Random: Does it exist? [The example of Deep Blue]

[Skakos]

In May 1997, an IBM supercomputer known as Deep Blue beat then chess world champion Garry Kasparov, who had once bragged he would never lose to a machine. Kasparov and other chess masters blamed the defeat on a single move made by the IBM machine. At the beginning of the second game the computer made a sacrifice that seemed to hint at its long-term strategy. Kasparov and many others thought the move was too sophisticated for a computer, suggesting there had been some sort of human intervention during the game. “It was an incredibly refined move, of defending while ahead to cut out any hint of countermoves”, grandmaster Yasser Seirawan told Wired in 2001, “and it sent Garry into a tizzy”.

Fifteen years later, one of Big Blue’s designers says the move was the result of a bug in Deep Blue’s software. Murray Campbell, one of the three IBM computer scientists who designed Deep Blue, said that the machine was unable to select a move and simply picked one at random. (source)

A random move. Regarded as one of the wisest ever.

But what is "random"? How can such a move, conducted by a programmed algorithm be "random" in any way? Someone programmed this chess program, someone built its circuits. Someone trained it. Someone debugged it. (probably not too successfully) Even though the move seems random, it is actually the deterministic result of many actions designed by conscious beings...

What do you think? Do you BELIEVE in random?

[Mortiscrum]

Oy vey.

Before I get on the the question, I'll note that if memory serves, computers CAN'T act randomly - we can't program true randomness even if we tried. Online die rollers for instance use some complicated formula to approximate randomness, but it's not actually random. Someone with more computer knowledge than me could give a much better explanation, but the algorithm must be reset periodically so that a pattern doesn't present itself. This makes me strongly doubt the story of Deep Blue making a random move because it was stuck.

Anyways - my intuition says yes, randomness exists. But what I think what this question really amounts to is "Does free will exist?" And that's a tough nut to crack.

If free will does exist, than randomness exists. If free will doesn't exist, than randomness can't exist. I do think there is this level of connection between the two as well.

[Renee]

Random movement and random choices both exist and do not exist. They can exist as "random" in one axis or form of movement, and be "not random" in another axis or form of movement.

Example: Neo-Darwinist concept of mutation. The mutation is random in the form of movement which is biological/ psychological, or put another way, life-based. But the change, which is random in the axis or movement form of biology, is not at all random in the sub-atomic change in the DNA molecule that precipitated it to happen. In the DNA molecule the change was not random, it is a change that follows all laws of chemical reactions.

Taking it to your chess-computer riddle or problem: the way the chess program works, is that it carries on and examines moves in ALL THEIR SUBSEQUENT POSSIBILITIES and chooses the most advanageous move. This can only work for X number of iterations of subsequent possibilities. This X is chosen as a limit, in order to make the computer program have a time limit between moves.

The program thus uses ALL possible moves at any time. It tries, for instance, to move its king into a check position, and rejects that move as illegal. It tries to move the knight to some place, where it is completely useless, but it tries that and rejects that move.

Once it has reached its limit of X, the computer then uses a weighted ranking of the current moves, and chooses the highest ranking movement.

It is possible that the highest ranking movement was distributed equally to all available moves.

In this case the computer will choose (and has chosen) a move that is though stupid, or inconsequential, or brilliant, has the same amount of ranking weight as all other moves.

this may be avoided by increasing the X (number of iterations), but that may still produce, by force of circumstances, yet the same result.

It is as if there was a draw of the best possible moves available in the computer's predictive stance, and it uses one move because it is needed to move a piece at the end of its computational iterations or by the time limit it is set at for each move.

[LuckyR]

In May 1997, an IBM supercomputer known as Deep Blue beat then chess world champion Garry Kasparov, who had once bragged he would never lose to a machine. Kasparov and other chess masters blamed the defeat on a single move made by the IBM machine. At the beginning of the second game the computer made a sacrifice that seemed to hint at its long-term strategy. Kasparov and many others thought the move was too sophisticated for a computer, suggesting there had been some sort of human intervention during the game. “It was an incredibly refined move, of defending while ahead to cut out any hint of countermoves”, grandmaster Yasser Seirawan told Wired in 2001, “and it sent Garry into a tizzy”.

Fifteen years later, one of Big Blue’s designers says the move was the result of a bug in Deep Blue’s software. Murray Campbell, one of the three IBM computer scientists who designed Deep Blue, said that the machine was unable to select a move and simply picked one at random. (source)

A random move. Regarded as one of the wisest ever.

But what is "random"? How can such a move, conducted by a programmed algorithm be "random" in any way? Someone programmed this chess program, someone built its circuits. Someone trained it. Someone debugged it. (probably not too successfully) Even though the move seems random, it is actually the deterministic result of many actions designed by conscious beings...

What do you think? Do you BELIEVE in random?

This is a perfect example of a "random" (meaning: not part of any intentional strategy, as opposed to truly random, which it clearly was not, classic lazy word selection) move appearing to be brilliant, specifically because the human opponent giving the machine undeserved credit for a deeper level of thought than the machine was capable of. In essence the human psyched himself out (as opposed to being psyched out by the machine), ultimately the human "lost", the machine didn't "win".

[Skakos]

Random movement and random choices both exist and do not exist. They can exist as "random" in one axis or form of movement, and be "not random" in another axis or form of movement.

You may be right. As in particles where you can only measure speed or position with a relative accuracy but not both. However I am not sure if in both cases the limitations are real or just "technical". In essence any random physical process - on which a random generator of a computer relies upon - could be completely deterministic and still look "random" to us because of our inability to analyze it fully.

[-1-]

You may be right. As in particles where you can only measure speed or position with a relative accuracy but not both. However I am not sure if in both cases the limitations are real or just "technical". In essence any random physical process - on which a random generator of a computer relies upon - could be completely deterministic and still look "random" to us because of our inability to analyze it fully.

Well... a computer program could use the atoms which decay in a slab of matter to point at the next number in a series produced randomly.

I.e. they could number the atoms in a slab, in any sort of way, even sequentially, and create a function that takes the currently decaying atom's "number" assigned to it as a feed to produce a random number in the series the computer is making up.

This is based on some QM theory, which has been proven true for all intents and purposes, and which states that a half life is precisely repeating, but the individual atoms decaying are random in their assignment to decay. (As well as completely indistinguishable from each other before the decay occurs, but that's beside the point right now.)

Mortiscrum's limitation of randomness in computer programming can therefore be defeated.

(

Mortiscrum wrote,
Before I get on the the question, I'll note that if memory serves, computers CAN'T act randomly - we can't program true randomness even if we tried. Online die rollers for instance use some complicated formula to approximate randomness, but it's not actually random.

)

[Woodart]

Mortiscrum's limitation of randomness in computer programming can therefore be defeated.

How? I don't see how the computer could be programed in your example?

[Skakos]

It seems that the algorithm would be based on the randomly decaying atoms. However even in this case one could argue still that the atoms do not decay randomly, it is just that we do not know or cannot know how they decay...

[Surreptitious57]

Randomness pertains to any event with a probability higher than 0 and lower than l. Any event with an absolute probability of 0 or l is non random as it is guaranteed to happen or not happen. However events with a non absolute probability of 0 happen all the time