>In the case of testing it’s because you choose to focus on the obviously bad thing (current state of testing) rather than the very complex and difficult question behind it: HOW do you measure knowledge? And when you decide how, how do you scale it?
This sounds like an entirely different question. When you have a method for testing, you have at least two different measures of effectiveness:
- how well the test measures knowledge when it is taken honestly
- how likely is the test to be taken honestly vs. subverted
I thought this thread was about the second question, but you seem to be focused on the first. But these problems require different kinds of solutions, and crucially, it is much easier (but still not easy) to verify success or failure in addressing the second question (cheating) than the first (predictivity).
This sounds like an entirely different question. When you have a method for testing, you have at least two different measures of effectiveness:
- how well the test measures knowledge when it is taken honestly
- how likely is the test to be taken honestly vs. subverted
I thought this thread was about the second question, but you seem to be focused on the first. But these problems require different kinds of solutions, and crucially, it is much easier (but still not easy) to verify success or failure in addressing the second question (cheating) than the first (predictivity).