[chrispenycate]

The "grey goo" argument only holds for self replicating nanotechnology. I hold it is far more probable that the first generations of nanobots will come out of factories, and be replaced as they wear out, rather than being of the "hunt around for materials to build more of myself" philosophy. This is far more practical, both for the level of complexity of the device itself (imagine having to fit a "genetic" instruction with all the details required to recognise, purify and put together correctly another of itself inside something that size, let alone the power supply – much easier to grow hundreds of millions of them on a substrate) and means the entire instruction set can be dedicated to the task they're intended to perform. What is more, every time they wear out you're forced to buy barrels of new ones, which keeps your corporations happy (well. perhaps the "factory" in which they are made could be small and cheap enough for individuals, but raw materials? Much more efficient to buy concentrated and purified than pass kilotonnes of seawater throuh to extract micrograms of platinum)
And while this doesn't eliminate the danger, it does make it directly comparable with chemical spill, and less than nuclear.