Please allow me to cast a bit of skepticism on this topic. I have an MA in Math Education, and am currently working on a PhD in Education, so I have had a lot of first-hand experience with "how the sausage is made" with regards to education research.

Teaching fads are all coming from research-oriented schools of education like those that I am getting my degree from. While the new methods that are being introduced are "research based", I find much of the research being conducted as highly questionable. Since around the 1970s there has been a strong shift away from quantitative research towards qualitative research in education, which IMHO allows researchers to pretty much "prove" anything that they want. I think that when most people hear the phrase "better teaching methods" they tend to assume that this means "better ways of getting students to learn the target curriculum", but in my experience actual mastery of, say, math skills now takes a back seat to, for example, higher self esteem or even just hating math less.

The result is that over the past several decades students in the US, at least, where such trends towards new teaching methods seem to be the strongest, are performing worse and worse on international comparisons of math and science skills (see, for example, the TIMMS: http://nces.ed.gov/timss/). Meanwhile the countries that now have the best performing students in terms of mathematics ability are largely from Asian countries that predominantly use methods now considered passe here.

The elephant in the room is that the biggest reasons for poor math skills are not related to teaching methods, but rather are cultural issues---starting especially in the middle school years, most students just don't want to be seen as being good at math, as that can be a social stigma for many students of that age (I'm sure that many readers of this site can relate...).

A student who wants to learn will do so, no matter how bad the teacher or the teacher's methods. A student who doesn't want to learn will not, no matter how good the teacher or the methods.

How much research is done on "Drill (and kill)" methods versus the more "modern" alternatives? Probably none. The dogma is that "drill and kill" is really bad, so no one actually does research on whether it works. It seems to me (not in the math education research field, but not that far away) that the "research" being done is often bedevilled and ham-strung by these preconceived dogma.

(Primarily I'm agreeing with you)

I raise a cautious note about one of your points though:

  > A student who wants to learn will do so, no matter
  > how bad the teacher or the teacher's methods

I used to believe this, and I want to believe it, but I've now seen several instances of otherwise bright children with bad teachers making no progress. Then a brief intervention by a gifted tutor has led to the child catching up and surpassing the other in their class.

You might claim that this is a gifted tutor by-passing the cultural issues by providing enough interest to make the child engaged despite the peer pressure. Perhaps.

But I now do believe that a poor to middling teacher in math will actively prevent many otherwise perfectly capable students from achieving their potential. That's why I'm heavily involved in math enrichment programs.