Category Archives: Lectures and Meetings

Demon-haunted worlds

“No equations! It’s after dinner.”

So Jim Crutchfield promised. He was as good as his word, too: for the duration of a lecture that spanned an hour plus, there was not a single equation or table of data to be seen.

The talk was one of the Center for Complexity and Collective Computation’s ongoing series of John von Neumann Public Lectures. No equations or charts, but there were plenty of interesting ideas. I scribbled like whoa to remember even a handful. Before dinner, even getting that many down would probably have been impossible.

The central part of his exposition was, puckishly, taken from a Honda commercial. If you don’t feel like watching the video, it shows a disarticulated Honda Accord, put together to form a elaborate Rube Goldberg/Heath Robinson (your pick) contraption whose execution takes up the whole 2 minutes. He apologized for the product placement, and then gave his team’s back-of-the-envelope calculation of how much energy of the initial impetus to the system was remaining by the time it reached its end: a mind-boggling 10-60 of it. So, he concluded, it wasn’t energy that had moved through the system – it was information.

“But,” an audience member demurred, “energy was involved in setting up the system so that it would work.”

Which Crutchfield cheerfully agreed with. What we’d seen, he argued, was an example of what he dubs intrinsic computing – a physical system that processes input (the initial push) based on past state (the actions that had put it in its current configuration) and transduces it to an output and a final state. Again, much like a big state-machine. (Fair warning: this sketch of the idea is written by a computer science Ph.D. dropout with not much background in physics or biology, based on a one-hour lecture, so I make no guarantees for its accuracy; it’s the impression of the idea that I took away. Time permitting, I would like to dig into some of the actual publications and get a better idea of the math, but I haven’t done that yet.)

Obviously humans make such systems for our own purposes, like writing natural history blogs and playing Skyrim; but if you look at biology and even fundamental physics through this lens, intrinsic computing appears everywhere, from single cells down to turbulent flows – phenomena that don’t have brains or nervous systems. Half-facetiously Crutchfield tossed out the question of what Nature does all this for – “a big optimization problem, to maximize the glory of God?”

Which let him loop back to another major, historical, point – faced with a phenomenon that does apparently complex computation, such as the physics of solar systems or life on Earth, there’s a human tendency to abstract it to a mind with a personality – a “demon”. As these phenomena become understood, the demons are “exorcized” – mystery, as he put it, becomes mechanism.

I’ve given shorter shrift to another major thread of the talk, which is the idea that humans call the behaviour of a system “interesting” when it lies in between the utterly predictable and the totally chaotic – that it has both predictability and variation. And in fact that this isn’t just a human aesthetic, but a characteristic of successful life; and given that, yielded an answer to another half-facetious question:

“Why isn’t everything disorganized? In fact, what are we all doing here tonight?”

Not questions with obvious answers, but being there and not some place else turned out to be a pretty good choice. Thanks of course to James Crutchfield for an engaging talk, and to the C4 for putting it on.


Leave a comment

Filed under Lectures and Meetings