As with DNA sequencing and semiconductor design, the fundamental
tools for nanotech are physical, but after the fundamental tools
are available, subsequent advances in design efficiency depend
primarily on software.
Drexler, Merkle, et. al. are design the equivalent of transistors
and resistors. After you have enough basic building blocks, your
design can proceed to the next level. Yes, better building blockk
will enable better next-level designs, but the first small set of
building blocks, the set that can be assembled by an assembler
built from that set, will IMO also provide a basis for "Goo".
The problem with defensive nanotech IMO has never been a lack
of potential capability, but rather stems from the lag time
between the introduction of the goo and the devevelopment of
the defense. the goo can replicate and overwhelm the defense
before the defender has time to design an effective counter for
tbe goo. The existing immune system works because it can learn
to build a new kind of defense from standard parts very quickly.
When it cannot learn fast enough, it fails.
In the scenarios we are discussing, the first goo will be
designed either by accident or deliberately, by humans at
human design speeds. It will have to be countered by
human design efforts operating at human design speeds. Design
of a flexible nano-immune system capable of adaptation to a
full range of threats is an entirely different level of
design complexity. Thus the problem is that the time from the
beginning of the attack until the deployment of the defense is
long enough that the defender is already dead, or assimilated,
or whatever.