Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)

LOL - yes, you get it!

The whole premise of the question is flawed - a 1300 and a 2700 do not even play the same game.

rooperi wrote:

Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)

LOL - yes, you get it!

The whole premise of the question is flawed - a 1300 and a 2700 do not even play the same game.

Estragon wrote:

rooperi wrote:

Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)

LOL - yes, you get it!

The whole premise of the question is flawed - a 1300 and a 2700 do not even play the same game.

This idea has been proven many times by immunologists: If you kill something often enough, it will develop an immunity to death.

rooperi wrote:

Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)

Actually elephants are scared of ants. So the elephant will usually resign even before the fight starts. And if the elephant tries anyway the "fight" will most likely end in a draw because the ant will find enough space under the elephants foot to survive.

In other words: it's one of the worst analogies I read on these forums so far

Valentin, your point is evident enough, so it's slightly odd that you feel the need to spell out in detail again that writing sentences and playing chess are not completely random endevours for humans. Of course that's not the case for our monkeys and our random move generator, so regrettably you remain incorrect with this one:

"*However, this same argument cannot be applied with Shakespeare and chess -- because the chance of winning a game depends on executing a sequence of strong moves (at least stronger than your opponent's). While a single strong move here and there might just come up for the beginner, it won't suffice in the end for winning, because the game is cumulative, i.e., the moves are not independent the same way that the numbers in the lottery are.** *

*So I suggest studying some probability theory first before commenting on the subject of it!"*

Again for simplicity, we can *entirely* put this in the context of flipping 10 heads. All we are doing is increasing the available desirable outcomes from say 2/17 (2 good moves, 15 bad) to 1/2 (we're calling a heads a 'good' move). Now, you would probably argue that this isn't a good analogy because we are only arbitrarily defining a head as a 'good' outcome, well here's the crux: *in the game of chess, we are only arbitrarily defining certain moves as good moves.* Of course chess is a complex game and there are countless reasons why we say certain moves are good and certain ones are bad. But this is exactly as unimportant as my reasons for desiring a string of heads over a string of tails when discussing probability.

In exactly the same way, there are many reasons why we define "Alas poor Yorick, I knew him well" as a more desirable outcome than "JIIU(*Y£(**HEWHOIOI" - but in the context of the infinite monkeys both are equally likely to occur, they will find both.

However, as you say, the element of skill does change things when we part with our monkeys or our random move generator. Again, putting this in the context of a coin toss - let's say I've developed a slight technique and I can hit heads 55% of the time. In one respect I am analogous to the 1300 player (though of course, the 1300 has much less than 55% chance of playing a good move), that is simply that I will most likely outperform pure chance. But the 2700 might be analogous to someone who could hit heads 80% of the time. He of course will be far more likely to string 10 heads together before the random coin or the 55%er. Nonetheless if we put these 3 partipants into endless strings of contest against each other, they will all be the first to hit 10 heads at some point.

A recurring response to this seems to be that a 1300 possesses so much mis-information about how to play chess that he will in fact do worse than the RMG. While it's true that his faulty logic will in many cases hold him back, I would argue that it's infinitely more true that the RMG's lack of *any* logic whatsoever will hold it back far more. A 1300 still likely understands opening prinicples, basic checkmate patterns, the concept of taking hanging pieces etc.. what does an RMG know? Absolutely nothing. People are imagining situations where the 1300 plays a move because of faulty logic, well in that case his faulty logic is just as likely to hit the 'right' move as the RMG, so long as it doesn't directly contradict the 'correct' logic. And why should we assume it does? Just imagine playing an RMG! It would be tedious to say the least.

And Valentin, the numbers in a lottery are very clearly not independent in the sense you are using it. You need to have a string of 5 (or is it 6?) numbers or you might as well have nothing, so in that sense they are required to cumulate. You must be being terribly misled by the fact it doesn't matter in what order they come

On another note, arguments involving physical parameters such as the mosquito having a chance to survive an elephant's foot, or me having the chance to outrun Usain Bolt are misleading. Again the probabilites are non-zero (yet for many reasons, incalculable), but in these cases we are involving the laws of physics, which tend to resist being broken.

Estragon wrote:
rooperi wrote:

Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)

LOL - yes, you get it!

The whole premise of the question is flawed - a 1300 and a 2700 do not even play the same game.

Ok, I understand your point, and I do agree that there is a 0% chance the ant will win with his own strength, but alas, that doesn't apply here.

All we need to know is that a 1300 player is capable of seeing a simple blunder sometimes and taking advantage of it, and a 2700 player is capable of making it. It's extremely unlikely, but again, possible.

"Still, the chance of that happening is so small (someone mentioned 1/30^40, which is close) that in practice no one will live long enough to experience this happening in our Universe."

**We agree. ** At least about the random number generator. That's all we want to know -- is it possible.

It has already been brought up that there are drawbacks to the 1300's nature of not making random moves in terms of the question, but it has *already been addressed* that **despite these drawbacks**, you're still probably better off being a 1300 than a random number generator.

I think the answer is yes.... but, to put it in perspective, there is a much better chance of Accrington Stanley beating FC Barcelona 4-0, or San Marino thrashing Brazil 6-0 in the Maracana...

Actually, I see a slight chance: Playing against Tkachiev.

The problem is that Tkachiev is not 2700 (his best FIDE rating has been 2672), and he does not seem to booze that badly since that famous tournament where he went playing being hopelessly drunk.

I have lost a game myself against a player rated 300 points lower. I was totally drunk, and fell asleep (like Vlad) for the best part of the game. I woke up some eight minutes before my time run out, I blizted a dozen of moves to an interesting position, and then... I blundered.

That was a very good lesson. Now I do not go to play when I'm hopelessly drunk.

Tmb: To be fair, there might be some extremely engine-like advanced moves that a 1300 might never end up playing, because it is simply too unintuitive (I could look for an example of what I mean by crazy computer lines if you are curious). The RMG will most likely miss it too, but not always, of course. That is probably the point of the "impossible" arguers.

Of course the above point is just speculation and it's probably not something that can ever be confirmed or disproven. In any case, I do think the 1300 has a better chance to beat the 2700 than the RMG.

Yeah I realise it is that point alone which makes the suggestion even slightly credible that a RMG might do it before a 1300. We could only formally settle it by putting the question into a precise mathematical formulation... and even then we would be able to argue on the scale of the parameters (for one, I wouldn't allow the probability of the 1300 playing this 'counter-intuitive move' as 0). So it was probably silly of me to assert that I know the 1300 would come out on top, as really I just believe so intuitively, and by a very big margin.

Haldane knew a thing or two about elephants and ants:

http://irl.cs.ucla.edu/papers/right-size.html

What if it's a fireant that would actually poison the elephant stepping on it? Then it wins, doesn't it?

Once when I was 1200 I played a series of 7 games against an 2150, and beat him on the 7th!

He got kind of sleepy from playing against me and answering my curious questions about the reasons he played this and that... his attention slipped and he allowed me a 5-move combination at the end of which I hang my queen, but if he takes it it's a back-rank mate.

Less than a year later I was already 1550 (that was when I was 17 and started to learn about chess) - but, of course, through a drop of alertness and a healthy dose of boredom stuff can happen...

Also, if it's a curious 1300 he'll learn things from playing the 2700 many times... he'll also learn about the 2700's penchants, favorite openings and tactical schemes... by losing and losing his rating will stay 1300 of course and not rise - but he learns and learns, about the game, about his adversary...

I guess he'll draw at least several times though before he wins, so if the games are rated, he'll no longer be 1300, his opponent no longer 2700... oh well..

Another thought: if they are later allowed to mix with other players before resuming their games, the ex-2700 may well return to his heights or at least come close - bad habits get formed when you play 1300's too frequently... the 1300 will probably discover that he's now 1700 or 1800... other 1300's will suddenly look silly to him

solskytz, if we say the 1300 is a fast learner and soon becomes higher rated, then he isn't a 1300 anymore. Similarly if we say he gains enough insights into the GM to beat him without improving his game, then we are not asking the same question. For similar reasons Pfren's response is uninteresting, no-one cares about the answer to whether an awake 1300 can beat an asleep GM.

A way to ask the question to avoid this kind of thing, would be to ask:

"If an infinite number of sober, wide awake and in every way in perfect frames of mind players rated 2700 played a an infinite number of 1300's in the same frame of mind, would any of the 1300's win?"

Though not likely it is not impossible !! Anything is possible a 1300 rated player can beat a 2700 rated player...... but a 1300 stregnth player can not beat a 2700 rated player. UPSET happens in chess for example a ZERO rated player from Nigeria Adeyinka Adesina (now rated) defeated a Swizz Grand master http://www.chessgames.com/perl/chessgame?gid=1688509 if anyone got bigger chess upsets please post! thanks

This is just to show that masters may not be as good as they seem...http://www.chess.com/games/view?id=198942

Right, an elephant steps on an ant enough times, and eventually the ant will win. Got it now :)