Hacker News Reader

Comment by Almaviva | Hacker News Reader

Connecting this with the article about the person who smelted his own sword, I wonder if it's possible for one person to single-handedly build a computer starting from Stone Age technology and raw materials?

View on HN

Sure it is, but you have to define it better. He could build a difference engine much easier than he could build a system-on-chip. The difference engine would require metalworking and precision tools; the SoC would require a working digital computer to design, which requires lasers and other production technologies that in turn require advanced production technologies to make. Even if you just wanted to make a basic 4-bit machine with vacuum tubes, you'd still have to invent all of the preceding requisite technologies just to make the constituent components.

A mechanical device would be much easier, and would be well within the capability of a single person with all of the prerequisite knowledge to make. Just look at the Antikythera Mechanism.

This french HAM just about has it covered. He makes his tubes from nearly scratch.

http://boingboing.net/2008/09/18/video-of-a-guy-who-m.html

That's hardly from scratch, though. Just look at all the different tools he uses. We didn't even have glasswork precise enough to use until the 18th century. Add arc welders, all of the chemicals (and chemistry, mining, etc) needed to make filaments, and everything else required and I doubt a person with full knowledge and access to all of the base materials, as they exist in nature, would be able to build a computer in their lifetime.

> full knowledge and access to all of the base materials, as they exist in nature, would be able to build a computer in their lifetime.

Even that's a huge head start over having to mine the raw materials.

It would probably take several lifetimes to build even the most basic computer from complete scratch even with all the manufacturing knowledge available.

For an electronic computer you also need a source of electricty -- and debugging is going to suck without a voltmeter, oscilloscope, logic analyzer or even LEDs.

I'd go mechanical any day. I recall the Soviets had a computer based on liquid that could compute certain differential equations.

There was a mechanical Turing machine shown at Makers Faire UK. Supposedly it was built only using hand tools and scrap parts. I'd say that's close enough to a hand-built computer.

http://hackaday.com/2011/03/25/mechanical-turing-machine-can...

Related: the Global Village Construction Set by Open Source Ecology (slightly confusing name IMO), a set of 50 machines supposed to bootstrap a small civilization. In other words, this is supposed to be enough to build your own computer, or mostly anything. (I won't say it is really enough, especially since most of those aren't even prototyped yet.)

http://opensourceecology.org/gvcs.php

I'm going to say "no", assuming that when you say "computer", you mean transistor-based.

I am not an expert on the subject, but I suspect making a facility that can manufacture thousands or millions of transistors, and then operating that facility would take obscenely more man-hours than one man has. Even if his goal was only a 4004, or even just a bunch of breadboards full of TO-92's.

Vacuum tube based might be possible for a very simple computer, if automating the glass-blowing and delicate assembly isn't too hard. After all, at least it's macro- rather than micro-!

Or a mechanical computer like the Digi-Comp [1]

Getting through smelting to machine tools would be impressive enough for me. Say stone-age to mid-industrial revolution.

[1] http://en.wikipedia.org/wiki/Digi-Comp_I

Jeri Ellsworth has succeeded in making individual transistors on her own, but she didn't have to purify the materials herself. But it didn't take her a lifetime either.

In "Designing Analog Chips", Hans Camenzind tells the story of designing the 555 timer chip; he did the whole design, without a computer, up to cutting the masks out of Rubylith. It took him several months of nonstop work.

Based on these, I extrapolate that one person could do all the physical steps involved in bootstrapping to a computer from stone-age technology. (I won't say "one man", because that person might have to be Jeri.) And I think they could probably do it in substantially less than a lifetime.

But I don't think it would be possible for one person to do all the intellectual work, or it would have happened long before the advent of agriculture.

And I think it would be a lot more fun to do it with some friends.

The mechanical-Turing-machine link http://hackaday.com/2011/03/25/mechanical-turing-machine-can... posted by chris11 is some kind of evidence here. I've also been thinking about this problem myself.

http://lists.canonical.org/pipermail/kragen-tol/2010-June/00... are some thoughts about possible low-tech digital logic techniques. https://github.com/kragen/calculusvaporis has some thoughts about how to build a reasonably practical non-bit-serial CPU out of only about 1000 NAND gates (or the equivalent.) I think that with the height-field approach described in the kragen-tol post, you could probably make it out of 1000 moving parts — although most of them would have to be machined to fairly tight tolerances. I think the level of mechanical difficulty is similar to a 1960s V6 engine (fewer parts, but much higher force and demands on precision), or a little higher than that of a Curta calculator (600 parts for Type I).

The Clock of the Long Now uses steel gears on stone and ceramic bearings. It contains some mechanical binary calculation (and the prototypes are the most beautiful machines I've ever seen) but it's not a programmable computer.

I remember a "joke" mathematical recreation in Scientific American years ago (decades ago), where they describe a computer found in the jungle, operating with levers and ropes.

Reply on news.ycombinator.com