May 29, 2002

9:03 AM

You may have heard of Conway's Game of Life, where pixels "live" or "die" based on a few simple rules about how many neighbors they have. But did you know that in the 30 years since the game was created, Life enthusiasts have (created? discovered?) an extensive catalog of (objects? creatures?) which interact to form some amazing, nifty, grinning, sometimes beautiful, rube-goldberg, occasionally even a little scary patterns often starting from the simplest of building blocks? (Including a Turing machine!) Or that a lone pixel can exert remarkable control over its environment? Now you can see in a few seconds in a java applet, on your desktop, or even on a PalmOS handheld the outcome of simple patterns that, when first discovered, no computer could handle. A mind blowing example of the power of emergent properties.
posted by straight (22 comments total)

One of the interesting things to me about Life is the way people talk about "discovering" these patterns, when it seems to me that, at some level it becomes people creating things. Like that Turing machine, or the firecracker. You wouldn't say that Shakespeare "discovered" Hamlet as one possible arrangement of the 26 letters of the alphabet. On the other hand, the fact that some patterns die out quickly and others keep going forever seems like a discovery, but it's hard to say where the line between the two is.
posted by straight at 9:08 AM on May 29, 2002

What a great post. That lot should keep me busy for a few hours ...

Just previewing your comment, I agree that there is a fine line, but I would err on the side of discovery for a few simple circumstances: simple rule set, some random or pseudo-random element in the algorithm, and some form of localised feedback. It's a bit like fractals: one can draw a great abstract pattern, but only "discover" a visualisation of a fractal. I guess the interesting difference here is that the Game of Life is about discovering processes, rather than patterns.
posted by walrus at 9:16 AM on May 29, 2002

Another way of getting at what I meant: "discovery" in this context requires a non-deterministic solution, such that the outcome cannot be predicted from the initial set of inputs, but can only be arrived at through iterative manipulation of the data set. For me, it's about the amount of intention in the process, and in some cases the art of making a good, informed guess.
posted by walrus at 9:33 AM on May 29, 2002

For the record - this didn't start with Conway. Stanislaw Ulam came up with cellular automata in the 40's, and John von Neumann first used them to prove that a self reproducing machine was possible. Their work was all theoretical, of course. What Conway did was come up with the first practical application and his relatively simple but ingeniously effective rule set, "The Game of Life". But that is just one rule set in a large universe of rule sets. (Wolfram explores the first 256 rule sets in that universe pretty thoroughly in his new book.)
posted by badstone at 9:36 AM on May 29, 2002

Right on, badstone! I was going to include this link to a bunch of other cellular automata rules, but my post was getting pretty long. It seems Conway's version is one of the most popular, though.
posted by straight at 9:39 AM on May 29, 2002

Harry Reid created a two-person abstract game based on "life" entitled Happy Camper. It's quite enjoyable.
posted by Shadowkeeper at 9:40 AM on May 29, 2002

"discovery" in this context requires a non-deterministic solution

In a sense, this dooms us to only being able to make "discoveries" about complex systems, never "creations". However, I expect major changes are afoot insofar as what we will consider "deterministic".
posted by badstone at 9:42 AM on May 29, 2002

Link-er-riffic! Thanks, straight.
posted by evanizer at 10:09 AM on May 29, 2002

What gets my noodle going about Life is that, unlike 'real life' it's not reversible. At every stage when something dies, it's completely lost, and you can't 'step backwards' to see how you got to where you are, unless you happened to record the steps as they happened.

Except possibly at the quantum level, life isn't like that, and it makes a fundamental difference between a chaotic yet deterministic system which appeases our pattern-recognizers, and the amazing thing that is real causality...
posted by kfury at 10:16 AM on May 29, 2002

unlike 'real life' it's not reversible
er... since when was life reversible? yes, at a large scale, life appears to be entropy in reverse, but in the background of the islands of order that life creates are oceans of disorder in the form of heat. If you could reverse heat to get gasoline or corn, we'd have a lot fewer problems in the world. the emergence of Emergence is not the death of thermodynamics. Life is more or less an ordered eddy in the turbulent flow from higher (low entropy) forms of energy (e.g. solar radiation) to lower (high entropy) forms of energy (e.g warm rocks).
posted by badstone at 10:28 AM on May 29, 2002

since when was life reversible?

Dude, didn't you see "Superman: The Motion Picture"?
posted by Shadowkeeper at 11:10 AM on May 29, 2002 [1 favorite]

For a fascinating take on alife by an anthropologist, I recommend Stefan Helmreich's Silicon Second Nature.
posted by muckster at 11:12 AM on May 29, 2002

Entropy, State function collapse, radioactive decay, and the Kaon. The Universe is irreversible on so many levels.

Our universe is not even symmetrical, as Madame Wu has shown us.
posted by vacapinta at 11:23 AM on May 29, 2002

David Brin went crazy for the game of life, apparently...he featured it prominently in his book Glory Season
posted by taumeson at 11:41 AM on May 29, 2002

That space-filler pattern is like the pocket universe created by accident in Greg Egan's Schild's Ladder, right down to the fact that it expands at half of "lightspeed."
posted by kindall at 12:01 PM on May 29, 2002

Besides being something fun to play with, the real value of Conway's Game of Life is that it demonstrates that complex behaviors can come from simple rules. (To say this is certainly not new or original.)

But what I find lacking in the Game of Life--it's major difference from real life--is the ability to tolerate unpredictable conditions. A glider-reflector may very prettily reflect a glider at a 90 or 180 degree angle, as long as the glider hits it in the proper place, and in the proper phase (of both the glider and the reflector). Hit the reflector in the wrong place, or the wrong phase, and you destroy the reflector. Likewise, take a look at the spacefiller in this applet. Very pretty on its own. But place a single block (a 2x2 square of cells, stable on its own) in the field with the spacefiller, and the spacefiller will be utterly destroyed after it comes into contact with the block.

If real life were like the Game of Life, we'd be fine as long as gas molecules and bits of food came into contact with us in the usual, expected way. But have a single atom hit you at the wrong angle, or in the wrong place, and you would rapidly disintegrate.

Anyone know of any cellular automata where structures have more tolerance for running into random, unexpected other structures?
posted by DevilsAdvocate at 12:31 PM on May 29, 2002

That space-filler pattern is like the pocket universe created by accident in Greg Egan's Schild's Ladder, right down to the fact that it expands at half of "lightspeed."

Kindall, when Egan's book came out, some rec.arts.sf.written folks said that the whole "expanding at c/2" doesn't have any real justification other than giving Egan a threat that moves quickly but still allows the rest of the galaxy to find out about and react to it.

Then when I saw that space filler, I made the same connection you did, and knowing Egan's interests, I feel sure he's familiar with Life, so it's very likely this was the inspiration for the book.
posted by straight at 12:46 PM on May 29, 2002

That space-filler pattern is like the pocket universe created by accident in Greg Egan's Schild's Ladder, right down to the fact that it expands at half of "lightspeed."

Kindall, when Egan's book came out, some rec.arts.sf.written folks said that the whole "expanding at c/2" doesn't have any real justification other than giving Egan a threat that moves quickly but still allows the rest of the galaxy to find out about and react to it.

Then when I saw that space filler, I made the same connection you did (it's the main reason I thought of it as "scary"), and knowing Egan's interests, I feel sure he's familiar with Life, so it's very likely this was the inspiration for the book.
posted by straight at 12:48 PM on May 29, 2002

Oh man, I so need to do research today...and yet I can't stop playing with these. Too darn cool. :) I'm with you guys on the spooky space filler...wow.
posted by dejah420 at 1:16 PM on May 29, 2002

Anyone know of any cellular automata where structures have more tolerance for running into random, unexpected other structures?

Yes. You can do this by adjusting the sensitivity of the Life automata. Sensitivity arises from this and a dependence on local vs. non-local interactions as well as the absence of a dampener.

The best example in the real world of the sensitvity of local interactions is the spread of an irregularity through a crystal.
posted by vacapinta at 1:28 PM on May 29, 2002

Yeah, Egan is pretty obviously implying that the "novo-vacuum" (and possibly our own universe) is composed of cellular automata, although he doesn't quite come out and say so. His fictional "Sarumpaet Rules" might be intended to be the rules for a game much like Life.

I just finished reading that over the weekend so it was fresh in my mind to make these connections. Synchronicity.
posted by kindall at 1:33 PM on May 29, 2002

Synchronicity.

I haven't read the book but I was just reading about it (and its physics) here.
posted by vacapinta at 1:43 PM on May 29, 2002