Citing low quality papers (solutions) in surveys?

Question

I sometimes find papers that aim to solve the same problem that I am working on, but these papers are of terrible quality. Some of these papers:

1. introduce wrong solutions,
2. re-introduce a solution that is already published, or
3. are just very hard to understand because of the low quality of writing.

Should I cite these papers in my "related work" section when writing a paper?

Would it be offensive to tackle these papers and prove that they are wrong (especially as the most of these papers are published in not very good conferences)?

Or is it perhaps a waste of time to pay any attention to these papers?

Answer 1

Speaking from the perspective of a no-longer-quite-so-young mathematician, this is a serious question. I think one should acknowledge "prior art", even if one disagrees with many aspects of it. That is, to pretend that something doesn't exist when one is aware of it is dishonest. (I do not think that one's bibliography must only include things one has used... that can be subverted to argue, as I have heard a distinguished mathematician say, that one need not cite anyone else's work _if_one_is_careful_not_to_look_at_it_.)

Yes, there is the awkward issue of giving an opinion on "prior art" that one finds deficient. As @Shion comments to the question, one probably should hesitate before being too sure. The universal non-commital (therefore slightly dismissive, which is the right amount) comment is something like (at the end of introduction) "Compare [A], [B], [C]." Not saying that they're crap, or failures, or anything else. Just admitting one is aware of them, and pointedly not endorsing them... if that's one's intent.

That is, I think that published papers should not just be update-reports, but have sufficient scholarly context-setting to orient a genuinely interested reader not already completely expert, for example. I realize that the literal function of many "published papers" is merely "making a living", but it is not profoundly difficult to do somewhat better.

Answer 2

In general I cite a paper in two cases:

1) I use something introduced there (from the problem statement to some theorem) in my work.

2) I think it would be a good idea if the reader takes a look at that paper when reading mine for some reason (historical, mathematical, whatever).

I don't think I am either able or obliged to put even everything good related to my current work into the literature list: I don't know the whole history myself and nobody will be able to navigate through the whole history anyway. Beyond certain length, the "additional literature" part of the list becomes equivalent to the empty one: if I am suggested to read 3 papers, I'll look at all 3; if 10, I'll look at 1-2, and if 50, I'll look at none, and I suspect that most other people have similar attitudes.

The only case in which I would put any reference to a terrible (in your sense) paper into my work is if I choose to write the sentence "There is a lot of junk written on this subject as well, see, e.g., [][][]" in the main body, but I usually prefer to make enemies in more sophisticated ways :-).

Answer 3

I would:

• Not cite them in your related works section as you clearly do not build on top of that work because of the bad quality.
• If these papers do not have a lot of citations or attention, I would not waste my energy on writing responses to these as they will remain obscure. I would recommend putting your energy into writing papers yourself and reading quality material.