Re: Reforming Education

Eliezer S. Yudkowsky (
Wed, 06 Oct 1999 20:40:47 -0500

Dan Fabulich wrote:
> 'What is your name?' 'Lee Daniel Crocker.' 'IT DOESN'T MATTER WHAT YOUR
> NAME IS!!!':
> > I don't buy that argument at all: first, nothing prevents teachers from
> > doing any kind of testing or evaluation they want at any time.
> This claim is false. Teachers cannot force students to take tests AT ALL.
> If students have no incentive to take the tests, then students simply
> won't show up on test day, or if they do show up, they will not have
> bothered to study.

That's exactly right. If the students aren't learning anything, there's no reason why they should show up on test day. There's no reason they should be in school. If the students want to pass their Verification, they'll learn by the most efficient means available - whether that's showing up in class, taking tests, home schooling, or internships.

> > The fact that such tests "don't count" should make them more honest
> > evaluations because the student feels no economic pressure to cheat on
> > them.
> No pressure to cheat on those tests, yes. Or study for them. Or show up
> to them. What good are the tests at that point?

Absolutely none, unless they're diagnostic tests to determine future coursework, in which case it's in the student's interest to take them.

> > The credentialing authority's reputation and the value of their service
> > will depend on how useful their evaluations are, so they have incentive
> > to do a good job of testing real understanding of the subject. As for
> > whether or not a student has "worked" during the year, well "who cares"
> > is precisely the correct response here. Fuck effort, results matter.
> I'm fine with saying "fuck effort." I'm not saying that effort is somehow
> an integral part of learning. The problem here is with cramming. A test
> can only show that you know the material on the test day; it has no
> ability to determine whether you know the material a month later. When
> students cram, they take large quantities of information and put it into
> short term memory. Students who cram still pass their tests, and often do
> quite well, but shortly thereafter they forget the material. On the other
> hand, if you've been studying/using/recalling the material over a long
> period of time, then you'll be more likely to remember it for a long time
> to come. Moreover, repeated testing is more likely to even out
> statistical flukes, in both directions, thereby making the test an even
> MORE accurate measure of the student's abilities.

If that's the case, then Institutes of Verification that require re-passing a written exam or one-day hands-on demonstration once a year will be more in demand by employers. And again, how do you "cram" for a hands-on test?

> Teacher's evaluations IN ADDITION TO third-party standardized testing is
> SUCH a good idea, that this very system is EXACTLY what we have in place
> NOW in high schools all across the country with the SATs, SAT II
> achievements and Advance Placement testing, all run by a (generally) well
> respected private organization, the College Board. Colleges then look at
> both the student's grades, being given out by teachers, as well as the
> results from the standardized tests administered by the College Board.

I scored well enough on the SAT at age eleven; should I have been admitted to Harvard? Multiple-choice tests just don't provide enough information, whether they come once a week or once a decade. They're too easily defeatable, either by cramming, or by knowing the secret mystical patterns formed by those little dots (of which it is said that seeing the complete and true pattern would drive one mad).

           Eliezer S. Yudkowsky
Running on BeOS           Typing in Dvorak          Programming with Patterns
Voting for Libertarians   Heading for Singularity   There Is A Better Way