the great odds ?

Ahh, but in an infinite number of universes....

 True. I did say an infinite amount of time, and that does mean the universe would probably succumb to heat death before it happened, so it is academic.

Frankly I'm skeptical of all of this because I'm skeptical (and perhaps a bit suspicious) of the conclusions which math can reach, or to which it can be put (perhaps it is best to say its "underpinnings").  But that is doubtless another matter entirely...

It does get pretty abstract pretty fast.  It has been said that nature abhors infinity....

Since GMs can beat other GMs, all the beginner has to do is make GM-like moves. And eventually that would happen.

something as simple as someone who doesn't know how to play yet makes a random pawn move and plays 1.e4 which bobby fischer say's is best by test is someone randomly making a good move. When playing outside I asked about 8 random people do they know how to play if answered no I would then ask them to make a random move and one guy played 1.g3 I asked does he at least know the moves he again say's no  but of course another simple fact that he could have just lied to me makes this inaccurate but still you never know...

 Exactly. Which proves that it wouldn't be impossible for me or anybody else to beat the WC. But as I said before, the odds of it happening at any one time are so vanishingly small you might as well say it's impossible.

The problem with this idea is the notion that the patzer plays randomly as well. But they don't play any more randomly than the GM does. 

What a patzer understands and doesn't understand about a position are a function of a lower level of understanding of hte game, bad thinking habits, etc., but they are not making random moves. Rather, they are systematically choosing bad moves frequently! That is a crucial and important difference.

Oddly, a GM playing a machine that did make purely random moves would eventually lose. But playing a person they won't, because the patzer isn't random, but rather, simply flawed.

 You are not thinking on this clearly. The patzer can still make all the right moves for all the wrong reasons.

If you go to http://senseis.xmp.net/?EloRating

and look at the chart, you can see % chances to win based on rating differential. If I am rated 300 higher than you I have a 85% chance to win, 400 points gives me a 91% chance to win.

There is no cutoff. It does not ever reach zero. Given enough games my 6 year old will eventually beat Carlsen. It will be a lot of games, of course. But if we had her rating then we could plug it into a calculator and figure out what the chance is, and it would be a non-zero number.

Now, how about a master winning 100 five-round tournaments in a row against class A players? What are the odds?

Kasparov losing against a/an U1800

The Cardinals making the playoffs this year ... or winning the World Series.

No, the problem is that those odds are not causal, but correlated. It's not rolling a die and the worst player wins on occassion. 

What is at issue is how moves are choosen. They are not choosen randomly by either participant. 

Now, there is a reality that ratings are estimations of ability, and what is crucial here is how accurate the rating is. So with the provision that the rating we're talking about is actually representative of playing strength, there is a limit at which you simply won't win a serious game, let alone a series, no matter how often you play. 

 But the worst player does win on occasion. Otherwise who would bother to play the game, just look at the rating "I am 1597, you are only 1596, I am the better player, you lose."

If you look at the page I linked to, the person rated 400 points above you, statistically, has a 91% CHANCE to win. It is not certain. Play him 10 games, and the odds are you will win one of them.

 Intresting test what the chance of a 1500 Class C player beating a 2000 player.

 Its going to be around 6%.

 really gald to here cause I played a 5game match with one with 75min time control and I won 3.5-1.5

Congrats you did awesome!

 thats why I asked ;) but it seems strange it's only around 6% cause even the game I lost I understood most of what he was doing and where I went to say it's around 6% that would mean all his ideas are alien to me which clearly wasn't the case.

So why play at all, just roll a 10 sided die and write the score down.

There are several issues here:

The %-age chance to win represents observed events. It does not predict future results.

This is because of several factors:

1 ratings are estimations and not measurements of actual ability

        2 skill levels at most lower levels involve a large number of people who are increasing their skill ahead of their rating.

This leads to my caveat that the rating has to be representative of actual ability. 

If we're talking about people who's ratings are not reasonably stable, then all bets really are off. My kid's rating is 700 and his last two tournament he has a 1500 and a 1700 performance. He won't be 700 next month. But if we're talking about well established people who have a solid history of performing at or near to their rating, then it applies.

At that point there's really no question. Because we're not talking about random processes, simply rolling the dice isn't how the question gets answered. 

In a serious tournament, should an expert win against 2400s, that expert isn't going to be an expert very, very shortly and what you're really seeing is rating lag.

In a serious match, an established, stable 900 won't ever beat an expert. A stable, established class C player won't beat an IM. Those things aren't just statistically rare, they're impossible. It's not random.  In a single serious game a stable, established expert isn't beating the WC.