I'm a PhD student in my third year (4-6 is common in my country) and seriously consider abandoning my current topic. The new topic is in the same general field (CS related), yet in a vastly different domain and would need a quite different methods. My advisor suggested this switch, he could keep me funded in both cases, yet probably better with the new topic.

Arguments for switching are both personal interest in the new topic (it's recently trending, I was interested from the beginning, yet few positions were available) and lack of progress in the current area:

  1. I could produce some publications, yet not up to my advisors expectations (should be easier with the new topic, given the impact factors of the journals my advisor suggested)

  2. For the last 6-8 month I made barely any progress (lots of failed experiments)

  3. I would probably have to abandon my current methods anyway due to 1./2., so half a year or so will be lost learning new methods no matter how I decide

Yet I shy away from switching, mainly due to already being quite old (combination of personal problems and a switch of my major as an undergraduate) and fearing how my C.V. would look if I did take about a year longer and had this second switch...

Thanks for any input.

  • 4
    Beware of the «the grass is always greener» fallacy when switching subjects. What is your precise question? Commented Jun 2, 2015 at 13:44
  • It's more about the long term effect of making such a switch. I should probably add my field (CS related) for further context. I think/hope I'm aware of the fallacy, yet I'm rather confident about the new topic being easier to publish in (remarks from advisor about desired venues and their impact factors).
    – throwAway
    Commented Jun 2, 2015 at 14:50
  • "the new topic being easier to publish in" The notion that there are high-impact journals that are rather easy to publish in seems somewhat contradictory. In my experience, literally every venue that is well-respected is also hard to get into.
    – xLeitix
    Commented Jun 2, 2015 at 19:42
  • The new topic is not as competitive/established, so its top venues are easier to get in.
    – throwAway
    Commented Jun 3, 2015 at 6:23

2 Answers 2


For the last 6-8 month I made barely any progress (lots of failed experiments)

This sounds like you have not treated your prelims as a contract negotiation.

The best advice I've received is that when you pitch your thesis topic to the committee, make sure that everyone agrees on the structure and methodologies involved. That way, if your work falters or your experiments fail, but you followed the guidance outlined by your committee, then you have still earned a pass.

Another piece of great advice is this: nobody is going to read your thesis.

Do not switch your program and do not start over, just complete the tasks you were given and move on. If you decide to switch fields later on, fine... that's normal, reasonable, and expected.

And just because you're studying one thing in school doesn't mean that's what you have to do for the rest of your life. There have been plenty of PhDs that completely jumped fields of study, out of CS and into sociology for example... or vice versa. Besides, the latest rage these days is adding "multi-disciplinary" to your grant proposals.

  • 3
    This is frankly horrible advice, particularly if there is a publication requirement for graduation. If the experiments fail or the work falters, there are no publications, and you can't "earn a pass." Moreover, sometimes there isn't a "pitch to the thesis committee" until close to graduation. So there might be plenty of time for things to go south before the thesis committee gets involved.
    – aeismail
    Commented Jun 2, 2015 at 15:57
  • @aeismail Like it or not, the fact is that a publication requirement for program completion is more of the exception than the norm. And if you're pitching your thesis right before graduation, that's in poor taste.
    – Twitch
    Commented Jun 2, 2015 at 16:11
  • @Alexandros there are multiple ways to get a PhD in CS, and CS isn't a unique or magical program in academia. I've seen all engineering programs - not just CS - require either a dissertation bundled from multiple publications or a single body of work. It usually boils down to whats easiest for the student and the committee agrees on.
    – Twitch
    Commented Jun 2, 2015 at 16:11
  • @Alexandros My program didn't require it. I just looked at Stanford's and didn't see a requirement. Are you talking about a conference paper? I haven't read of a requirement for a conference paper, but I know that conference papers are more common in CS programs. Regardless, conference paper != a peer reviewed journal publication.
    – Twitch
    Commented Jun 2, 2015 at 16:24
  • 1
    I think this may be valid advice if the OP has no interest to stay in academia. If (s)he wants to, then finishing a bad thesis is career suicide while nobody cares if you spent more time on your thesis.
    – xLeitix
    Commented Jun 2, 2015 at 18:00

Most people won't care.

The time to PhD isn't really considered all that important unless it's highly anomalous (much shorter or longer than standard), and if you get good publications out of it, nobody is going to make a big deal about switching topics. It happens for all kinds of reasons—funding changes, or because the original project doesn't pan out for whatever reason (technical or logistical).

It will also not impact your career much, unless you're planning to continue studying one of those areas as your post-graduation career. Again, the overall quality tends to matter much more than the actual topic in most cases (particularly if you're moving into a different area from your graduate work).

  • Thank you very much for your feedback. Any idea if this is also true for industry careers?
    – throwAway
    Commented Jun 3, 2015 at 9:30
  • Again, I think it's not important, unless you're being trained in a topic the particular business you're applying to is interested in.
    – aeismail
    Commented Jun 3, 2015 at 11:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .