Answer by Holt for Why are questions on the algorithms tag downvoted/closed so aggressively?

Hook! There are problems, and there are algorithms. The problem is that the algorithm tag contains more problems than algorithms.

TL;DR

In my opinion, problems that are NP-hard (are alike) are probably better suited on CS than SO, because there are no "easy" answers. Knapsack problems are the limits, and it is why they can be either well or badly received.

To me, it looks like there are different kinds of questions in the algorithm tag:

1. Questions about (to be) implemented algorithms;
2. Questions about existing algorithms;
3. Questions about problems.

The first category is obviously on-topic on SO, the second one is often "ok" for SO people if the question is well written because these can be answered quite simply.

For me, the question Knapsack with mutually exclusive items and 0/1 knapsack with dependent item weight? are in the third category.

Why was one more rejected than the other? I think there are multiple facts:

• The proportion of people on SO really familiar with such problems is low;
• As one of these, I think the first question is much easier to answer than the second one.

You can answer the first questions by providing an (almost) complete algorithm (as this was done), you (probably) cannot answer the second question without having to refer to external sources. If I were able to answer that second question, I would either:

• Have to assume that OP's is already in the OR field, in order not to have to explain basic topics of OR, but if this was the case, then OP would probably have already found his solution;
• Have to assume that there are no existing algorithm, but there are whole books on the knapsack-problem, if such version has not been tackled yet, it is probably a good fit for a PhD-thesis, so how would I fit that in a SO answer?

While these two questions are interesting (to me), I don't think they fit on SO. I think there are different type of "problems", and some of them are fitted for SO and others not (and should probably be asked on CS).

As a rule of thumb (do not take this as a universal truth, I am just trying to illustrate my words), I would say that if your question can be asked in a programmer interview, it can fit on SO, but if you need to fund a PhD-these to solve your problem, it is probably more suited on CS.

Also, if you want practical answers (which is, I think, the purpose of SO), you should provide practical information, which these questions do not¹ (e.g. the size of the instances, the constraints on the dependency, ...) - Without these, OP is asking for a study about this kind of problem, which is certainly not the purpose of SO. These information are part of the context of the problem, anywhere on SO, questions without context are closed as too broad, the algorithm tag is no exception to this.

_{¹ The second got an answer, so you might say "But these are answerable out of context!", but... the only answer had to made a strong assumption on the instances in order to provide a reasonable algorithm.}