Eliezer S. Yudkowsky wrote:
> Now _this_ is something I dealt with last time. If you take a lot of
> lasers, maybe (for nanotech) a solid sphere-surface of hyperefficient
> quantum-well lasers, all pointing inward, you can zap a pellet of
> deuterium ond it will fuse. The last time I suggested it on
> the group, someone else wrote back that I'd disclosed something he was
> going to patent.
It might work for the trigger, if you can make it small enough. That would remove the need for plutonium, but you still need a very large amount of deuterium and/or tritium to get a large explosion.
> As for power, there are all kinds of burnables in the Earth; or if not
> that, you can use geothermal, the energy differential between one mile
> down and ten miles down. As for getting geotherm set up, I'm
> sure that you can build some pretty efficient electrochemical batteries
> and charge 'em up beforehand. If you've got a large enough resource
> base you could give 'em a small supply of antiprotons. (Remember the
> antimatter leak at Fermilab? 10^12 antiprotons - almost bright enough to
> see, I think.)
You can't burn anything 1,000 meters down - no oxygen, right? Antimatter reactors are possible, but would have to be rather large (you need to absorb the gamma radiation to capture the energy, and that takes a lot of matter). Conventional batteries don't hold enough power to dig themselves down to a deep construction site. IMO, your best bet is actually to drop a power line from the surface - you'll need a small pipeline to move material in and out of the worksite anyway. Failing that, build a burrowing robot with a nuclear reactor. Either way, however, you need something much bigger than a pack of nanobots.
> And remember, the best place to find deuterium is in the
> unoccupied and already-pretty-noisy oceans.
True enough - but to make gigaton-range weapons you need *tons* of the stuff, and probably tritium as well. Even then, you need several dozen bombs to do a good job of wrecking the world (or hundreds, if diamondoid buildings are common). Its hard to collect that much without making it obvious what you're up to - either your robots end up hot enough to show up on orbital surveillance, or you end up with a lot of gigantic, heavily-shielded vehicles that the neighbors are going to wonder about.
Now, all of this is not to say that it can't be done. It just can't be done by diffuse, nearly-undetectable swarms of nanobots. You need macro-scale machines for a lot of essential roles, and macro-scale machinery is hard to hide.
Billy Brown, MCSE+I