RE: Challenge of Design Complexity

Billy Brown (
Tue, 15 Dec 1998 11:18:05 -0600

<continued from previous post>

Let us suppose for a moment that we accept the above arguments as substantially accurate (not that I think anyone will :-) ). What could we then deduce about the next century?

Well, for starters we can say that we aren't going to go from the first assembler to mature nanotech overnight. Nanotech simply makes physical construction more like software design - you still have to design something before you can build it, and the design complexity will choke even 2020-era computers (not to mention human engineers) long before you get to utility fog.

Fortunately, that means that a gray goo scenario is unlikely. Actually making a nanodevice capable of out-competing everything in Earth's biosphere would be a daunting project during the early nanotech era. It is much more likely that weapons research will create limited versions, and defenses against them, in a continuation of the same patterns we are familiar with.

As computers get faster, and software design methods improve, our ability to build complex nanotechnological devices will advance rapidly. If we are still relying on humans at that point we will probably see something that looks like the computer revolution, but applied to all material goods - every year everything is smaller, faster, better and cheaper. If we have AI or human intelligence enhancement at that point the rate of change will be much faster.

Of course, even moderately mature nanotechnology will soon lead to intelligence enhancement, which leads to better nanotech and more intelligence enhancement, and so on. We still get a Singularity, but there are constraints on what kind we can have:

  1. It is unlikely to be driven by any single technology. You need broad advances in many different fields to sustain the process.
  2. It is unlikely to be dominated by any small group, let alone an individual. There is no single breakthrough that will give you the whole enchilada, so you get a faster version of the tech advance we are familiar with.
  3. It is unlikely to be precipitated by a seed AI. Unless it turns out to be very easy to become an SI, it will not be able to invent everything it needs by itself.
  4. It is likely that a substantial portion of society will have the opportunity to become Transhuman during the early nanotech phase.
  5. By the time we have moderately advanced nanotech, we will probably also have lots of people with Transhuman IQ and assorted other enhancements. By the time society has mature nanotechnology, it will be dominated by proto-SIs and other nigh-incomprehensible beings.

OK, here I am, waaaaay out on this limb. Anyone have a saw?

Billy Brown, MCSE+I