It seems to me that we should compensate candidates for their time.
From an employer's perspective, it seems like it might be a good thing to do regardless of whether it's the right thing to do or not
The $500 push would hopefully result in more in-demand candidates taking interviews.
What does HN think? Is it crazy? Is $500 too much, or too low? Is the risk of non-serious candidates too high?
If they don't get the job, gain is zero. If they do get the job, the gain is their salary. EV = P(job) * E[salary|job]. Someone who believes there is a low probability of them getting the job will be drawn to these sorts of interviews, but the sharp people who want the job are not going to be more interested.
If you're trying to optimize for amount of good candidates, maybe. If you're trying to optimize for share of good candidates, absolutely not.
What could be interesting is a simpler test. How about the following?
* On the first of every month, the company publishes a list of random numbers.
* The job interview consists of a written test on how well you've remembered these, and them some code stuff.
Fast, cheap, efficient. Should be a reasonably good selection criteria.