25

I am a Ph.D. student in theoretical computer science. I have recently come across one problem on which somebody has already done work, but in that work I figured out one problem which may be small (I don’t how significant it is). I asked my colleagues about this, and some of them suggested me to pick a new problem for your first Ph.D. problem. I am in confusion what to do.

My question is: Is it okay to work on a problem which is incremental as your first Ph.D problem? Incremental in my context simply means designing a faster algorithm.

3
  • 5
    What did your advisor say? Commented Feb 11, 2018 at 17:48
  • 9
    Isn't it the job of your advisor to give you advice on this kind of thing? Commented Feb 11, 2018 at 19:06
  • 1
    The general answer to the question contained in the last two sentences is "yes -- obviously so". The specific answer for your actual situation (which is the only answer that matters) is a firm "maybe". I don't see how you expect to get a useful answer from random strangers who know none of the relevant details. Commented Feb 12, 2018 at 11:32

2 Answers 2

35

Yes!

Your goal for your first problem is just to publish something. It doesn’t have to be significant. It doesn’t have to be in a top conference. It doesn’t have to be related to your eventual thesis topic. It doesn’t even have to be in the same subfield as anything else you publish in the future. Just publish something.

This is just getting you used to the process of writing and revising and submitting and being frustrated by bad reviews and rerevising and seeing the results in print, so that none of that stuff is a barrier later. This is inoculation against the inevitable future Impostor Syndrome telling you that you can’t do this, because you already did this.

2
  • 8
    I know a couple of people who encourage (or even require) their new PhD students to work, usually in their first 6 or so months, on small problems with the intent of publishing their work, even if just on arXiv. Commented Feb 11, 2018 at 11:44
  • 4
    @101010111100 Yup, I’m one of those.
    – JeffE
    Commented Feb 11, 2018 at 14:36
2

To add to JeffE's answer: Yes, and it is encouraged for inexperienced researchers.

Check Eamonn Keogh's awesome slides on how to do research:

http://www.cs.ucr.edu/~eamonn/Keogh_SIGKDD09_tutorial.pdf

In particular, read from page 20: "Finding research problem":

Some people have suggested that this method can lead to incremental, boring, low-risk papers...

  • Perhaps, but there are 104 papers in SIGKDD this year, they are not all going to be groundbreaking.
  • Sometimes ideas that seem incremental at first blush may turn out tobe very exciting as you explore the problem.
  • An early career person might eventually go on to do high risk research, after they have a “cushion” of two or three lower-risk SIGKDD papers.

Data mining is not my field, but I heard that KDD is the top-tier conference there.

You must log in to answer this question.