If the type of scenario outlined in Michael Crichton's Prey is correct, nanotechnology may indeed destroy science fiction. Crichton's novel (including a foreword by the author purportedly discussing thedangers of new technology) is an attempt -- moderately successful, given the credulity of the general population -- to scare the shits out of its readers with a tale of invisible submicroscopic entities that take over humans, leading in many cases to the customary gruesome death that makes for great box office.

Clearly, Crichton believes that nanotechnology could have devastating consequences because of the potential of nanomachines to alter the structure of matter at the molecular level. And worse, if at some point those unhinged power-crazed scientists decide that the best way to make more effective nanobots is to use self-reproducing and evolving nanotechnology, then it's only a matter of time before the principle of natural selection kicks in and the bots best able to reproduce themselves out of whatever raw material is at hand dominate. Dominate, indeed, to the extent of wiping out whatever other forms of "life" are around, and supposedly ending in the grey goo state when the Earth's surface is composed mostly or entirely of nanobots.

If and when this happens, any science fiction books, authors and readers will have been converted into a swarming mass of nanomachines, thus nanotech will have destroyed science fiction.

However, as usual Crichton's egregious scaremongering has a very tenuous basis in fact or probability. There are strong reasons why a ruinous takeover by nanobots is impossible.

First, conservation of energy and the Second Law of Thermodynamics. The nanobots have to get free energy from somewhere, and need to get rid of their excess entropy (and waste heat) since in performing their molecular rearrangements they will generate huge amounts of entropy. If one wishes to keep them at bay, the simplest solution is to leave them in a sealed room with the lights turned out. Now of course good enough self-reproducing bots will find a way out sooner or later, but they won't have the energy to fly, jump or crawl anywhere. The problem of energy supply and waste heat disposal is a major one for any substantial application of nanotechnology, as evidenced by the problems of cooling the smallest and fastest CPU's.

Now of course the nanobot (or cluster thereof) with a good survivor manual will drift on the wind until it encounters a source of energy. This will be a source of chemical energy, i.e. a configuration of atoms not in its lowest energy state. Like a bowl of cornflakes or the guy eating them. The next problem is that it needs to start digesting the energy source, since burning is a no-no for its delicate internal constitution. So it needs a reasonable complement of enzymes or other catalysts on board, otherwise (given the known laws of chemistry) it's not going to get anywhere; or, if it's really unfeasibly lucky, it could get by with a way to pervert the metabolism of its victim to its own benefit. But note that this requires an exact adaptation to the particular food source involved.

The requirement of free energy also gives us a sure-fire way of halting the bots in their tracks: use a load of stuff which is already in its lowest energy state. Like sodium chloride or nitrogen gas or silica.

However, it's still going to be difficult to defend against things that are molecule-sized and can't be filtered out: the bots can just survive, if they're robust enough, until they do meet a source of food. Next we meet the problem of raw materials. If the bots are to reproduce they need to find some of the same stuff that they are made of. So this rules out gallium arsenide-based scenarios immediately, the gallium market being what it is. Moreover the energy cost of turning the raw material into the nanobot shouldn't be too large. This rules out silicon-based structures since the source of Si, rock or sand, is very thermodynamically stable. More or less the only bet for a really dangerous bot is organic compounds, since these are around in large quantities and can be rearranged and structured with relatively small expenditure of energy.

However, this brings us to the crunch point: Nanotechnology is biotechnology. We already have nanotechnology of the type I've been describing. It's called viruses and bacteria, and the air, earth and water are saturated with them, metabolizing the heck out of any free energy that's around. Any supposedly-dangerous nanobots released by feckless scientists drunk on their own god-like powers of world domination would have to compete with the fantastically efficient and well-adapted microbiology of the world outside the lab. However much training and evolution they undergo inside, on the outside they would face adversaries with millions of years more experience. The chance that a few scientists would be able to spot and exploit a weakness in the biome, a niche for nanobots, is zero.

So I seem to be arguing that escaped nanotechnology can be no more dangerous than escaped biotechnology. But this doesn't seem particularly comforting: the deadly viruses that have come around in recent years aren't very nice to meet. Three points give us hope: scientists working on biotech already have rigorous codes of conduct and safety procedures designed to prevent any release of potentially dangerous material. Also, the vast majority of mutations of already-existing viruses create less effective viruses: if "something goes horribly wrong" with an experiment, it's overwhelmingly likely that whatever it is will just fizzle out and die. Lastly, the number of mutations that occur in nature will always hugely outnumber the number that can occur in the lab, simply because labs cover such a minute fraction of the Earth. All the dangerous virii so far have come from nature, since that's what nature is -- a vast outdoor laboratory dedicated to developing new and more successful self-replicators.

To conclude: Science fiction will only be destroyed by nanotechnology if its authors can't find a way to write better than Michael Crichton. On the other hand, don't cut funding for the Centers for Disease Control and Prevention...