37

What should you do when you see a question raised in public, say here on stack-exchange, that you know the answer to, because you are looking into as part of current research project?

For example, I see a TCS.SX question that I know the answer to, because I worked on the problem recently. I haven't finished writing up the results yet and am trying to get a few more results to make an acceptable paper.

Should I contact the asker personally? Claim credit publicly by publishing answer on the site? What should I do?

András Salamon
  • 19,000
  • 3
  • 64
  • 150
asterix
  • 331
  • 2
  • 4
  • 12
    This should be in meta.cstheory.stackexchange.com – didest Oct 20 '10 at 04:32
  • 14
    why should this be on meta ? this is not about the running of the site really. It's a genuine research process question. – Suresh Venkat Oct 20 '10 at 05:18
  • http://meta.cstheory.stackexchange.com/questions/214/should-there-be-a-protocol-to-notify-authors-if-we-find-an-error-in-a-paper seems mildly related, but is on meta. – RJK Oct 20 '10 at 22:06
  • 1
    Voted to migrate to Meta. From the way it is stated, the question is about how to behave on this website. – Tsuyoshi Ito Oct 21 '10 at 03:23
  • 1
    @Tsuyoshi: I had the same feeling when I voted to migrate to meta, but after Suresh's comment I changed my opinion. The situation is not specific to the questions on this site and applies to other situations, e.g. questions posted on blogs, or even to off-line discussions. – Kaveh Oct 21 '10 at 06:36
  • 1
    I cannot understand why Suresh and Kaveh think that the question is not specifically about this website, because it plainly asks specifically about what we should do on this website. I opened a thread on Meta: http://meta.cstheory.stackexchange.com/questions/545/migrate-to-meta-or-not-when-should-you-say-what-you-know – Tsuyoshi Ito Oct 21 '10 at 12:23
  • 1
    I think the question as stated belongs on meta, but if it were edited to say something along the lines of "What do I do when I know the answer to someone's question -- be it on SE, at a conference/workshop, in a blog post, on arXiv, on someone's website, on a research visit, i.e. any venue other than a refereed publication -- particularly when it is a major component of one of my ongoing research endeavours?" then it could have an excellent home here. – RJK Oct 21 '10 at 13:06
  • 3
    I've edited the question to avoid the TCS.SX-only interpretation that most people have chosen not to follow. – Charles Stewart Oct 21 '10 at 14:24
  • 1
    I might add that in the revised form it should definitely be CW, since there is no one "right" answer. – RJK Oct 21 '10 at 15:41
  • @Tsuyoshi: Because the answers are applicable in other situations, and I am not sure that being about the site is a sufficient condition for belonging to meta. – Kaveh Oct 21 '10 at 18:22
  • 1
    @RJK: I don't think it needs to be CW, not having a right answer is not enough for making it CW. (ps: It seems that the new SE policy on CW says that if you think a question should be CW, you should only flag it for moderator attention, no need to comment that it needs to be CW, in fact such comments are discouraged.) – Kaveh Oct 21 '10 at 18:23
  • @Kaveh: Could you please provide a pointer to the new policy? I only read our official FAQ on CW -- http://meta.cstheory.stackexchange.com/questions/225/official-faq-for-theoretical-computer-science/239#239. Anyway, I will flag as you suggest. – RJK Oct 21 '10 at 22:33
  • @RJK: See this and this. – Kaveh Oct 21 '10 at 22:46
  • How are questions of this kind about TCS? They should belong to a more generic "research.stackexchange.com", where we could discuss research practice and advice of everyone's interest. The same goes for Kaveh's question: http://cstheory.stackexchange.com/questions/2953/advice-on-good-research-practices . Many answers are not tailored for TCS, they could be in MO as well. – Alessandro Cosentino Nov 16 '10 at 18:29
  • @Alessandro Cosentino: I personally don't think being of interest to people outside theoretical computer science makes the question inappropriate for cstheory. This question should be posted somewhere and this site is the natural place to post it as it is the site for theoretical computer scientists who are more likely to have related experience. I also don't see any problem in posting a similar question on MO and having one answer linking to this question. – Kaveh Nov 16 '10 at 23:12
  • @Kaveh: Right. What I meant is that for this kind of questions there seems to be too much redundancy among all the "stackexchanges". Also, many answers look suitable for a broader audience. Anyway, I am not against posting these questions on cstheory, mine was more a proposal for another website. – Alessandro Cosentino Nov 19 '10 at 14:54

5 Answers5

29

I'll share my side of the story that is mentioned in Shiva's answer.

Shiva had several questions related to my own research (space-bounded algorithms for reachability problems). We were currently putting the final touches on a full draft of the proof, and were not ready to announce the result. However, it was clear that Shiva was very interested in this area, so I emailed him saying we had a result in the works and gave him just the theorem statement. A week later, we had our draft complete and I sent him a copy. A few weeks later we shared our updated draft on the ECCC.

This worked well for several reasons.

  • Shiva and I have never met in person, but now we know each other as colleagues interested in similar problems.
  • Shiva also sent me a copy of his most recent work, which is also related and interesting.
  • The questions gave me an extra boost of motivation to complete a full draft quickly.

I recommend this course of action in the future: partial disclosure via private communication, full disclosure, and public announcement.

Community
  • 1
Derrick Stolee
  • 2,044
  • 1
  • 21
  • 34
20

Incidentally this happened to three of my questions. The authors contacted me through email and stated the main theorems from their work in progress that answered my questions. Later they were kind enough to send me a preprint before making it public. I was very happy and excited about their results.

I think it is a good research practice to share at least the main theorem to the questioner by email. If the questioner is working in the same field he might be able to appreciate your work and give very useful feedback. This will help in expanding your research circle which (in my opinion) is always good.

Once you mention that you are actively working on a problem and have partial results, people are often "ethical enough" not to work on the problem (or) discuss it without your permission. This universal strategy works very good for the entire research society.

Shiva Kintali
  • 10,578
  • 41
  • 74
15

Here, where public interest in the problem has been expressed, I think it's good for science to make known the fact that you believe you have solved the problems and provide some concrete details of what you have achieved. If you have reason not to show your hand right away, I think the goal should be to figure out what you are now comfortable saying in public, and how to present it. So I argue against private email.

Something I've seen some people do is write research bulletins, a bit like personal technical notes, that summarise findings on a topic in some degree of rigour, but without attempting to be comprehensive in the way that makes writing papers for peer-reviewed publication so time-consuming. The progress reports that Harvey Friedman used to send to the Foundations of Mathematics mailing list would be a good exemplar of that kind of thing, e.g., Self-contained posting 82: Simplified Boolean Relation Theory.

Starting such a bulletin series to deal with this, and then posting an excerpt here seems like a good strategy, since it allows you to quickly identify your achivements, while being in control of what details you make public.

I don't recommend regular blog posts for this, since they carry some unwanted associations, that they are conversational, open to revision, and not properly referenceable documents. Writing in a form for publication on Arxiv would make sense, but a Research Notes section on your publications page linking to an html page would work.

Charles Stewart
  • 4,316
  • 28
  • 45
  • 4
    I think one reason that Harvey can use this method is that he is really really really well-established researcher, i.e. a researcher with a similar research record can claim whatever she wants, but the situation is completely different for a junior researcher (or even for many senior researchers). Simply saying I have proven something does not have the same effect, IMHO, a technical report on university website or a draft on arXive/eccc are more applicable here. – Kaveh Oct 21 '10 at 09:18
  • 4
    @Kaveh: Yes, definitely, but I know of less well-established people who have used this strategy effectively. I don't dispute the existence of risks, but it's best for science if you do publish details, and you can manage the risks by choosing which details you make public. And sending private emails also carries risks. Writing such notes does not, of course, mean that you have anything concrete that carries weight on your CV, it is only a mode of engaging in scientific conversation. – Charles Stewart Oct 21 '10 at 10:38
11

I was originally inclined to close this question, but then felt it's worthy of an answer. In this situation, since you aren't comfortable releasing your work into the wild, the best solution is to email the questioner privately, if you really want to give the answer.

Alternatively, you could wait, write up the results, put them on the arxiv, and then point the questioner to the answer.

Suresh Venkat
  • 32,071
  • 4
  • 95
  • 271
  • 5
    I think ArXiving your results ASAP is a good idea. Please keep in mind that an ArXiv manuscript does not need to constitute a minimum publishable unit. I think it is perfectly ok to submit a 2-page proof to ArXiv, even though it would be obviously too short as a conference or journal paper. Resolving an open problem that other people would like to solve is more than enough. – Jukka Suomela Oct 21 '10 at 10:04
  • Other arXiv-like websites are: ECCC (for complexity theory) and ePrint (for cryptography). – Sadeq Dousti Oct 26 '10 at 11:54
6

The way to handle this situation naturally depends on a number of factors.

  1. How confident are you of what you know? (This relates to the next factor.)
  2. How much of what you know is rigorously written down? In particular, how much time will it take to finish the full write-up?
  3. How much attention did the question receive? Is it likely that many solid researchers will begin to work on the question after it was asked publicly?
  4. Is there a natural public venue to partially announce/sketch what you know in reply, e.g. via the question-asker?

Weighing these factors depending on your own experience and judgement will lead you to different actions according to one of the other responses in this thread.

RJK
  • 1,952
  • 1
  • 19
  • 27