Re: Proteins and processors (WAS: Singularity in 18...)

Eliezer S. Yudkowsky (
Thu, 17 Jun 1999 02:11:27 -0500

Gina Miller wrote:
> It depends on the size of the protein sequence that you want, for small
> peptides like less then 50 amino acid residues it's very efficient to
> synthesize them chemically. You can get gram quanities of peptide after
> purification, on the other hand if you want to make a fifty thousand dalton
> protein that is about 500 amino acid residues then you have to express it in
> a biological system. There are many different vectors to use in biological
> expression and yields vary enormously for different proteins, and you almost
> always have to do a lot of purification. So it depends on how big a gene,
> which protein, how much you need, if you need it over and over again, and
> how much work this particular protein is to purify.
> Gina "Nanogirl" Miller

Actually, my comments about the online STM and the custom-protein labs wasn't intended to evoke an AI trying to work with protein CPUs. I was positing that the current level of technology would be enough for... well, imagine that somebody arrives from eight million years in the future with a backpack full of '99-readable DVDs; what can they build in our time, and how long will it take them? This is essentially the question of "What can a transhuman AI running on the Internet do?"

Anyway, I was positing that the AI/backpacker could just send a DNA sequence to a sequencer, then send the DNA to a protein synthesizer, and out would come... drextech. An assembler, something you could use to build the diamondoid assemblers that your next DVD has, and then the diamondoids could build the massive quantum computers in your third DVD, and so on.

Or alternatively, that the right sequence of actions sent to an STM probe, acting on the right substrate, could build up a surface that would self-assemble into a matrix when an electrical current was applied, and then you could put "doping" atoms in the matrix to get a custom-shaped molecule that would react in interesting mechanical ways to various electrical frequencies. And so on and so forth.

Not being a specialist in nanotechnology, I don't know how plausible this all is. I may be spoiled by being in AI and neurohacking; after all, I *know* that those DVDs could contain the code for a transhuman AI, or the DNA for ultra-nootropics, or a human-transhuman's DNA. But as far as I can tell, our current hardware plus the right information equals fast infrastructure. Anyone have some more-informed speculations?

           Eliezer S. Yudkowsky
Running on BeOS           Typing in Dvorak          Programming with Patterns
Voting for Libertarians   Heading for Singularity   There Is A Better Way