Random Generator vs Perfect Computer

I am wondering if anyone has ever programed a random legal move generator to play against a perfect computer model. If the games were run continuously with a supercomputer, eventually the random generator should win and new strategies could be studied. Has this ever occurred or is there just not the computing power to make this feasible?

I imagine the computing time would be far too vast until the a random move generator would beat the best current models.  Remember that there are a billion move combinations after move 4 alone, and that number of combinations rises exponentially.  Games involving the best computers can easily exceed 100 moves.

I realized what you may be asking is effectively trying to solve chess by brute force.  I recommend checking out this quick wiki read to show why it's impractical: https://en.wikipedia.org/wiki/Shannon_number

I completely understand this is not an easy task. It's more of a theoretical question. Yes, this is very much like random letters and Shakespeare. If extremely powerful computers were running the program for years, theoretically, wouldn't something interesting happen eventually?

Theoretically, we would find interesting things and likely "solve" chess.  But computationally, it is purely not feasible.

I thought about this before, but the chances of the random number generator beating the computor is like that Infinite Monkey Theorum.

The Earth would probably end before anything could happen in your scenario.

Afaik in talk chess forum, with monkey theorey in chess assuming monkey doing random moves,

if you have 10^52 monkeys, you have 1 chance in winning against Stockfish.

But against perfect computer, there is "zero" chance, as chess is a theroetical known draw game. 

Also, people misunderstand what 10^50 represents.

They think "that's a really big number" and "computers can do really big numbers" so it would take a weekend, or a month... certainly not more than a few months right?

Yes, an infinitely long series of random numbers contains every possible combination of numbers.

A random move generator left to play for an infinite amount of time will play all possible games of chess.

Running the program for "years" would still be an infinitesimally small chance. One billion years is only 10^9 years.

Here's a quiz for you. Which of the following is closest to 1% of 10^50?

a) 10^0
b) 10^1
c) 10^5
c) 10^10
d) 10^20
e) 10^50

Humorously though, a random move generator would fair much better against a perfect engine than a human would... that's because a random move generator isn't biased towards any move, while even a beginner human will play some types of moves more frequently than others.

Theoretically, the OP might have a point, but I don't think he realizes just how bad a true random mover chess engine is. I've watched a couple of them, and it's truly agonizing to watch. :-)

The answer is e by the way. 1% of 10^50 is 10^48.

Super computers operate in the 10^15 to 10^18 range. The age of the universe in years is of the magnitude 10^9.

If you took the world's fastest super computer and mass produced it so that every single human owned one billion of these computers, and then you had all of these computers generate random chess games, and you ran all of these computers for 10 billion years, that'd get you to about 10^46... which is not even 1% of 10^50.

And the 10^50 number is only (about) the number of positions not games.

64 choose 32 for which squares are occupied, then multiply by 32! for which pieces go on them = ~10^50

The number of possible games is usually quoted as around 10^120.

And remember 10 times bigger than 10^50 is 10^51... so 10^120 is pretty big.

@llama47 - So you're saying there's a chance? ;-)

Haha, yes

e

Hush you. You have a US flag. The fact that you can do math on the level of a 10 year old in Europe or Asia shames the rest of us Americans who can barely count to 10.

Also I would say the odds of the rng generator even lasting 20 moves is under 50%