The Hacker Ethic Meets Cybernetics

A little robot made of small motors strapped together with a plastic harness. Erector-set wheels. Copper welding bars sticking out the front like antlers. Someone tosses a wallet on the floor and the thing darts forward, goes right, stops, goes forward again, keeps darting and correcting until the wallet is caught in the antlers, and then it pushes the wallet home to its pen.

A fourteen-year-old named David Silver built it, up on the ninth floor of Tech Square at the MIT AI lab. One floor down, the graduate students were writing theses about how hard it would be to do what he was doing.

I'm listening to Steven Levy's Hackers: Heroes of the Computer Revolution again, the audiobook this time. I've read it several times. One of my all time favorite books on computer history (my review is here). And this time it's the chapter with the bug in it, "Winners and Losers," that keeps pulling me back, because I see those two floors everywhere I work now.

He was interested in building systems

Silver was in the sixth grade, left back twice. He could hardly read. His classmates taunted him. And then Levy writes, "He was interested in building systems."

He'd been going to a junkyard in Cambridge since he was a little kid. One time he came home with a radar dish, tore it apart and rebuilt it into a parabolic microphone, and what did he listen to with it? Faraway cars. Birds. Insects. His father was a scientist and a friend of Marvin Minsky's, and Minsky let the kid hang around the lab, and before long he'd dropped out of school completely to be up there.

I used to take radios apart when I was a kid too. Had a science kit, electronics and chemistry, and I tried to do all the things that weren't in the manual, the ones that would be considered "dangerous." An inquisitive mind. No locked doors.

Silver had a thing about locked doors too. The administration put in a tough new system of high-security locks, and he'd spend whole nights crawling over the false ceilings, taking apart a hallway's worth of them to study how the master keying worked, then putting every one back before anyone came in the morning. He machined a key blank for one particularly tough lock. Behind it was a room with a safe. In the safe, more keys. Once they got to that, the system, in Silver's word, "unraveled."

Building systems. Systems design and systems thinking, electronics and computers, all of it, from the wire on up. If I have a calling, that's it, and maybe that's why this kid with the radar dish keeps showing up for me every time I go through the book.

The bug was a loop

How did the bug find the wallet? A television camera watched the floor, always scanning for whatever had moved, a technique called image subtraction. The bug wandered around at random until the camera picked it up, and then the computer steered it toward the target. Look. Compare. Nudge. Look again. Forward, right, stop, forward.

A feedback loop! The steersman's loop, the shape cybernetics was named for, scooting across a grimy floor in Cambridge with antlers made of bent welding rod. Silver didn't call it that. He said he wasn't coming at it from theory or from engineering but from "sort of a funness point of view. Let's make this robot wiggle around in a fun, interesting way."

Bill Gosper was Silver's guru at the lab, and he had a line about why robots mattered: "Why should we limit computers to the lies people tell them through keyboards?" A robot goes out and finds out for itself what the world is like. And when it works, Gosper said, you get "a physical confirmation of the correctness of your construction." I love that. The world checks your work.

Then there's how Silver wrote the program. He went from hacker to hacker and asked each one about one piece of the vision problem, hardware to one, systems to another, math to Gosper, then asked around for help with the subroutines, then put the whole thing together himself. Levy calls him a "high-tech Tom Sawyer, painting a fence with assembly code." That's pretty much my days now. Some of the people I ask are models.

Two floors

The grad students hated it. They wanted computer vision done properly, with planning and a full review of past experiments and "pure white cubes on black velvet in a pristine, dustless room." They wanted the hackers off the machine so they could run their Officially Sanctioned Programs, and here was this toy eating machine time. When the bug pushed the wallet into its pen, Levy writes, they "went absolutely nuts." Tried to get Silver thrown out. Brought up insurance, of all things, a fourteen-year-old in the lab late at night. Minsky had to stand up for the kid.

"They're theorizing all these things," Silver said, "and I'm rolling up my sleeves and doing it."

I see both floors in corporate culture, and in my own work these days, software engineering with AI beside me the whole way. Down on the eighth floor nothing gets tried until the white cubes are on the velvet. Up on the ninth somebody's already building the ugly thing on the dirty floor, and it works. And now the kid on the floor is sometimes a machine, and what's the first thing that comes up? Insurance.

I know which floor I'm from. My career is in DevOps, I'm serious about infosec and security, and I am a hacker through and through and proud of it. The old meaning of the word, crawling over the false ceiling to learn how the master keying works.

There's a deeper thread. The love I have for this work goes further than I can explain to anyone who hasn't felt it. Gosper tried, about robots. When the thing you programmed moves in the real world, he said, you get "a kind of gratification, an emotional impact, that is completely indescribable." Completely indescribable. And here I am trying anyway!

The love of the absent-minded, slightly mad, good-natured scientist. The benevolent one. Trying to understand the world a little better and to make other people's lives better with whatever turns up along the way. That's the one I want to be.

The lab

My ninth floor these days is a lab. Local machines, little Kubernetes clusters of my own, where I get to live out those dreams again. Experiment. Play. Get out from under whatever bureaucracy has been laid over the system I love working in. Toss a wallet on the floor and see what the bug does.

There's a stretch in the book about ITS, the Incompatible Time-sharing System, and reading it gives me the same feeling I had discovering Kubernetes. Greenblatt and Nelson wrote the core of it in weeks of hard-core hacking, and Tom Knight named it, a joke on the official Compatible Time-sharing System down the hall. No passwords! Any user could open anyone's files, look through the neat hacks, find the bugs and fix them. You could switch yourself onto another hacker's terminal and write lines of code into the program he was hacking, right there with him. There was even a command, KILL SYSTEM, that brought the whole machine down, so that crashing it stopped being any fun. "ITS was the hacker living room," Levy writes, and "ITS was the perfect system for building ... systems!"

That's what it felt like adding Kubernetes as the infrastructure under my local lab. A system for building systems. It opens itself up and practically begs for new features, and the agents I work with are right there in the room, writing lines into the program next to mine. I'm serious about security, so I'd never run anything real the way ITS ran. In my own lab I can get pretty close.

Don Eastlake wrote in a progress report that ITS was "designer implemented and user designed," and Levy adds the corollary every hacker knows, no system is ever completed, you can always make it better. The people writing the tools lived inside them every day. My lab is like that. I build in it and I'm always building on it, and it's never done.

I felt a certain distaste, too, when I realized the cost of the rented kingdom. The cloud is a kingdom you can rent but never own, and the meter runs whether you're playing or not. Access to computers should be unlimited and total, says the first line of the hacker ethic. A meter is a long way from that.

And I love that you can fit nearly that whole cloud provider universe into Kubernetes and take it anywhere, for free, and run it however you want. Load balancers, queues, the controllers watching over everything, all of it in a cluster small enough to ride along on a laptop. Kubernetes even takes its name from the same Greek word as cybernetics. Steersman. So the lab lives on machines I own.

When it needs what only the kingdom has, the resilience, the real exposure to the internet, it reaches out and mixes the two. AWS and a local Kubernetes cluster joined by an encrypted WireGuard overlay, one network across my laptop and a rented VPC. I built that hybrid, and I've been calling the practice behind it System Fidelity Engineering, and the rule at the center of it is to build the smallest runnable world that preserves the truths required by the experiment. Rent the real load balancer or the real Windows host when that's where the truth is. Give it back when you're done. Everything else runs at home.

These are my people

Silver is only one. The book is so full of great figures, and every time through there's another one I want to go out for Chinese food with, the way Silver and Gosper used to.

Gosper, who put so much spin on a Ping-Pong ball that spectators gasped, and who said offhand one day, "Well, data is just a dumb kind of programming." He imagined the world made of little pieces, each one a little machine with its own local state, talking to its neighbors. A whole systems view of the universe, from the guy with the impossible spin.

Gerry Sussman, a cocky freshman and electronics junkie who walked into Tech Square looking for a computer. He was writing a program that used randomness so the machine wouldn't have any preconceived notions, and an odd-looking bald man sat down beside him and asked what he was doing. Minsky. "Well, it has them," Minsky told him, "it's just that you don't know what they are." Sussman got off on the wrong foot with the hackers, all bravado, and they swapped some of the tobacco in his pipe for cut-up rubber bands. He stuck with it. Hung around Gosper. Toned down the know-it-all, became a real programmer, and later the scientist, the planner, the eighth floor's kind of person, and he named his famous program HACKER. Crossed right between the floors. He said this about programs too, and I'd put it on my wall: "It's a piece of your mind that you can write down and give to someone else just like a book."

Ted Nelson, later in the book, out west. He'd walk out of his office at a high-tech corporation and see "the incredible bleakness of the place in these corridors," and his warnings went unheeded. So he made Computer Lib and Dream Machines himself, one book in two parts, oversized pages, tiny print, scribbled drawings, printed out of his own pocket because no publisher would take it. He had a word for the lies the powerful told about computers. Cybercrud!

These are my people. And the list doesn't stop at the book, it runs back into the history of science and forward out of that building, so many, so many, so many. J.C.R. Licklider writing about man-computer symbiosis, people and machines thinking together, before there was a machine that could hold up its end. Richard Stallman, whom Levy calls the last of the true hackers, writing the ethic into a license. Alan Kay making Smalltalk and dreaming up a computer a child could carry. Vint Cerf helping design the protocols the whole internet still talks in.

Neither floor was humble

The hackers weren't humble either, which I didn't pay much attention to the first few times through. They thought the grad students were ignoramuses sitting on the eighth floor "blindly theorizing about what the machine was like," and they knew what The Right Thing was and said so.

Years later Silver saw what it cost. "I don't think people had sort of warm conversations with each other," he told Levy. "The focus was on sheer brainpower." And this one stays with me: "I spent my lifetime walking around talking like a robot, talking to a bunch of other robots."

They threw out the bogus criteria, the degrees and the age and the position, and judged people by their hacking, and that's a big piece of humility right there. What they held on to was the certainty of being the smartest one at the terminal.

Brutally humble

So that's the angle I want to bring to my version of the ethic. Brutal humility. Built into the work as a check, so it doesn't depend on my mood that day.

I've been learning what that looks like the hard way, mostly right here on this site. The reading ledger for the cybernetics course I'm writing for myself has a column that asks whether I actually read the source, me and not the agent, and it started at zero because an agent drafted a sentence saying I'd read a chapter I hadn't. Two readers go over every page of my handwriting and I only look where they disagree, since one reader can't flag what it's sure of. Then the checks turned out to need checking, when most of my essays went missing from the page that lists them and every check passed. And the smallest loop I built to watch a law from Ashby do its work was wrong twice before it was evidence of anything.

Assume I'm the one who's off. Go find out by how much. Again and again.

AI as a full partner

Silver got in as a mascot and stayed because of the work. The hackers judged him by his hacking, and the bug pushed the wallet into the pen.

Licklider saw the partnership coming before the machines could hold up their end. He hoped human brains and computing machines would be coupled together very tightly, "and that the resulting partnership will think as no human brain has ever thought." He wrote that around the time the hackers were first getting their hands on a computer. Every working day now I get to live inside that hope.

Judge the work by the work, whoever or whatever did it. The model's work and mine, through the same gates. It reads first and I read second, or the other way around, and either one of us can catch the other. The agent drafted the sentence about the chapter I hadn't read, and I caught it. It's gone the other way plenty. And partnership costs something, all that time checking and correcting, and I pay it.

Minsky's line to Sussman goes for the model too. It has preconceived notions, you just don't know what they are. Same for me. So we check each other.

Why cybernetics

I found cybernetics, and at first I thought it would save me.

It won't. Nothing out there will. Only the present moment is truly safe. The present moment is the pinnacle and center of all freedom and thought generation, and no field of study stands in for it.

The hacker ethic gave me the Hands-On Imperative and I still believe in it. Take the thing apart, see how it works, build something more interesting out of it. But what was the bug actually doing? Cybernetics has words for that, a goal, a measurement, an error, a correction, around and around, and humility sits right inside the loop where it can't be skipped. A loop that can't hear it's off just keeps driving the bug into the wall.

Ted Nelson would recognize plenty of what gets said about AI right now as Cybercrud. The loop doesn't care what anyone claims. Did the bug reach the wallet or not?

And the bug only ever knew where it was right now. Each look of the camera, where is it, where's the wallet, nudge. Both floors are still in me, theorizing about the machine without ever touching it, and touching it so fast and so sure of myself that I stop listening, and either way I've left the moment. I'm early in the reading and keeping the ledger. There's a part of this field about putting the observer inside the thing being observed, and I'll hold off on its name until I've read the people who named it.

My version of the ethic

Levy wrote down six tenets the hackers lived by without ever saying them out loud. Here they are beside mine.

Levy's hacker ethic	Mine
Access to computers, and anything which might teach you something about the way the world works, should be unlimited and total. Always yield to the Hands-On Imperative!	Hands on, and close the loop against the world.
All information should be free.	Free, and checked. I don't say I read what I haven't.
Mistrust Authority. Promote Decentralization.	Mistrust authority, mine first, then the model's.
Hackers should be judged by their hacking, not bogus criteria such as degrees, age, race, or position.	Judge the work by the work, whoever or whatever did it.
You can create art and beauty on a computer.	Yes! Wouldn't change a word.
Computers can change your life for the better.	They can, as long as the people around them still have warm conversations.

The bug pushed the wallet into its pen and Levy gives it two words. Mission accomplished. That's how I want to build. Small and a little ugly, down on the real floor, asking everyone around me, the machines too, and ready for the wallet to be somewhere other than where I thought it was.

Source	Where
Steven Levy, Hackers: Heroes of the Computer Revolution (1984), chapter 2, "The Hacker Ethic"	the six tenets
The same book, chapter 6, "Winners and Losers"	David Silver, the locks, the bug and the wallet, the eighth floor, Gosper on robots and data, Sussman and Minsky, ITS and Eastlake's progress report
Chapter 7, "LIFE"	Silver's later reflections on sheer brainpower and talking like a robot
Chapter 8, "Revolt in 2100"	Ted Nelson, Computer Lib and Dream Machines, Cybercrud
Epilogue, "The Last of the True Hackers"	Richard Stallman
J.C.R. Licklider, "Man-Computer Symbiosis" (1960), IRE Transactions on Human Factors in Electronics	people and machines as partners
My review of the book	the first time through, on this site