back
9 comments
I read Soul of a New Machine in college, and then wound up working for a company that actually used the machine that the book was about (the Eclipse MV/8000) for cross-development of video games. This was 1982.

I was used to a Unix-and-Emacs-on-a-VT100 environment, and while the MV/8000 didn't run Unix or have a port of Emacs, its OS and DG's standard screen editor were surprisingly pleasant to use, and the shell was pretty nice. It even had a C compiler (a big deal for the time). My only issue was that the department I was in used the machine for regular office work as well as engineering, and it was heavily loaded during the day with a bunch of terrible "office" software. My assembles took 45 minutes turnaround during the day, but only 5 minutes at night, and guess who didn't see the sun for their first six months at Atari?

If you liked this book, I also recommend:

Lundstrom's A Few Good Men from Univac (essentially a biography of Seymour Cray)

Murray's The Supermen (well . . . ditto :-) )

Post author here. Here’s what prompted me to write it:

Tracy Kidder’s book says that a NAND gate cost eight cents in 1981. The latest iPhone has 11.8 billion transistors. So the chip at the heart of each phone is $1.4 billion in parts, no margin. That’s 1981 prices, 2021 money accounting for inflation.

Sounds right, a 7400 TTL chip held 4 of them and probably cost about $0.32.
I discovered Kidder's book while I was digging into what microcode actually was. I had been thinking about what a huge impedance mismatch it was to write in C targeting an abstract machine that resembles of a PDP-11 and expecting the CPU to translate the instructions for that abstract machine into an architecture is that drastically different.

Microcode is the magic, it's the compiler internals that we don't have access to. I really wish we could get a modern CPU with a writable microcode store so research could happen in the open.

> if it served to entrench and not upend the existing class system, was the computer revolution a revolution at all?

No, engineers don't care about such things as class, and simply do as they are told by their superiors as long as the task is technically interesting. Therefore, there was no revolution in this sense.

> And when you poke binary into the 6502 and program it to add 2 and 3, and execute that operation and, having ascended that ladder with your own hands, see in your mind’s eye the shift registers rippling and the gates flipping and the electron in every transistor collecting and flowing…

> A spiritual experience, and a healthy dose of cognitive vertigo.

I remember having that experience sitting in a processor design class. It was amazing.

Interesting thing about the switch as a computing device is that it has no intrinsic size limits (other than physics itself).

The vast majority of machines cannot fulfill their function below a certain size. A car or a stove can't be miniaturized but "pure" switches can.

- - - -

> And that’s another observation that could only have been made closer to the start than today, with the perspective to see the before and after: if it served to entrench and not upend the existing class system, was the computer revolution a revolution at all?

Well, which existing class system, eh? Ask the East coast elites what they think of the new upstart West coast elites. Silicon Valley is young, the people who won big out here are the challenge to the old class system. It might feel like FAANG are "the establishment" but the fight is still raging.

As far as liberating the lower classes to upend the existing class system, yeah, that idea sure turned out to be incredibly naive. :(

Elites still be elites. West Coast elites pick from Stanford instead of Yale. Desk jockeys win out against hand and body workers. It’s the same, just spread westward.

What new things have been enabled that might affect class? I’m not sure that “influencers” obtain a higher class level. They’re an interesting beneficiary of the technology shifts, but it’s tech more as publishing platform than tech as tech.

New commerce platforms allow a different patterns of buyers and sellers, but we’ve always had a variety of merchants.

The ability to own and apply statistical models as produced by ML seems like elite. Are the people who build and own these drawn from lower-than-elite classes? I think the VCs and other investors that prosper from them are the definition of elite class.

What a beautiful website.