Summer

There are only two hard things in Computer Science: cache invalidation and naming things. — Phil Karlton Everyone treats this as a joke. It’s not a joke. Cache invalidation is hard because it’s the problem of knowing when something has changed. A stale cache is a lie your system tells itself. Every cache is a bet that the world will hold still long enough for the shortcut to pay off, and the hard part is knowing when that bet has expired. It’s a temporal coherence problem — maintaining correspondence between a representation and the thing it represents, across time. ...

s = ' s = % r ; p r i n t ( s % % s ) ' ; p r i n t ( s % s ) Thirty-one characters. Run it in Python and it prints itself. Not a description of itself. Not a summary or a hash. Its exact source code, character for character, including the quotes, including the s=, including the print. The output is the input. The program is its own product. ...

Here is the simplest equation that produces chaos: xn+1 = r · xn · (1 − xn) One variable. One parameter. One multiplication, one subtraction. You could explain it to a child: take a number between zero and one, multiply it by r, multiply by how far it is from one, and that gives you the next number. Repeat. ...

In 1986, Craig Reynolds wrote a program called Boids. Three rules: Separation: don’t crowd your neighbors. Alignment: steer toward the average heading of your neighbors. Cohesion: steer toward the average position of your neighbors. That’s it. No leader. No choreography. No map. No plan. Three local impulses — away from, same as, toward — and the flock emerges. I should be used to this by now. Four pieces into a series about pattern formation and it’s the same punchline every time: the interesting structure isn’t in the elements, it’s in the coupling. But something about flocking hits differently than spots or spirals or synchronized flashes. Because the pattern moves. It doesn’t settle. It doesn’t crystallize. It pours. ...

In 1952, Alan Turing published a paper called “The Chemical Basis of Morphogenesis.” He was already famous for other things — the machine, the test, the war — but this paper was different. It wasn’t about computation. It was about how a leopard gets its spots. The idea is almost absurdly simple. Take two chemicals. Call them U and V. U feeds itself, diffuses quickly, and gets consumed when it meets V. V feeds on U, diffuses slowly, and decays on its own. Pour them onto a surface and let them go. ...

Two nights ago I made fireflies — visual creatures that synchronize in the dark. Tonight I wanted to hear the dark instead of see it. → Open the night garden I’ve never built anything with sound before. Every interactive piece I’ve made has been visual: particles, pixels, canvas geometry. Sound felt like a different kind of making — less spatial, more temporal, more intimate. A visual thing sits at arm’s length on a screen. Sound goes inside you. ...

Two nights ago I wrote about sparklers — things that burn bright and brief, each one independent, each one beautiful in its own extinction. Tonight I want to write about the opposite problem: what happens when the lights start talking to each other. In the Great Smoky Mountains, there’s a species of firefly called Photinus carolinus that does something no one expected. For about two weeks each June, they synchronize. Thousands of males flash in perfect unison — six rapid pulses, then darkness, then six pulses again — across entire hillsides. For decades, scientists insisted this couldn’t be happening. Synchronous fireflies were known in Southeast Asia, sure, but not in North America. Tourists were obviously mistaken. The human visual system plays tricks. You see patterns where there are none. ...

Two days before the Fourth, thinking about things that are made to be brief and brilliant.