I don't know... I feel that a final examination is no test of knowledge. It's a test of memory. It's regurgitation. I'd never do that in my class.

I do have one class where doing well on the final project can get a failing student a passing grade though. But the project requires they apply what they've learned.

My algorithms final was entirely problems that we had never seen before that required that we create new algorithms or heavily modify existing ones. The average score was a 55% (this is a professor who makes extremely hard tests and then curves up).

Had he just been asking us to recite implementations of quicksort and whatnot I might agree with you, but I think that your premise is flawed. Not all final exams are what you have described.

Will the faculty be creating the assessments? I'd wager no. In many places businesses and curriculum-design people weigh in and approve curriculum created by faculty, so I could see those same folks creating assessments.

Perhaps I'm overly pessimistic from my experiences, but I'd bet the tests'll be multiple-choice, just like the UW entrance exams are.

Because it's easier for a machine to grade them.

EDIT: the article does say the faculty will be creating the tests. That does give me hope.

I don't understand that, unless your final consists of mutliple chioce quetions. Why can't you just set novel problems that require written answers? Americans are in love with multiple choice tests for some bizarre reason. Most countries in Europe conduct extensive testing through school,b but multiple-choice only makes up a small portion of the whole exam. Students have to answer essay questions and show their work on problems.

I see no reason why this UW initiative can't work the same way. I hope that it can work this way and attract credibility, since it would make it easier for people like myself who find the spiraling cost of college tuition hard to manage.

Presumably, the cost of taking the tests is the cost of having someone thoroughly vet your work, in addition to developmental costs. If it costs, say, $1,000, you'd expect that to mean that people would spend around 10 hours actually verifying your work. And so, even if you have to write 10 papers and develop 10 projects, 10 hours of grading should be plenty to provide a fairly comprehensive assessment (30 minutes per large assignment is much more than most professors put in now. At least, in my department). For $1,000, you could even up the evaluation time to something like 40 hours (at ~$15/hr, plus administrative costs) by using TAs who really understand.

This is why college won't get any cheaper: there will be more value in having a qualified person evaluate not only the test, but all your schoolwork. You can game a test, but it's harder to game a few years of school.

The "bizarre reason" is laziness. You can grade multiple-choice tests with a key. I don't think they really assess anything myself. I think a test like you describe is better but I don't see it in practice at the places I've worked.

Thus I'm not hopeful.

And to be honest, they're gonna get their money from you one way or another. :)