assess drawing chances of a given elo strength against 'near perfect play' (in the computerish sense of the word).

Or to give a practical example : what would be the chance of a strong GM to at least draw a match against a 'near perfect player', if we assume every draw counts as a win for the human ?

Come on guys, give me a mathematical model for this rather than splitting hairs endlessly with your randomly generated 'to be or not to be here till the end of the known universe' monkeys...

hicetnunc wrote:

"Perfect play a win does not guarantee" (solskytz)

That's actually a very good point. Rather than arguing endlessly that epsilon > 0 (I think everybody understands that), I would be interested if our apprentice mathematicians come up with a model to assess drawing chances of a given elo strength against 'near perfect play' (in the computerish sense of the word).

Or to give a practical example : what would be the chance of a strong GM to at least draw a match against a 'near perfect player', if we assume every draw counts as a win for the human ?

Come on guys, give me a mathematical model for this rather than splitting hairs endlessly with your randomly generated 'to be or not to be here till the end of the known universe' monkeys...

If we can solve this, we actually answer another very interesting question.

If a superGM (2800+ rated) has a non zero score against perfect play, we can deduce the ELO rating of a perfect player, therefore the maximum possible rating. I always thought claims that 10000 ratings are possible is nonsense, I actually wonder if it's as high as 4000