Thursday, February 10, 2011

The Role of CS Postdocs

The Computing Research Association (CRA), the major professional organization for Computer Science in the US, has a new white paper on the role of postdocs in CS Research. It's excellent, very thorough, and has several interesting graphs. (Including who's hired in what areas over the past few years.) The white paper is up for people to leave comments if they are so inclined, until March 15th.

I haven't had time to read the report in depth, but a few things jumped out at me as interesting:
  • 42% of all CS PhD graduates are hired into industry immediately after completing their PhDs. (I affectionately refer to this as the Google Slurp). However, there is a large category of "other", which sadly includes the unemployed. So maybe that old statistic that less than 1% of PhDs in Computer Science are unemployed is no longer true, eh? 



  • Here's the graph of who got hired in what. Everybody is going downhill, but it looks like Architecture and Theory people are hurting the most in hiring. Which correlates with my anecdotal information. 

(Why aren't Security, CS Education, or Ubicomp listed as fields? I would not consider any of them to be subfields of any of these really. They are their own fields).

  • This table is really interesting. Compared to other fields those of us in CS/Math hardly postdoc at all:



Anyway, interesting stuff - check it out and leave comments if you have them. CRA is a well-run organization, if you comment I strongly suspect your opinions will be taken into account - not just for future white paper drafts, but also for recommendations they make to Congress and various funding agencies. (I know CRA and NSF in particular are usually fairly kissy kissy. CI Fellows anyone?)

7 comments:

  1. :) My current postdoc is a CI Fellow.
    The financial support is VERY generous, much higher than a typical postdoc salary. We heart the CRA deeply.

    ReplyDelete
  2. Nice! Do you know if they're planning to continue it this year? I hadn't heard anything.

    ReplyDelete
  3. In the last table, 67.5% of all postdocs are in biomed compared to 0.9% in CS. What percent of all scientists are in biomed, and what percent in CS? While I'm sure it's not 68:1, I bet it's still highly skewed in favor of biomed.

    For the first one, it sure sounds to me like most of those "others" are just considering multiple options. When I was job-searching, I considered postdocs, industry research, and tenure-track positions (I had offers for all three). Had I been asked at that time, I probably would've had to say "other".

    The second one is just hires into tenure track positions (I had to check the report to find out). It's probably not surprising that the overall percentage is down in a bad economy---many places have been canceling planned job searches over the past couple of years.

    Overall I feel like there are a lot of possible confounding effects and priors that haven't been taken into account in this data. I'm not sure what to make of it.

    ReplyDelete
  4. FCS, I don't know if they will be continuing. But I know that last year the renewal was really cutting it close; it wasn't clear when, if at all, year 2 funding would come in until the very last minute. Not sure if the reason was the NSF timeline for releasing funds, or something on the CRA side. I can give you (offline) a couple of names whom to contact if you are interested in upcoming year's opportunities.

    ReplyDelete
  5. Anon, good questions. I would also like to know about confounders and priors. It's worth commenting on the white paper, it's possible they are sitting on some of this information.

    GMP, thanks for the offer, but I think I'm ok for the moment. Mostly I was just curious because I'd heard they had funding issues.

    ReplyDelete
  6. security and ubicomp are most likely under "operating systems/networks"
    cs education is not a field in which you can get a phd.

    ReplyDelete
  7. Ubicomp is just real life now, as is computer-supported cooperative work. Once it's successful, it's not AI.

    ReplyDelete