>From: "Eliezer S. Yudkowsky" <firstname.lastname@example.org>
>Subject: Re: Reforming Education
>Date: Mon, 11 Oct 1999 13:09:46 -0500
>John Clark wrote:
> > I'm almost embarrassed to point out the obvious but the only way to tell
> > teacher from a bad one is to see how well they teach. Have all students
> > a given grade take a standardized test, this would really be a test of
> > the teachers not the students. Teachers with students in the top 20%
> > would get a raise, those in the top 5% would get a big raise, those in
> > top 1% would get rich, those in the bottom 20% would get a pay cut ,
> > those in the bottom 5% would get a big pay cut and those in the
> > bottom 1% would get fired.
> > I'm sure the National Education Association would really hate this
> > policy and that's pretty good evidence it's a good idea
>I hate this policy. It's improperly controlled for good and bad
>students, whether you're teaching in the suburbs or some inner-city
>hell, and so on.
>A better way to do it, as Crocker and I keep saying, would be to break
>up the system into modular components that could be more highly
>specialized and could integrate new methods in particular areas. Once
>the credentials are given by specialized credentialing organizations
>that compete for credibility with employers, then the teachers and
>schools can be credentialed by their own independent systems which would
>compete for credibility with parents.
They both can work. I don't believe in bad students. Of course there are students who misbehave (kick them out or find a new way to teach them) and students who don't learn very well (personalized learning programs in personalized-specialized-schools).
2 class rooms each containing 1 teacher 10 good students 5 bad students and 3 retarded students.
If students in teacher one's class perform better than students in teachers two's class then teacher two is doing something wrong and if teacher two wants to get that raise or kip its job then teacher two will make an effort to learn teacher one's methods.