Via Andrew Sullivan, I see that Kevin Drum is banging the drum for universal pre-K, and good for him. I'm also glad to see that Drum is acknowledging that the earlier we intervene, the better. The metrics for students who start out disadvantaged are quite discouraging; typically, those who are behind grade level early stay behind throughout organized education. What Drum doesn't say, but what is glaringly obvious to me, is that parenting is the great unspoken factor in education reform. But it's a classic wag the dog situation; we lack a framework to intervene in parenting in all but the most extreme situations, so we don't talk about it in our policy debates.
Drum also calls for a lot more empirical experimentation in his post. And there, too, I'm with him in the abstract. I should point out, though, that there is already lots of educational experimentation going on. Like, lots and lots. If Drum is frustrated by the pace of change, I would point in the direction of what many tell me is the single biggest obstacle: ethical concerns. The simple dynamic is that, while standards vary depending on individual institutions, institutional review boards, and researchers, the ethical interdiction against research that might set back students relative to their peers makes a lot of controlled experimentation impossible. If you know, thanks to prior research, that certain conditions are correlated with poor educational performance, it's unethical to expose students to those conditions in a controlled experiment, even if you believe that the benefits of the research will be great in the long run. What's more, if you find in the commission of your research that one of your groups is significantly outperforming the other, many believe (including some professional groups that establish research protocols) that you have to change or end your experiment, in order to prevent leaving the underperforming group behind. And the younger the children involved, the more stringent these guidelines tend to be.
To take an extreme example, there's great debate about the degree to which a lack of exposure to a linguistically rich environment permanently disables a child's language ability. Obviously, you can't isolate a child from language in order to ascertain the degree of this sort of disadvantage; you'd be permanently disabling a living human life (or lives) in doing so. Researchers are therefore forced to rely on disparate, frequently obscure natural experiments, such as with so-called feral children. Similarly, there's a belief among some Chomskyian syntacticians that children who are not exposed to a fluently delivered or functional grammar (such as in efforts to reestablish a dead language) will spontaneously generate a grammar in order to structure the lexical information they are exposed to. Again, you can't perform an experiment to tell if this is true; you'd risk long-term harm to children.
As I said, those are extreme examples, but when I talk to my friends who do educational research on children, it's a constant frustration. They accept and support the ethical principles involved, but they feel that progress is also seriously retarded due to those principles. It's something to keep in mind when examining these issues.