"Come on, Bergmann. People have a couple hundred kinds of bacteria living on their skin, for starters. A bunch more in the gut. That's all it's counting."
He shook his head. "We already thought of that. The count's still too high."
"Well, I
"Fix it," he said. End of interview.
So we fixed it. The problem was simple enough, as it turned out: CAPTAIN was looking within the cells of my own body, and seeing multitudes.
A cell isn't just a cell. It's actually a colony of small cells inside a larger one. Some act like batteries, turning chemicals or sunlight into energy. Some keep the genes from getting all tangled up in each other. But it wasn't always thus: a few billion years ago, way before multicellular life evolved, the ancestors of those small cells lived free in the primordial goop. One day some larger cell tried to have them for lunch, only to find it couldn't finish the job: it engulfed its prey easily enough, but couldn't digest it afterward. The little cells found they could live quite comfortably inside the thing that had tried to eat them. In fact, both parties ultimately found themselves better off than they had been. The host cells could use the chemical energy the little cells put out, while the little cells were comfortably insulated from environmental changes — not to mention other predatory cells that might be a bit better at chewing once they'd swallowed. Everyone was a winner.
They never looked back. We don't even think of the little ones as "cells" any more — we think of them as parts, organelles, give them names like "nucleus" and "chloroplast" and "mitochondrion." But they've still got their own genes, left over from the days when they lived on their own. They still reproduce independently. In a way they are still individuals, living inside our cells like janitors in a high-rise.
That's what CAPTAIN thought, anyway. It had been counting all those different parts in my cells as separate individuals, even separate species. It wasn't even wrong, strictly speaking, just about a billion years out-of-date.
Bergmann wasn't impressed, though. This was the fifth time we'd had to go back and start from scratch.
The problem is, you can't just tell a computer to "go out there and protect the diversity of life." You've got to tell it exactly what life is. It's the sort of thing that geek teachers bring up in grade seven biology classes to try and spark the students' interest. What is life? they ask, thinking the class is going to get all excited about this garbage. Is a salt crystal alive? No? Well, it grows, doesn't it? It "eats" ions from the surrounding solution, it incorporates them into its "body", it "grows" — if you break it, it repairs itself — it heals, get it? — and a fragment, chipped off the main crystal, will grow on its own, so you could say that's reproduction —
Yeah, right. Whatever.
Everybody knows that trees and people are alive. Everybody knows that salt crystals aren't. If that bozo at the front of the class thinks he's going to turn anybody's crank with his lame droning about "living" salt crystals, he's even dumber than he looks.
But the old geezer has a point, you know. There's this list of features that gets trotted out every time you want to describe what "life" is: it breathes, it eats, it grows. It reproduces. The problem is, the list doesn't always work. Viruses can't reproduce on their own: they have to hijack the reproductive machinery of some other cell to do that for them. And common garden-variety Drano — the stuff you use to unplug toilets — that "breathes," if "breathing" is taking in one kind of gas and expelling another. For every rule, there's an exception — something that obviously is alive, even though it doesn't fit the list.
Something else that obviously isn't, although it does.
I had this teacher once. "I may not be able to define an elephant," she told me, "but I know one when I see one." Defining life is a lot like defining an elephant. You just sort of, well, know it when you see it.
But try telling that to a machine. You can wave your hands and fudge all you want, but computers are dumb: they just won't get it unless you spell it out exactly.
Actually, now that I think about it, there is one sure way to tell if something's alive: try to kill it. Only living things can die.
CAPTAIN seems to have figured that much out, anyway.
I’m checking CAPTAIN'S satellite feeds right now. I've got access to a million electronic eyes, I can watch the whole world in lovely shades of infrared and ultraviolet.
I can see chlorophyll sparkling where plants brighten the landscape; I can see black voids where they've been wiped out.
A lot of those, now. More every day.
I watch as the nukes and the lasers touch down at precise intervals across the planet, turning the world into a piecemeal firestorm. CAPTAIN'S hitting eastern Africa today. I try talking to it. If CAPTAIN was human, I'd grab it by the throat and scream, "What in God's name are you doing? You're supposed to be saving the world!"
But CAPTAIN isn't human. It doesn't understand rage, or frustration. My questions have to be precise, logical, framed in the language that software understands. So I type them at the keyboard, ask about biodiversity counts, species loss, remedial measures taken.
The computer responds in kind, gives me statistics and diversity indices, estimates that it will achieve the goals we've set it in two years or less. If you translated that response into English, it would be saying "What are you complaining about? I'm doing exactly what you told me to do. I'm protecting the diversity of life."
Another beam lances down from orbit, and Madagascar catches fire.
Back in test mode CAPTAIN was an obedient lapdog. The moment we committed it to a real mission, though, we let it off the leash. It's a military system, after all. Oh, we came in at the last moment and gave it all sorts of nice new mission objectives, but at heart it was always a creature of the battlefield. It was designed to keep going under conditions of total warfare, to interpret any shutdown command as an enemy
