Just to note, while the Stanford experiment was a pretty horrific abuse of scientific power (for which the perpetrator was rewarded with a chair at stanford), its not the reason why all experiments require ethical approval. That honour (?) goes to the Milgram studies on obedience http://en.wikipedia.org/wiki/Milgram_study
Also, that particular anecdote appears not to have a tremendous amount of relevance to the point, but on a completely tangential point, the stanford study was replicated by the BBC
ncpsychology.weebly.com/uploads/4/7/8/4/4784501/reicher__haslam_original_study.pdf
The results are extremely interesting, and somewhat counter to the interpretations put forward by Zimbardo.
The Milgram experiment is the other member of the "top two" that I refer to in the post.
I could be mistaken about the ethical approval being brought forward, but that is what I recall being taught while an undergrad at Stanford. The Milgram experiment occurred years before the SEP, and I seem to recall that the fact Zimbardo was able to get approval for this experiment (including getting the local police to help) led to a huge crackdown at universities.
I doubt there is any objective metric for choosing "the two most famous and influential psychological experiments ever conducted" but I immediately assumed you meant Pavlov's dogs as the other.
When I was an indie game developer (distinct from a start-up in that I wasn't externally funded, though at the time we were quite profitable), I gave programming tests that I could have passed with flying colors.
If I remember correctly, I only had two people every really pass my test. One I hired and he worked with me for years and went on to be the VP of Engineering at a Fortune 500 company. One wanted $80/hour as a contractor and I couldn't afford to hire him.
I did hire someone who had a positively mediocre performance (keep in mind most people failed outright, many answering NO questions correctly) on the test once, because I thought maybe my test was too hard. He washed out after a few weeks.
What's my point? Don't set your bar too low, or you'll regret it. On the other hand, if you feel you need programmers better than you are, you may already be in trouble.
If you need programmers at the same level but with different domain knowledge, that's not a problem, but it's hard to "look up" at a skill higher than you and be sure of what you're seeing. Even if you think you understand what the candidate is saying, it's hard to distinguish between someone smarter than you mapping out a realistic approach and a smooth-talker very effectively blowing smoke out of their nether regions.
I believe puzzle questions, made famous by microsoft, and then copied by google, are the exemplar here. The standard back-story is that it is not about getting the right answer, but explaining your thought process. Unfortunately, the act of explaining totally wrecks any deep or involved thought process, because you are constantly engaging the speech parts of your brain, which upsets your working memory. Your interviewer never had to go through this process while talking out loud, so has no idea what the question is like on the receiving end.
I am a bit suspicious in this regard of algorithm questions for which the interviewer already has an answer. The interviewer may have a solution in hand that they never solved themselves, or solved at their leisure in a different context (e.g. not in front of several people using a marker board with dried up markers).
That said, an interview in isolation is sometimes an attempt to get a sense of how a person reacts under pressure or constraints. It is doubtful that this really matches with the type of pressure that comes with most jobs.
Interesting question, but to me, were I a founder, I would want to hire engineers who were smarter than me anyway, so I would have no problem giving an interview I myself couldn't pass. In fact, even as a regular employee I want to work with people smarter than I am. But it's still worthwhile to do exercises like this specifically to trigger one's empathy.
For a while, I stopped asking questions that I knew the answer to, and instead I asked the candidate questions that I was about to work on. So I didn't know the answer yet. It was the closest I could get to a normal co-worker situation, where we both discuss an ongoing challenge.
The only reason this approach is not scalable is that you need a constant flow of new problems, otherwise you start rehashing the same old questions, and by then you [think] you know all the good answers.
This reasoning is flawed. You'll have no way of knowing if your candidate's answer is actually correct (they may come up with a working solution that's different than your ore-canned answer), or whether they have a flaw.
The whole point of hard problems is to see how people approach problems - and having them identify their own mistakes and correct for them is part of that.
There's no flaw here at all. There are plenty of puzzles I have difficulty solving, but when someone shows me a solution, it's usually pretty obvious why it's correct and how it works.
Using this same principle though we can argue that someone who knows nothing about brain surgery, such as myself, is qualified to hire brain surgeons based on whether they can explain how to do it in a way that I personally think makes sense to me. I might like to commend myself for being able to hire brain surgeons by virtue of my own ignorance, but it seems to me that the flaw here is I don't really know whether what he explained is true, or if it just sounds good.
You can't make too big a gap--I certainly wouldn't recommend non-technical people trying to hire engineers this or any other way--but maybe it's still possible for a reasonably smart engineer to identify an even smarter engineer.
Because verifying an answer is easy, but creating it from scratch is hard. Analogous to this is intuition is that verifying a proof is simpler than coming up with the proof.
Perhaps psychology experts could chime in here: I've always been a little skeptical of the Stanford Prison Experiment's findings. It was such a complex experiment whose results might have been drastically affected by each participant knowing that they are the objects of study, the experimenter playing the "warden", people playing along with the game, etc.
I guess I suspect some of the conclusions may be true, but I've never really bought that the experiment itself provided good evidence for them.
Also, that particular anecdote appears not to have a tremendous amount of relevance to the point, but on a completely tangential point, the stanford study was replicated by the BBC ncpsychology.weebly.com/uploads/4/7/8/4/4784501/reicher__haslam_original_study.pdf
The results are extremely interesting, and somewhat counter to the interpretations put forward by Zimbardo.