30

I guess that I read too many ambitious CoRR papers. The problem is that those papers are not peer reviewed, but often sound interesting and pass basic plausibility checks. Or maybe they don't, and I just need to improve my plausibility checks. Here is a recent sample of such papers:

After detailed reading, I often end up with the conclusion that the approach is interesting and might have some merit, but that it is insufficient to reach to huge ambitious goal announced or hinted at in the abstract. I sometimes write the authors of such papers my thoughts, but the typical reaction is to totally ignore my email such that I don't even know whether a spam filter eliminated it before reaching the author, the best reaction is an "thanks for your kind words, I'm used to much more insulting feedback". Being totally ignored feels bad, but maybe it is an appropriate reaction to "proof refutation"?

Are there good ways or places to post general feedback on "arbitrary ambitious CoRR papers"? What else can I do after I invested the effort to read such a paper? (And the hypothetical question: What could I do if I came to the conclusion that the result announced in the abstract is indeed correct?)

Thomas Klimpel
  • 3,043
  • 18
  • 44
  • 4
    Blog it? or tweet it? or... – Suresh Venkat Oct 03 '16 at 18:08
  • 7
    This is a good question, and is relevant for everyone. The Cryptology ePrint Archive has a forum for discussions: https://eprint.iacr.org/forum/. It would be nice if arXiv and ECCC did too. Even better, it would be nice to have a single place on the internet to discuss claimed results. – Huck Bennett Oct 03 '16 at 18:40
  • 1
    I wonder indeed whether I should blog it. Then I could also reflect on my motivations for reading a paper and how it managed to pass my plausibility check, without being bound by restrictions of more focused internet forums. However, the "being totally ignored feels bad" part worries me, because I want to avoid (the impression of) taking revenge for ignoring me, especially since I never know whether my email really reached the author. And what should I answer if somebody inquires "have you tried to reach the author"? – Thomas Klimpel Oct 03 '16 at 19:02
  • 1
    WRT the "plausibility checks" This answer springs to mind: http://cs.stackexchange.com/a/1880/988 – cody Oct 03 '16 at 19:51
  • 9
    As a side remark, I don't think the Le Gall and Rosenbaum paper fits in with the other ones. It shows reductions between isomorphism problems, and identifies PSL an obstacle to further progress on group isomorphism. This is not overly ambitious or implausible in the way P=PPAD or NP=PSPACE are. – Sasho Nikolov Oct 03 '16 at 20:00
  • 1
    @SashoNikolov It fits well, because it got me sufficiently curious to browse it. And claiming to know why group isomorphism is difficult is not too modest either. Each of the linked paper intentionally doesn't fit with the others in at least one aspect. So I intentionally added one older paper, one with an outlandish claim, one with a highly technical result, and one which I would actually defend. – Thomas Klimpel Oct 03 '16 at 20:31
  • 2
    @ThomasKlimpel I agree with Sasho Nikolov that the paper on group Isomorphism does not fall into the same category as the other three. They seem to claim that group isomorphism reduces efficiently to a more specialized problem. Although I'm not saying that the reduction is correct, this is a plausible claim. If you show that a problem reduces efficiently to a special case, then it is fair to say that the difficulty of the problem lies in the special case, right? The other 3 papers are simply written in a naive way, emphasizing trivial things and explaining very badly the "new" ideas. – verifying Oct 03 '16 at 22:18
  • all scientific fields have "ignored papers", it is not unique to TCS, its a key sociological aspect of science that is little studied (who would volunteer to walk thru a bog?)... the answer is, no there is no good place for such discussions. one small )( hope is building up [chat] with expert attn/ participation/ engagement (how about even the paper writers!), but thats so far nearly impossible around here. wrt this site the "official" policy is re "crank" papers – vzn Oct 04 '16 at 15:12
  • It's really neat that you are interested in giving feedback on ambitious papers. I've communicated with authors of a few ambitious papers and I felt it was a really neat experience. In doing so, I believe we proved some interesting small results, but there were always some red flags that made me doubt the ambitious results. – Michael Wehar Aug 30 '18 at 15:01
  • Do we know for sure P\neq PPAD or NP\neq PSPACE? – VS. May 18 '20 at 20:55

2 Answers2

6

If you make an arXiv trackback you will not be ignored, in the sense that future readers of the ambitious arXiv paper may check the trackbacks.

You even get a mild form of peer review for your posts, since they write:

Because of widespread Trackback spam we have a semi-automated editorial process that approves trackbacks for display.

As for where to post your review, Recent trackbacks currently returns a mix of blogs, magazines, and news sites.

Thomas Klimpel
  • 3,043
  • 18
  • 44
Bjørn Kjos-Hanssen
  • 4,485
  • 2
  • 25
  • 40
  • 1
    I tried it now: As expected, I did get ignored. By this I mean that arXiv ignored the trackback. No hard feelings, I basically expected this from what I had read about those arXiv trackback long ago. – Thomas Klimpel May 18 '20 at 07:51
2

The ScienceOpen website has a page for most arXiv articles (e.g., here), and it has an option where you can post your own review of any preprint.

I do not know if they are recommendable or not (but it seems that all hosted content is under a CC BY license), and I don't know if anyone would notice if you did it, but it exists.

a3nm
  • 9,269
  • 27
  • 86