Is a bad approach for a problem a good reason to close a question?

Question

Related meta question, regarding the inverse of this question: Is solely suggesting another solution a bad answer?

See Accelerate OpenGL 2D on Python3 and Finding Primes in Java.

These questions have one thing in common: They are about performance. They worded their question in a manner as "it reaches speed X, but I'd wish it would reach speed Y".

For the first linked question, there are (at the moment of writing) 3 close votes. Maybe 4, depends on how my "leave-open" vote works. Still.

We require that code works before we will review it. This is a good thing, because broken code takes many forms, and most of them inhibit review.

However, when it comes to performance issues, the code works. It's just not fast enough. Often, however, performance optimizations don't come from micro-optimizations - they come from using a different algorithm. I don't have statistics to back me up, but any answer that comes with a different (faster for expected \$n\$) time complexity sounds like a good answer.

That said...

My answer in the second linked question completely trivializes the asker's code. If the asker presents an implementation of algorithm A, and the best answer is "use algorithm B", then surely anything else after that is just a waste? Explaining to someone how algorithm A might be implemented better is a waste of time as they should use algorithm B instead.

For the first linked question: Should questions asking for a better algorithm be closed?
For the second linked question: Why does reviewing an inferior algorithm still have purpose?

Answer 1

Personally, I think the answer to this question "Is a bad approach for a problem a good reason to close that question?" is no, because it is only a bad approach because a better approach exists. And knowing that a better approach exists is no trivial matter.

Additionally, if X people sort using algorithm A in a certain year, and 1 year later algorithm B is invented by some clever person, would you close all the algorithm A questions on grounds of "use algorithm B"? That sounds silly.

Then there's something that feels wrong about closing a question whilst using an answer as the close reason. That is, if you were to copy the close reason and paste it into the answer field, the user would accept it. That just feels wrong.

My answer only covers half of the points I made, however. If the answer containing the better algorithm is posted quickly, what good would other answers be?

Other good answers would critique the code in question. Most likely with a disclaimer of "this answer over here is correct. Algorithm Y is faster, but here are some improvements that can be made in your existing code/algorithm." – ckuhn203 in the comments

Answer 2

It is my opinion that performance issues are bugs. A piece of software is not only subject to functional requirements (“program must find primes”), but also to non-functional requirements (“program must complete within x seconds”). A program with performance issues does not work as required.; the code is broken.

Code Review is not about solving problems, but about giving code reviews. Questions about algorithms may be suitable for Programmers, while optimizing may be on topic on Stack Overflow (Especially questions already containing an attempted solution make for great SO questions).

Should questions asking for a better algorithm be closed?

Usually yes, and especially when OP is not asking for feedback on the current code, but for a reimplementation: Questions about code not yet written are off topic.

Regarding the example: “Finding Primes in Java”: The question is borderline off-topic according to this. The accepted answer does answer the question, but is certainly not a code review. Such answers are a good indication that the question should have been closed.

Why does reviewing an inferior algorithm still have purpose?

If OP is not aware that a suboptimal algorithm was used, that can be a valid part of a review. However, the given piece of code will usually show many other things where you can give feedback on, such as formatting and style, general architecture, unexpected gotchas, ….

A code review is only superficially about perfecting a given snippet of code. In the long run, it's about enabling the programmer to write better code the next time.