[lordnacho]

While I agree with the idea that people respond to incentives, you are making it out to be a lot simpler to design these schemes than is actually the case.

The examples you give are straightforward. You already have a bunch of people who know how to do a job, so you pay them to do it quickly. Basically you are giving them money to go and tell their families they are going to be working late for a while and they have to postpone their holidays. These are both examples of a simple task with a definable, specific goal. Everyone can tell when the junction is built.

With this teaching math thing, there is no finish line. The people who decide if the kids pass are... teachers. Grading your own work is not going to lead to healthy outcomes. You want to adjust for how easy the task is because you don't want easy classes to get paid and difficult classes to be excluded from getting the bonus. But then who defines the baseline? Teachers again. Maybe not the exact same teachers but they are all part of the same system.

Finally there's the problem of feedback. Incentives work when the person who is incentivized knows how things are going and knows how to change the outcome. It is not clear at all that teachers know that if they just show Billy Bob the times tables as a rhyme then he will pass his test. It is not clear at all that teachers even know whether Billy Bob understands the times tables, or is just repeating what is being said.

This is the problem with all incentive engineering schemes. I'm an engineer too and I wish it were simple. But the history of it is rife with all sorts of catastrophes.

[WalterBright]

> Grading your own work is not going to lead to healthy outcomes.

Sigh. Why do people keep bringing this up? Of course you'd need an assessment test that is not under the control of the teacher. Nobody sets up an incentive program where the person being incentivized evaluates himself.

Come on. Give me a reasonable riposte.

[lordnacho]

I think you need to steelman my arguments, per HN guidelines. I didn't say that each teacher would literally mark their own work, the fair interpretation would be that another teacher or committee of teachers would do this.

Here's what I wrote:

> But then who defines the baseline? Teachers again. Maybe not the exact same teachers but they are all part of the same system.

So how do you intend to grade the teacher's work, except by other teachers, who are in the same position?

This is just like having board members appointing CEOs out of the same pool.

In fact, it's a pretty hard problem to deal with in general, and it appears many places in society, so it's fair to ask how this would be dealt with.

If we're going to build an incentive system, we don't want it to be gamed.