adsenanim comments on Rationality Lessons in the Game of Go - Less Wrong

42 21 August 2010 02:33PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Sort By: Best

Comment author: 24 August 2010 04:28:12AM *  0 points [-]

This reminds me of the general rules of games...

I was recently playing a game of mastermind with a friend

Mastermind is considered a “solved" game, much like Tic Tac Toe, or checkers.

This considered, I was given cause for thought that even though it is "solved" it still presents the ideas of "Learned Rules", "Intrinsic Rules" and "Trial and Error".

For learned rules the idea is that the rules are related or taught, how one should act according to circumstance.

Intrinsic rules are those rules that are obviated, that the situation itself causes the desire of a solution.

Trial and error is the process of clarifying the rules, related to Occam's in the idea of using the simplest rules to solve the game.

The real question is what do we do when a game situation presents us with a flip-flop such as explained in Charles Petzold's book code? (This is a basic computing concept).

Are games representative of real life or are they viable only as a thought experiment?

Can games be more complicated than physical reality?

Comment author: 24 August 2010 02:29:21PM 1 point [-]

The real question is what do we do when a game situation presents us with a flip-flop such as explained in Charles Petzold's book code? (This is a basic computing concept).

I don't understand the question. By flip-flop, do you mean an electronic circuit with 2 stable states? What did you have in mind, in the game world?

Comment author: 26 August 2010 08:41:04AM *  0 points [-]

Sorry for the delay.

Let’s start a Fire.

The Fire requires 3 things: Air(A), Heat(H) and a Combustible(C) so that:

F == A+H+C.

We know that there are many true statements about F:

F == H+C+A

F == A+H+C

Etc.

Let’s say that these are also true:

F != A+A+A

F != B+A+A

Etc.

We also, because of trial and error, can enumerate the false statements, starting with:

F != A+H+C.

Etc.

Continuing with:

F == A+A+A

Etc.

Now this is where the flip-flop comes in:

The true and false of the basic circuit have an extraordinary amount of combinations for the purposes of making fire.

I came up with this idea not only because people learn games through both negative and positive reinforcement, but that many times we only have a partial picture of the possible combinations for a win.

This is redoubled when we think of thing in terms of arbitrary meanings such as air, heat and combustible.

Comment author: 26 August 2010 10:44:09PM 1 point [-]

I still don't understand what the idea is.

Comment author: 27 August 2010 02:37:50AM *  0 points [-]

The idea is this:

Not only that people can learn as much about a game from losing it as they can from winning it, but that they need to loose in order to learn how to win. The flip-flop acts as a helper in the process of trial and error.

The feedback caused by the wiring of two NOR gates of the flip-flop allow this because the switches are controlled by the true and false sets exclusively; one switch is always associated with the true statements and the other with false.

When we start to learn, all possibilities are indeterminate, they can be either true or false; F == A+A+A is just as valid as F != A+H+C.

The flip-flop becomes sort of an ex post facto method of examining the data of the experience depending on win or loss. With a loss there can be mild sorting of possibilities, but the real sorting comes with comparing wins and losses.

Let me know if how I am representing this idea is to brief, it is still in its infancy, and as I have said elsewhere in my posts, I haven’t read everything.

Comment author: 15 April 2011 12:07:05AM 0 points [-]

but that they need to loose in order to learn how to win. Can't people learn from others' mistakes? What do you say?

Comment author: 27 August 2010 03:40:10AM 0 points [-]