5 ms·

Academic transitions are one-way only: Reflecting on my first Reddit AMA

41 points by JohnHammersley 11y ago

chriskanan 11y ago
I agree with his advice, and it feels like I'm giving that advice on a weekly basis to first and second year PhD students that think they can go to industry for 3-5 years and then get a professorship. Some students have told me it works that way in China, but it definitely isn't true in the United States. Unless you can publish prolifically at your company (Microsoft Research, FAIR, Google Research, etc.) or work at a government laboratory, your CV will quickly lose its competitive edge after you leave the university. As an aside, just managing to be competitive for professorships after one's PhD or postdoc is an enormous challenge due to the scarcity of positions and the qualifications of competitors. Based on some limited analysis I did previously, the number of publications required to get an interview is often in the range of 10-30 [1,2]. [1] http://www.chriskanan.com/statistics-on-ucsds-computer-science-faculty-candidates/ http://www.chriskanan.com/statistics-on-ucsds-computer-scien... [2] http://www.chriskanan.com/planning-for-life-after-your-phd/ http://www.chriskanan.com/planning-for-life-after-your-phd/
- pgbovine 11y ago
  That's 1000% correct. As someone who came from industry (Google) back to academia 3 years ago, and who has been involved in faculty hiring for the past 2 years, the odds are totally against traditional industry applicants unless they've been in a publication-oriented lab such as MSR for computer science (or unless they've won some really notable international award). [Edit: also, the longer it's been since your Ph.D., the higher the bar implicitly is for you. so if you haven't published much since finishing your Ph.D., with each year you wait, there will be a new crop of fresh Ph.D. graduates who look more competitive on the market.]
  - unixhero 11y ago
    Why does the World need to be like this? That edit there is soul destroying stuff.
jgamman 11y ago
caveat: it's hard to get back into academia _and get paid_. if you are sufficiently motivated you can join a research group gratis and just pitch in and do research - most PIs i've met don't mind other people generating results for them. it really is a gig though, maybe something you do when you don't need cash.
- pgbovine 11y ago
  That sounds good in theory but is unlikely to yield satisfactory results since you're not on their critical path. PIs are busy people; they will turn most of their attention on grant-funded critical-path projects. It's a similar instance of this problem: http://pgbovine.net/fellowship-warnings.htm http://pgbovine.net/fellowship-warnings.htm
tzs 11y ago
Suppose you go to industry with a Masters rather than a PhD, and then come back to get a PhD. Will you be on an even footing with other new PhDs for an academic career, or will that break after your Masters cause problems?
- JohnHammersley 11y ago
  The problem some friends (who did this) found with this route is that by the time you've finished your PhD after working in industry, you have other commitments (usually family), which makes the relocation & relatively low pay of post-doc positions difficult to cope with. It's not impossible to continue in academia this way, but it is harder than when going straight to PhD from masters, and my friends who went this route ended up back in industry after their PhDs.
shade23 11y ago
Would this also be true for a Master's program?Would it be harder to get back to academia after having spent 4 years in the industry which was more product oriented than research oriented?
- chriskanan 11y ago
  If your question is, "can I get into a PhD program after I do my MS and then work in industry a few years?" Then the answer is that's fine. But pay will be really low. We are talking about going from $90k+ in computer science to $20-30k for 4-7 years. For many people a MS doesn't help them graduate with their PhD faster for a variety of reasons (this was the situation I was in).

I agree with his advice, and it feels like I'm giving that advice on a weekly basis to first and second year PhD students that think they can go to industry for 3-5 years and then get a professorship. Some students have told me it works that way in China, but it definitely isn't true in the United States. Unless you can publish prolifically at your company (Microsoft Research, FAIR, Google Research, etc.) or work at a government laboratory, your CV will quickly lose its competitive edge after you leave the university. As an aside, just managing to be competitive for professorships after one's PhD or postdoc is an enormous challenge due to the scarcity of positions and the qualifications of competitors. Based on some limited analysis I did previously, the number of publications required to get an interview is often in the range of 10-30 [1,2]. [1] http://www.chriskanan.com/statistics-on-ucsds-computer-science-faculty-candidates/ http://www.chriskanan.com/statistics-on-ucsds-computer-scien... [2] http://www.chriskanan.com/planning-for-life-after-your-phd/ http://www.chriskanan.com/planning-for-life-after-your-phd/

pgbovine 11y ago
That's 1000% correct. As someone who came from industry (Google) back to academia 3 years ago, and who has been involved in faculty hiring for the past 2 years, the odds are totally against traditional industry applicants unless they've been in a publication-oriented lab such as MSR for computer science (or unless they've won some really notable international award). [Edit: also, the longer it's been since your Ph.D., the higher the bar implicitly is for you. so if you haven't published much since finishing your Ph.D., with each year you wait, there will be a new crop of fresh Ph.D. graduates who look more competitive on the market.]
- unixhero 11y ago
  Why does the World need to be like this? That edit there is soul destroying stuff.

jgamman 11y ago

caveat: it's hard to get back into academia _and get paid_. if you are sufficiently motivated you can join a research group gratis and just pitch in and do research - most PIs i've met don't mind other people generating results for them. it really is a gig though, maybe something you do when you don't need cash.

pgbovine 11y ago
That sounds good in theory but is unlikely to yield satisfactory results since you're not on their critical path. PIs are busy people; they will turn most of their attention on grant-funded critical-path projects. It's a similar instance of this problem: http://pgbovine.net/fellowship-warnings.htm http://pgbovine.net/fellowship-warnings.htm

tzs 11y ago

Suppose you go to industry with a Masters rather than a PhD, and then come back to get a PhD. Will you be on an even footing with other new PhDs for an academic career, or will that break after your Masters cause problems?

JohnHammersley 11y ago
The problem some friends (who did this) found with this route is that by the time you've finished your PhD after working in industry, you have other commitments (usually family), which makes the relocation & relatively low pay of post-doc positions difficult to cope with. It's not impossible to continue in academia this way, but it is harder than when going straight to PhD from masters, and my friends who went this route ended up back in industry after their PhDs.

shade23 11y ago

Would this also be true for a Master's program?Would it be harder to get back to academia after having spent 4 years in the industry which was more product oriented than research oriented?

chriskanan 11y ago
If your question is, "can I get into a PhD program after I do my MS and then work in industry a few years?" Then the answer is that's fine. But pay will be really low. We are talking about going from $90k+ in computer science to $20-30k for 4-7 years. For many people a MS doesn't help them graduate with their PhD faster for a variety of reasons (this was the situation I was in).