Interesting is a social construction. As such, I would define interesting as worthy of publication in a first or second-tier conference at some point in the history of the field.
> If you cannot demonstrate ability to solve a simple, well trodden problem how will you be able to solve a new one?
0. I've seen many well-respected researchers who have solved truly interesting, new-to-humanity problems stumble on well-known problems others consider easy and/or fundamental. In fact, this is probably true for every single researcher. And doubly so for practitioners. One person's fundamental technique is another's blind spot.
1. Simple != easy to come up with. Often exactly the opposite! Take the TH algorithm as an example. It's a very simple algorithm. Small. Simple. Optimal. Awesome. Not trivial to discover, however.
2. Well-trodden -> unless the algorithm is absolutely fundamental and taught everywhere (e.g. sorting), you've given up on testing problem solving skills. You're just testing whether the person's intro/DS/algo professor covered this particular algorithm (and maybe their memory). Or, more likely, rather their random interview studying happened to hit your pet problem. Since curriculum is not anywhere close to uniform in CS -- even among top schools -- this is just silly.
3. Simple problems don't always have (simple) solutions. Create a fair voting scheme.
It depends on why you are asking the question. Are you literally demanding the candidate spit out the optimal answer immediately? If they can they it means they can remember their algorithms class which is a good signal.
If not, can they solve it with a naive method and then make reasonable attempts to iterate on that, even if they never hit an optimal solution?
It is a little like a high end restaurant asking a trainee chef to make an omelette, which is apparently common in the culinary industry.
> Are you literally demanding the candidate spit out the optimal answer immediately?
I think the author of the article is indicting situations where this is the case. Perhaps he could have been more explicit up front, but reading between the lines this is the situation he's complaining about.
> If not, can they solve it with a naive method and then make reasonable attempts to iterate on that, even if they never hit an optimal solution?
If I were asking the question, absolutely.
> It is a little like a high end restaurant asking a trainee chef to make an omelette, which is apparently common in the culinary industry.
Are there omlettes that are "optimal" (very awesome in some sense)? I honestly don't know (and if so, man, I want one!)
Interesting is a social construction. As such, I would define interesting as worthy of publication in a first or second-tier conference at some point in the history of the field.
> If you cannot demonstrate ability to solve a simple, well trodden problem how will you be able to solve a new one?
0. I've seen many well-respected researchers who have solved truly interesting, new-to-humanity problems stumble on well-known problems others consider easy and/or fundamental. In fact, this is probably true for every single researcher. And doubly so for practitioners. One person's fundamental technique is another's blind spot.
1. Simple != easy to come up with. Often exactly the opposite! Take the TH algorithm as an example. It's a very simple algorithm. Small. Simple. Optimal. Awesome. Not trivial to discover, however.
2. Well-trodden -> unless the algorithm is absolutely fundamental and taught everywhere (e.g. sorting), you've given up on testing problem solving skills. You're just testing whether the person's intro/DS/algo professor covered this particular algorithm (and maybe their memory). Or, more likely, rather their random interview studying happened to hit your pet problem. Since curriculum is not anywhere close to uniform in CS -- even among top schools -- this is just silly.
3. Simple problems don't always have (simple) solutions. Create a fair voting scheme.