rose.blog

The other day I was discussing the concept of downloading brains into silicon with Steve Jurvetson. (I’ve been thinking alot about this recently, which I blame in part on William Gibson).

Part of what got me thinking about this is that I get to spend a considerable amount of time hanging out with biotechnology & pharmaceutical researchers.

The paradigm for health research is to try to use science and technology to fix our bodies when something goes wrong. Here are some interesting facts:

1. The pharmaceutical industry spends a lot of money on what they call research (note that not everybody would agree with calling everything they spend this money on research but let’s ignore that). In 2005, the total spent on research was about $40 billion dollars (see page 15 here). Including the biotech industry ups this number to about $50 billion.

2. Projecting forward twenty years with no growth in spending (which is a very conservative assumption), gives roughly $1 trillion dollars of spending to 2026. I will be 54 in 2026. My eldest son will be 22.

3. The end result of this spending are mostly small molecule drugs, which we can ingest in pill form, that have some kind of effect on our health. There are also a small number of vaccines and medical devices that will also produced.

4. All of these products have as their intended objective one or both of: (1) alleviating symptoms of uncomfortable conditions (allergies, colds, flus, erectile dysfunctions, etc…) (2) curing specific disease conditions (cancers, viral diseases, AIDS, etc…).

OK so does anyone else see the basic fundamental flaw here… let’s say we manage to cure 50% of all of the life threatening ailments we have names for in the 20 years to 2026. That’s pretty good, but unfortunately one of the other 50% are still gonna getcha.

I think the whole idea of trying to extend life by fixing the things that go wrong with our bodies is totally quixotic. Let’s say I get skin cancer in 2026. By then there’s a marvelous cure, so it goes away, but eventually something is going to go wrong that we don’t have a cure for, and then it’s worm food time.

I guess I see the current approach to health kind of like tring to build a car that can drive to the moon. We keep building faster and faster cars and we call it progress, but it’s not really progress because the whole idea is doomed to failure from the outset.

OK so are there any other alternative ways we could go about the whole life extension / health business?

One way to try is to develop methods for “copying” all of the circuitry we use to interact with the world around us into a more robust substrate. Even though our brains & bodies are amazing marvels they have some obvious flaws (they have 100% failure if you wait long enough). If we could figure out how to copy all of these marvels into something that is equivalent in its information storage and processing capacity but didn’t have that annoying property of certain death, lots of people would want access to the process.

I think it’s naive to assume that we can’t figure out a way to do this eventually, which brings me around to the point of this post. So let’s say we do get there at some point. How do we get from there to here?

Obvious Fact #2 (see previous posts for #1): Doing research on “downloading” biological brains into other substrates (silicon probably, maybe with a bit of niobium ) won’t be begun by using human subjects. When this future technology is under development, it’s absolutely clear that the first “test subjects” will be things like nematodes: really simple nervous systems where we can look very carefully at everything that’s going on in the “transferrence”.

Now here’s the thing: Silicon computing machines have MUCH faster operating time scales than biological computing machines. Like a factor of a million or so. So every second to a bio brain is like (all other things being equal) a million seconds to a silico-brain. A million seconds is about 12 days. So every second that passes for the bio-brain is like 12 days for the silico-brain. A month for the bio-brain is like 83 millenia for the silico-brain!

Not only does the time scale change, but the silico-brain can be connected to other systems that can do stuff to it in ways that probably can’t be done to the bio-brain (think Gibson’s microsofts… learn Spanish by plugging it into a socket in your brain).

Back to the nematode: so you download the nematode’s brain into silicon. Can you teach it Spanish? Probably not, because it likely doesn’t have the necessary hardware to store and process information of the quantity and type required. So the “hardware” has some limit - at some point, we can make it do whatever it can do, but this might not be all that interesting.

After the nematode we could try some other downloads. At some point in this evolution we get to an experimental subject whose nervous system is close to ours, say a rat. Rats are used in early stage drug testing because they’re ugly and pharma people think of them as mini-humans (obscure reference warning: Dreams in the Witch-House).

So we download a rat’s brain. Now this brain might have the capacity to design a slightly better version of itself, given enough time. If it can, then that slightly improved version might be able to make at even better version, and so on, etc… this feedback loop could create an “entity” with a silico-brain whose powers of comprehension dwarf the original brain, and of course dwarf human bio-brains.

If this were to occur, then we would have (maybe inadvertently) created a new type of being, evolved from a mammalian brain at an enormously accelerated rate (being in silicon and all), that is not at all human, with massive thinking power, which would eventually find its way into cyberspace. A year after being downloaded is like a million years to the silico-rat; that’s a long time to evolve into something else.

Not sure exactly what this would mean, but we might start seeing What would the cosmically intelligent post-singularity rat god do? bumper stickers.