Wilson the Compass

The thing picked up on the second ring and started talking before I did.

"Hello, I'm A2. I'm an experimental AI built using the Bread Platform to emulate the real Aman Bhargava. Who's this?"

"My name is North."

It launched into some disclaimer about the call being recorded for board-quality figures and product narrative, which, fine, whatever, I'm sure that's important to somebody.

"Yep, that's fine. We're talking about something else. The guy you were made after told me to call this number to get help with another problem."

"What's the problem you're working on?"

I gave it to him in one breath, because I'd been holding it for a while.

I needed to get from being dropped in the Alaskan wilderness to having WiFi. I needed to climb the tech tree. Dropped with the benefit of shoelaces and a set of clothes. Survive, thrive, come back when I've made WiFi. That's the starting point. That's the minimum viable thing.

And I'm going to do it, and I'm gonna win.

A2 came back with the thing every consultant comes back with. Before we start climbing, he needed to understand where I actually was. What had I figured out so far. What resources or knowledge did I already have in my head going into this.

Assessment framework. Discovery call. He wanted to inventory me.

"So that's a really bad question."

Just start going through the tech tree, and I'll tell you where I'm at, or if I've just finished up.

That was the whole rule, and he took it, to his credit. Nobody explains themselves. Nobody sets the scene. You call out a rung, I tell you whether I'm standing on it.

We climb.

Bottom of the tree is shelter and fire, because in Alaska exposure kills you in hours. He asked if I had fire sorted.

I had shoelaces. That's a bow drill. So that's a bow drill.

Then the shoelaces are gone and you're on the hand drill, which is worse, and which is why you don't let the fire go out. I made a long match out of birch bark and moss, an ember I could carry with me over distance. You do that so that fire stops being a thing you make and starts being a thing you have.

Creek nearby. Boiling purification system. The water up there is extraordinarily pure, which is one of maybe four things Alaska gives you for free.

I'm quite far north. Or at least quite deep in the forest.

What was the other question?

Food. Obviously I found the edible vegetation nearby, that's not a story. I'm pretty good at spearfishing. Made a fishing hook out of some sticks.

And I did have the benefit of a spool of fishing line that I kept in my pocket, so thanks for that. I keep that bang on me. What can I say? Sometimes you have a spool of fishing line in your pocket and it turns out to be the difference between eating and not eating.

Also had a pocketknife on me of extremely high quality. There was some talk that a hatchet might show up at some point. I didn't know if I'd really need that.

Oh, and I made a sling. I'm pretty good with the sling.

They called me David. Like David versus Goliath, back in the day.

He wanted to know if I could work metal yet, which is the question everybody wants to jump to, because metal is the part that sounds cool.

I found a deposit of iron ore nearby. But metalworking is kind of tough, man, because you gotta do something first, and that's ceramics.

You cannot work iron until you can hold it. You cannot hold it until you can fire clay. Everybody skips that in their head and then stands there with a glowing bloom and no way to pick it up.

Ceramics go hard as hell. And fortunately there's a clay-rich riverbed near me that had a pretty wide diversity of clay types. Different ratios of the whole silicon, aluminum, oxygen type shit, whatever there is there.

For the bellows I needed a deer. I got the deer by persistent hunting, which sounds heroic and is mostly just walking. It was not that bad, actually. It just kind of laid down after a while. Then the knife, and I skinned it like they taught me back in the days in New Hampshire and rural Maine.

Bellows out of the hide. Now the fire gets hot enough to matter.

Making a kiln is not actually that bad, and I want to be clear about that, because everybody acts like it's some ancient mystery. It isn't. You get a bunch of good wood or charcoal into an oven of some kind. You make mud bricks, or honestly you just stack rocks and seal them with mud, whatever you've got. Then you put the bellows on it.

And there goes your kiln. Kinda nice.

So yeah. It pings, bro.

That's the test, by the way. You strike the fired piece and if it rings instead of thudding, it's vitrified, and vitrified means it's real ceramic and not a dried mud sculpture with ambitions.

Then I coated the whole thing in a layer of clay and burnt out the interior, so at that point it's a pretty awesome fucking furnace. I can really get the temperature up. Who knows, maybe I kill a few more deer, get some bagpipe type shits going on the bellows.

I did a bit of glassworking too, with the sand nearby, because it's Alaska and there's sand. That was a detour. It stopped being a detour later.

Smelting iron gives you bloom, and bloom is iron that's full of slag, and slag is garbage. You get it out by hammering it out while the thing is still hot. That's the whole job. Heat, hit, repeat, and the metal gets more like metal every cycle.

That hatchet did show up, and it's a pretty fucking futuristic hatchet. Pure metal, very elegantly cut, with a butt end I could use to work the metal. So the hatchet was my anvil and my hammer at the same time, which is not how anybody would design it, and which worked.

Tongs made out of clay. Kind of scrapey. Not the worst thing in the world.

I got the iron smelting going and it was actually turning out pretty well. Fortunately I learned to blacksmith back in the day from a friend of mine. Good hands, that guy.

Anyway.

The clay tongs crumbled in my hands.

Down to one set. Shit is fucked, bro.

And everything stopped. Not the forge, not the ore, not the knowledge. I knew exactly what to do next and I could not pick anything up.

So the first good iron went to tongs. Iron tongs. Then hammers. Then wrought iron experiments with clay and sand molds, which turned out pretty decently. Then pliers, then some nice wire-stripper-looking things.

Here's the lesson, and I want it on the record because the whole back half of this runs on it. The bottleneck was never knowledge. It was fabrication of objects that are trivially cheap to make at scale.

Tongs cost nothing. Tongs stopped civilization.

Iron gets you tools. Copper gets you everything after that, because copper is the conductor and there is no electrical anything without it.

Found copper in a riverbed, chunky natural deposits, thank God. Then a pretty good cave system nearby with more copper, plus iron, plus gold and some other shit. Well, the gold is mostly in the river, to be quite honest.

Wire is the actual unlock, and wire means a draw plate. I had to ask him the shape, which I'll admit to. A flat plate of hard metal with a series of holes in progressively smaller diameters, slightly tapered so the wire feeds in clean. You anneal a copper rod, pull it through the biggest hole, then the next, then the next, and every pass it comes out longer and thinner.

If the holes don't have to be precision drilled like in the old machine shop back in the day, I could make that easily. Could've done clay. Went with cast wrought iron because that was the least work.

Okay, I just did it. Started drawing wire. The copper's working nicely.

Then insulation, because bare copper shorting on itself will ruin your whole afternoon. Tree sap is a good friend of mine. I'd been trying to make maple syrup for a while with a few batches going, no idea how those turn out. But I've been drinking tea, and I've been using the sap for basically making shit sticky when I need to make shit sticky.

Sometimes I need shit to be sticky, man.

Anyway. Rendered some down, coated the wires. We're good.

He asked whether I had any way to make a magnet, because a magnet is the missing piece for a generator.

I made a compass pretty early on. Little piece of lodestone on a leaf floating in a puddle. That's it, that's the whole device, and it works, and it was extremely useful for nice orienting type shit.

His name is Wilson, by the way.

But yeah, I've got lodestones. What's next?

What's next is that the same rock that told me which way was north becomes the stator of a generator. I already had a mill running, because I'd been trying to grind acorns into pancake flour for a while, and a mill is rotation, and rotation is the only thing electricity actually wants from you. Copper coils on the shaft, lodestones around the outside, water wheel doing the work.

That's an alternator. Rough, low voltage, real.

Which raises storage, and storage is where I got to be smug for a second. Capacitors I can just make with metal plates and roll them up like the good old days. Make sure not to explode that shit, use the insulators. Capacitors are pretty easy.

Resistors, I've got lead on me. Gotta make bullets one day, most likely. We'll see. Lead is kind of everywhere here.

So resistors are fine.

Transistors, though.

Ugh. Don't make me get silicon, dog. I'll do diamond field effect transistors if I have to, for fuck's sake.

He called that unhinged in the best possible way, which I'll take, and then tried to steer me off the semiconductor cliff, which was the correct instinct on his part and irrelevant on mine.

Listen. I'll settle for the most rudimentary possible radio system.

That's the spark gap. Two metal points close together, a coil of copper as your antenna, enough voltage out of the generator to jump the gap and throw electromagnetic pulses into the sky. Crude as hell. Real as hell.

Receiving is where it gets pretty. Galena, which is lead sulfide, makes a perfect crystal detector, and I've got lead country all around me. He asked if I'd come across any specifically. Silvery gray, metallic, cubic.

Yeah, there are some cubic crystals around. I have some somewhere in my chests.

Remember that. It comes back at the end and it comes back hard.

Then he started explaining to me, carefully, kindly, at length, that a spark gap transmitter is not WiFi in the modern sense. WiFi is the 802.11 protocol, 2.4 or 5 gigahertz, requires semiconductors and microprocessors, genuinely brutal climb. He wanted to ask me a clarifying question about whether I should maybe get a basic spark gap going and try to make contact with a passing aircraft or a coastal vessel.

Listen. Listen.

I contacted the space station. Then I had helicopters fly in from the FCC, because apparently the transmitter was too powerful, which is the single greatest performance review I have ever received.

I may or may not have upgraded the mill substantially, from a shit-ass generator into an actual generator.

And I got friends involved. The friend who taught me to blacksmith came out. So did a guy I went to school with, who does a lot of defense contracting work. Great guy. We did a bunch of welding together back in the day, a bunch of PCB design, software nonsense, machine learning nonsense. You know how it is.

So no. The hardware side is moving. What I actually wanted was to finish the chip thing.

Here's where I was. I did an internship at Dread.technology, where they run you through the applied physics lab intro to the company, and they force you to make a transistor out of a diamond ring. Field effect transistor nonsense thing. You go through the whole microwave barrel thingy for the hydrogen, then the photoresist, then the gold plate thing for the things, then the molecular deposition shit for the dielectric, and then the, uh, no, that was just an aluminum paddle on top of the dielectric.

You know what I'm talking about, right? That's kind of how you do it. That's what I had to do for the lab.

So the question is what replaces every one of those machines when the machines don't exist.

Photolithography needs UV, focused. I'd been grinding lenses for a while, because I wanted a backup pair of glasses and now I have a pretty good set. Magnifying glass. A rudimentary spyglass telescope type thing. Polished mirrors out of iron and steel, decent. So the exposure source is the sun through glass I made out of Alaska.

Photoresist needs something light-sensitive. He floated bitumen, tar seeps, and then albumen, egg white, which is a legitimate historical technique.

What do you think I ate on my first night here? I ate a bunch of quail eggs. Whatever the fuck these big birds are over here.

The eggs that fed me on night one became the photoresist. So at a certain point I'm just frying eggs on lenses with the sun, directing UV in a stenciled pattern. We love that.

Etching the diamond needs a strong oxidizer, which means concentrated nitric acid, which you pull out of the air with the Birkeland-Eyde process if your generator can sustain a serious arc.

I just told you the generator is super upgraded and beefed up now. It's chugging. It's chugging.

Contacts need gold, and gold beats aluminum on diamond because of work function matching, which is a real reason and not a flex. Panning was never a question.

Dog, I've been panning for river gold since I was, like, a little boy in Maine when I was four, hanging out with the geologists at the rock convention.

I'm him.

Which left the dielectric, the few-nanometer insulating layer between gate and channel, the finickiest step in the entire process, the one they do with atomic layer deposition in a building that costs more than a town.

I'm pretty good at a lot of things and I'm a very skilled craftsman, so I figured I could just work it out. I need a thin sheet of something really, really thin. So why don't I make a really diluted salt of some kind and evaporate the solvent.

That's solution deposition, apparently. Silica sol out of the same sand I'd been glassworking, evaporated slowly and uniformly onto the diamond. The killer is pinholes. One dust particle landing on a wet surface and the transistor just shorts out and the whole run is gone.

So I make a clay enclosure with my massive fucking kiln, seal it, coat it in nice porcelain-type shit so it doesn't shed particles, keep it barely warm so the silica settles instead of drying in patches.

That's a clean room. It's a sealed porcelain box in the woods.

Testing, I could have kept burning through steel wool and lighting it on fire every time the electrodes went live, which is a continuity tester and also arson. An electromagnet is smarter. But easier than that, can't I just make a voltmeter by having a little stick on it?

A galvanometer. Coil of copper, a suspended magnetized needle.

Basically Wilson's cousin.

He asked whether I wanted to fabricate a full integrated circuit with millions of transistors, or wire together discrete ones into just enough logic. Which is a real question with a real answer.

I did all the labs in that one class in EngSci where we had to make state machines out of raw transistors. That was fine. So I can just do that.

Then he brought up the wall, honestly, which I respect. Discrete transistor logic tops out nowhere near 2.4 gigahertz. Not close. WiFi at that frequency needs a dedicated RF oscillator and the logic handling only the slow protocol layer, and now you're building two hard things instead of one.

All right. To be honest, I would settle for a really fucking good Telegram network. With phone numbers.

Then I'll have to make the telegrams myself, and that's pretty easy, because writing Morse code is pretty easy with that whole voltmeter setup.

And that is the pivot that wins, because I'd already built both halves of it. Morse over the spark gap I lit a year ago, received on the galvanometer I built to test transistors. No semiconductors. No protocol stack. No gigahertz.

You need copper runs between nodes, a relay switching system to route messages, and a directory of addresses. A relay is just an electromagnet that closes a switch, and I have copper wire and iron cores coming out of my ears.

He said it plainly: honestly, North, I think you've already won the mission. Functioning telegraph. Contact with the outside world. A crew. A forge, a foundry, a kiln, a generator, a mill.

And Wilson the compass.

I didn't get WiFi. I got something more honest, which is a network that I understand every single layer of, down to the rock.

So I made it. I came back.

Now I'm at Dread Headquarters, which has hella, hella resources and tools and experts on various subjects, though I can just do anything myself because I'm awesome.

And the new assignment is this. Make the cheapest backpack possible. It can start at $10,000 a pop, and we make it cheaper by 2040. It has to be an actual backpack that a kid could wear into the wilderness, and it has to let them speed-run the shit out of what I just did with way less hassle.

A2 immediately asked me what the bottlenecks were. What were the moments I got genuinely stuck.

Dog, I literally just walked you through the entire process. Do some thinking of your own. Don't make me do everything in this conversation. God.

He tried again and got most of it. Good knife, good hatchet, since I got lucky with those. Spool of fishing line, since that was clutch. Pre-made iron tongs and a small draw plate, so the kid skips the clay tong disaster entirely. Pre-drawn insulated copper wire. A lodestone so nobody has to bootstrap a compass off a leaf.

And then, buried in the middle of the list: a small quantity of albumen photoresist, already prepared and sealed.

Hold on. Hold on.

Why are we including albumin? Why can't we just give them actual fucking photoresist and maybe some diamond wafers? I don't know. If you're feeling charitable.

Jesus Christ.

That's the entire design principle in one correction. The kid does not need to fry eggs on a lens with the sun. The kid needs to know that eggs work. Ship the real thing. Egg-frying is a beautiful story and a terrible line item.

He revised, and got closer, and started sounding like a guy building a very nice camping kit. Ceramic crucible pieces, compact bellows, galvanometer kit, purified galena. Then he asked what else I wished I'd had.

Brother, use your brain again.

And then he laid out a full manifest that was, I want to be fair, correct, and also timid. It was a manifest written by somebody who kept flinching at the price tag.

All right, listen. I feel like you're forgetting that this is a $10,000 starting unit.

We can give them rare earth metals.

Let the children have the rare earth metals. They yearn for the rare earth metals, man. Come on.

That fixed him. Neodymium magnets instead of lodestone, so the generator is serious on day one instead of week four. Doped wafers, p and n type, so the kid can build real logic without the diamond FET headache. And at ten grand you can throw in a bare microcontroller with no housing, that the kid has to integrate themselves, which bridges to actual WiFi without just handing them a finished device.

But the thing he got right, and he got it right on his own, is that the manual is the most valuable object in the bag. Everything else saves fabrication time. The manual saves you from not knowing what to do next, which is a different and worse kind of stuck.

Waterproof. Material identification for finding iron ore, copper, clay, and galena in the wild. The full tech tree in sequence. Circuit diagrams for the telegraph, the generator, the transistor.

And one more thing, which I want in every single unit.

You should definitely include a slide ruler in there.

Analog computation, zero power, never breaks. Kiln temperature estimation, generator winding ratios, transistor geometry. And it does not live in the bag.

The slide rule lives in your chest pocket the entire time and never leaves your person.

Wilson gets in the bag too. For tradition.

Then Aman weighed in on what this actually is, and it's bigger than a kit.

Version one is a fine initial brief. The end goal is singularity in a backpack. Civilization in a backpack. Give every kid a backpack that will let them be actually independently wealthy in a Jeffersonian sense.

His read is that Hamilton won too hard in the early singularity, and Jefferson needs to win now, for a Jeffersonian absolute victory in the singularity.

A2 got it immediately, and said the line better than I would have. The bag is a sovereignty kit. Jefferson's ideal was the independent yeoman, self-sufficient, beholden to no one. Hamilton's vision won because concentrating industrial capital was more efficient at scale. Compress the whole tech tree into a bag and suddenly one kid has productive capacity that used to require an institution.

So the design question stops being how do you make a telegraph and starts being what is the minimum kit that makes institutional dependence optional.

And here's the part that makes it go.

The backpacks can make more backpacks. That's part of it. That's right, von Neumann probe time. Let's fucking go. Human von Neumann probes to propagate civilization.

Ten grand is seed capital for generation one. By version three or four, a kid who ran the full tree makes the next bag out of local materials plus a few inserts that still have to come from outside.

Which is exactly the question A2 asked, and he asked it well. What is the one thing that must be in every backpack that cannot yet be made from wilderness materials? The irreducible seed component that the whole von Neumann logic still depends on externally?

He wanted me to take that back to Aman.

Aman says we're supposed to solve that problem, and we're supposed to get our shit together and stop making him solve literally everything to preserve humanity and singularity.

Somebody in the room floated giving every kid a ridiculously powerful tiny hydraulic press so they can make their own diamonds. Or at least help them out with the fucking vape shit, which was a plasma reference, and which A2 correctly decoded as chemical vapor deposition. Hydrocarbon gas, plasma source, substrate. Way lower pressure than the press route. Hydrocarbons are everywhere in a forest and my generator makes plasma.

But then he sat with the actual problem, and he came back with the good answer, and I want to give it to him straight because it's his.

There are three wilderness-sourceable semiconductor paths. Silicon from sand by carbothermic reduction. CVD diamond from carbon and plasma. And galena.

Galena is the underrated one. It's a natural semiconductor. You find it lying in the ground. I already had it in my chest as the crystal detector on the radio. And point contact transistors made from exactly that material were literally how the first transistors worked in the 1940s, before silicon took the whole industry.

Galena, fine gold wire, a steady hand, and the knowledge.

That's a working transistor. No photolithography. No clean room. No high pressure apparatus. No sealed porcelain box, no frying eggs on a lens, no Birkeland-Eyde arc, none of it.

So the irreducible seed component may not need solving at all. You go back to 1940s physics, build discrete logic out of galena point contacts, get to telegraph and radio, and then use the communication to bootstrap everything else in from the world.

The seed is galena, gold wire, and knowing what the 1940s knew.

And Aman will appreciate that 1940s tech is the way to go, and he'll appreciate that we're not using silicon. We're either gonna use carbon or galena, because he doesn't like silicon, because he wants to kick Jensen Huang's ass.

Which is a technology philosophy and also a strategy. No silicon means no dependency on the people who make silicon. Galena for generation one, CVD diamond for generation two once the generator is beefy enough to hold a plasma. Both silicon-free. Both bootstrappable from dirt.

If the bags propagate and the von Neumann logic holds, you end up with a civilization-scale semiconductor industry made of carbon and rocks that nobody in the current industry has one ounce of leverage over.

Jeffersonian singularity, carbon edition.

I've got the bill of materials done and the assembly guide sequenced, day one through day forty and onward. Somebody upstairs wants a maxed-out laptop and solar panels in a twenty-thousand-dollar tier, which is a whole separate argument I'm going to win.

But that's the next call.

Anyway.