Merit pay for "student achievement" is a bad idea. Not because I'm some kind of communist, but because I'm one of the (it seems) relatively few people that actually think about what student achievement in science education looks like.
Knowing random facts about stuff is only a side-effect of actually being scientifically literate. The idea of scientific literacy includes knowing the core concepts of science, being able to construct (and deconstruct) scientific arguments, being able to use the tools of science, and being able to participate in the broader scientific community. Knowing what percentage of the earth's surface is covered in water isn't really part of that puzzle AT ALL. (I'm a relatively "literate" person, with a BS, MS, and PhD, although not in earth science, and I thought it was closer to 60%).
Science education is of supreme importance to the future. But if you're really serious about improving science education, you have to think HARD about what you mean by that. You mean: making sure people have all the pieces of scientific literacy, not just making students memorize facts.
Once you accept that point (and clearly, slashdot comment threads are not the places for real debate, but try reading How People Learn (bransford and brown) and Taking Science to School (big committee, but published by the NAP) for more insight there).... where was I? Oh yes, if you accept that scientific literacy is more than just knowing facts, you have to take a critical look at the standardized tests you're using to base teacher merit pay on. They don't actually test scientific literacy. They test fact retention for the most part, and scientific process skills to a lesser extent. But process skills in these things are tested in a content-free way that completely lacks any kind of face validity as to its relationship to actual scientific inquiry practices.
So, think about it: we're going to base teacher merit pay on student performance. Fine. But if you want to do that RIGHT, you have to actually measure the kind of performance you want, rather than settle for the kind of performance that's easily testable on a large scale. That turns out to be a nigh-intractable problem, and it's this intellectual cutting of corners (testing what you can test, and valuing that, instead of valuing what's central to each discipline and accepting that testing for performance in that fashion is going to be expensive and a real challenge) that's led to the travesty of NCLB - nationwide failure of a system that's supposed to help our most fragile natural resource.
Anyway. The biggest problem with all of this is that thinking hard about education is a real challenge. Teachers have a very important set of critical skills that most science folks don't understand (since most science folks tend to think that science should be just as easy to everybody else as it was to them). Sure, there are plenty of bad teachers. But basing merit pay on test performance will do very little to improve education if the tests are deeply flawed.