Monday, February 07, 2011

The Theory Postdoc Culture

FOCS Call for Papers posted. Deadline April 13.

The CRA organized a committee that put together on a white paper on whether postdocs are healthy for Computer Science. They are looking from thoughts and comments from the CS community that you can leave on that page. Suresh gave his thoughts last week.

Theoretical Computer Science, for better or worse, is ahead of this game. We have a plethora of postdoc positions available in our field. Combined with a relatively tight job faculty job market in the past few years, it is quite rare to see students in our community going immediately into a tenure-track job without a postdoc and a number of people are doing second and third postdocs. We haven't quite hit the point where it is impossible for a student to go right from Ph.D. to a tenure-track job at a major research university, but we are awfully close.

The computer science academic job market has ebbed and flowed since I was a graduate student. Most students believe the academic job market they see when entering graduate school will be similar when they get their Ph.D. Most students are wrong. Postdoc positions give some elasticity, filling in the gaps until the market moves back the other way.

When I got my Ph.D. I had a choice between a tenure-track at a good liberal arts college or a two-year assistant professor, basically a teaching postdoc, at the University of Chicago. I took the latter to keep my research career going and it did pay off for me--I had a good rookie year and the U of C kept me. It doesn't work as well for everyone but if postdocs can keep people's dreams alive that's a good thing.


  1. UCSD has been trying like crazy for the past 7 years to hire a CS theory person straight from grad school. But they might be the exception.

  2. Macneil: a long-term vacancy like that suggests that UCSD is overestimating its competitiveness and only making offers to theory superstars who take offers elsewhere. If they were to lower their standards a bit to say the quality of their own median graduate I'm sure the slot would be filled in no time.

    -Mouse 42

  3. I have a strong suspicion that more men than women are willing and able to put up with jobs that require moving to another city every couple of years. An expanded use of postdocs is therefore likely to interfere with the goal of increasing the representation of women in academic faculties.

    -Mouse 42

  4. Related to "Mouse 42's" post is the observation that high-reputation users on MathOverflow and TCS StackExchange are overwhelmingly male ...

    ... and here "overwhelmingly" means the proportion of males is 1-ε.

    A world of multiple post-doc "careers" and/or hiring evaluations that correlate with StackExchange reputation will foreseeably be associated with a "lost generation" of women in CS and math.

    This is not what was intended or foreseen ... but that is what is happening nonetheless.

  5. UCSD seems to no longer be hiring, what with the california budget crisis... Seems they missed the boat.

  6. "Most students believe the academic job market they see when entering graduate school will be similar when they get their Ph.D. Most students are wrong. Postdoc positions give some elasticity, filling in the gaps until the market moves back the other way."

    Close your eyes and whistle.

    Postdoc positions have quickly come to dominate in TCS, and with more and more postdoc factories being established (such as CCI and the Simons center), the trend has institutional and funding momentum that cannot be reversed.

    I don't fault the government or donors, as who can blame them for wanting short-term research returns?

  7. I have stated before: Lance, in the best case you are too optimistic about the future of academic job market, and as a senior person you are misleading students make wrong decisions. The academic job market in North America is not going to get back to what it was. One or two one year post-docs are OK but post-doc is NOT a job, a post-doc is severely underpaid, and have to move from one place to antoher continuously. It is really difficult to have a family and raise a child in such a situation, and I am saying this as a male, the situation is worse for females. Theory is important for us but it is NOT everything, and in my opinion, a job in the industry with some research elements is a much better option. We shouldn't move toward a math-like culture where graduates need to do multiple post-docs sometimes for decades. It is just inhuman, a typical young person already spends around 6 years as a graduate student. You are looking at the problem from the sole point of the faculty members and computer science research. I think we should also consider the effects of this change on individuals who are going to be under paid and keep moving from time to time for a longer period.

    I think the modern research business model is broken. There is a very small number of well-payed (way above average income) tenured positions with complete job security. That is the top prize. Then there are a HUGE number of graduate students and post-docs that are severely underpaid (way below average income) that carry out a large part of the research.

    My solution is to get smaller number of PhD students, but every faculty member and every department is going to act selfishly and get as many students and post-docs as they can, they are the cheap labor in the market after all. Most of the students thin they can join that small group so they have incentive to live in these underpaid positions, and they are also tricked by "for the pure sake of research and knowledge" statements from seniors (how many of tenured researchers would continue if their income is cut to the amount they pay to their students?). So the solution is to have graduate students and post-docs force the departments and faculty to do so by asking to paid more decent wages. The graduate students and post-docs in computer science need to get organized to force the universities and faculty members pay them enough to live decent lives, and that will also decrease the number of PhD students because there is not enough money to pay decent wages to so many students and post-docs, and in turn will cause the pressure on the job market to decrease.

  8. The postdoc system works quite well in mathematics. In fact, the average time to tenure and full professorship in mathematics is the same as in Computer Science sine it is usual to get tenure credit for ones postdoc years. I do not see why theoretical computer science shouldn't have the same system -- as it is, it is becoming usual for theoretical CS Phds to compete for the same jobs and postdoc positions as (say) discrete math Phds.

  9. >The postdoc system works quite well in mathematics.

    I think that depends from whose side you are looking at it. You can also say that slavery was working quite well for a very long time.

  10. It is common in medicine that localized symptoms are diagnostic evidence of systemic disorders; in such cases treating the symptoms has small utility.

    If we postulate that what Lance calls "the Theory Postdoc Culture" is a localized symptom ... then what is the systemic disorder?

    One Polya-style strategy for solving this diagnostic problem is to invert it. So let's try that.

    The Time Magazine issue for Monday, April 29, 1957 has a cover story titled "The New Age." This 20th century "New Age" heralded the greatest economic boom in human history. It was an era in which every issue of publications like Scientific American was stuffed with job ads.

    In effect, what Lance is really asking, is "Where is our 21st century 'New Age'?

    Do Lance and GASARCH know the answer? Probably not. And I don't either.

    So ask yourself ... and answer for yourself ... well ... where is it?

  11. You know, comments about how bad postdoc culture is would carry more weight if they didn't suggest mathematicians can spend decades in postdoctoral positions or that this is tantamount to slavery. Sometimes I wonder how many of the comments here are meant to undermine the position they are nominally supporting.

    I also find it strange that whenever the topic of limiting graduate admissions comes up, it always seems to be focused on other people. People seem to think "if most of my competitors had never been admitted in the first place, life would be good", rather than "what were they thinking when they admitted me? They should have realized I didn't deserve a chance to become a computer scientist."

  12. A world of multiple post-doc "careers" and/or hiring evaluations that correlate with StackExchange reputation will foreseeably be associated with a "lost generation" of women in CS and math.

    Is there any reason to think there's much correlation between StackExchange reputation and hiring? Currently, the high-reputation users seem to be a pretty random mix of people with great jobs, people at somewhat out of the way places, and students. We'll see how things develop over time, but I expect this will evolve somewhat like math contests: male-dominated (for now, in the US), correlated with a certain form of talent but by no means identical to it or representative of other forms of talent, and only modestly predictive of professional success.

  13. I tell my prospective grad student to make sure they have programming skills and to be happy with the prospect of doing research as a grad student but then going off to industry. (Then I wonder why I have no grad students.)

    In general students should be told that there are options aside from academia and to prepare for them.
    Doing research as a grad student and then doing something else with your career is a fine option.

  14. @John Sidles: That isn't it. The race to more postdocs is a classic tragedy of the commons situation, where the entire field is being harmed because the PIs who can control the flow of PhDs ignore the jobs problem. The good news is that it will eventually lead to better community awareness, but the bad news is that things will get much worse before they get better.

  15. Hmmmm ... a well-established principle of evolutionary biology is associated to the isomorphism "StackExchange reputation ... is modestly predictive of professional success" ⇔ "Social dominance ... is modestly predictive of reproductive success."

    Modest evolutionary forces, when sustained over time, exert cumulatively huge influences on both communities and species, and in consequence. In consequence, even modest evolutionary forces can play crucially important roles.

    This is not a social message that sites like TCS StackExchange intend to send ... but none-the-less, the message is received and understood, whether consciously or unconsciously, by pretty much everyone.

    As Katherine Hepburn famously reminds Humphrey Bogart in the movie African Queen: "Nature, Mr. Allnut, is what we are put in this world to rise above." The challenges that are associated to "rising above human nature" help us appreciate why policies that go beyond treating symptoms are so very difficult (in practice) to organize and implement.

    Fortunately, of all the tools we humans have for rising above our nature, mathematics is among the most powerful ... it is hard to name any great enterprise that succeeded without having solid foundations in mathematics. And very fortunately for us in the 21st century, these foundations are nowadays far stronger and broader, than they ever have been before.

  16. Where are STOC 2011 accepted papers? It's a long time since notifications already...

  17. Anonymous asserts: @John Sidles: The race to more postdocs is a classic tragedy of the commons

    As Jean luc Picard (in his assimilated Borg identity) once asserted: "That is a limited point-of-view." It's true that it's not easy being one young hominid, among 3×10^9 other young hominids, on an increasingly hot, resource-limited planet ... but associated with this sobering reality there are some really wonderful opportunities too.

    History teaches that it is not realistic to expect textbooks and/or professors to point-out these opportunities, for the simple reason that there is at present no consensus even what they are.

    As von Neumann put it: "'If people do not believe that mathematics is simple, it is only because they do not realize how complicated life is." Helping us in recognizing and realizing the opportunities that are associated to the complexities of everyday life (especially in medicine) is for me what complexity theory is all about.

  18. The list of STOC accepted papers (with abstracts) is now accessible from the STOC11 homepage:

  19. ----
    The list of STOC accepted papers (with abstracts) is now accessible from the STOC11 homepage:

    Really? I still get: (This Information is Not Yet Available.)

  20. Almost all papers are complexity (ok just a very few exceptions). It means PC choices were not balance. Also I think this STOC was supposed to accept at least 10-15% more. The number accepted papers is around 84 which is even less than some previous years.

  21. I agree that complexity theory papers seem way over-represented in the stoc accepted papers list. However, I found the list of topics diverse and it is nice to see many areas that are not usually well-represented. Also, isn't complexity (as of late) generally over-represented in stoc/focs?

    There seem to be few papers that can be classified as "Algorithms", although the ones there seem quite interesting. I don't think that is a problem, though, since this general trend of turning STOC/FOCS into mostly complexity conferences will just serve to make SODA that much stronger.

  22. ya stoc/focs are complexity conferences, soda an algorithms conference. and all take unusually large number of bad game theory papers. in general CS game theory is a self-deluding area of research so far removed from any real application.

  23. @Anon #11,

    I have expressed my view, I think the multiple post-doc culture is a bad thing for people, and yes, there are math PhDs spending very long periods as postdocs and at the end they don't find any academic jobs.

    My comment about slavery was not a comparison, I wanted to demonstrate that "things works quite well" depends on from which side you are looking at them.

    I personally have passed the point to make a different decision and also the point that changing these things would have any any effect on my job future prospect. Your personal attacks won't undermine the point I am making, I find multiple post-doc culture inhuman, and have asked a very simple question: how many of tenured researcher would stay in their jobs if they are paid what they pay to their students (or even twice that amount)? That is a quite simple question, isn't it?

  24. @anonymous
    " asked a very simple question: how many of tenured researcher would stay in their jobs if they are paid what they pay to their students (or even twice that amount)?"
    Here in France, a postdoc is paid 2103 euros (net income per month from the INRIA (France leading CS institute)) and a tenured associate professor (like myself, maitre de conference at la sorbonne university, paris 4)) is paid 2300 euros net after 4 years (national salary grid, no negociation, nothing only our 5 weeks of holidays per years, during which we work of course because teaching 200 hours a years eats all our productivity time).
    Any postdoc in CS will be paid 1.5 or 2 times this salary in the industry after 1 year.
    So yes, the salary is important to keep driving people to this area but I don't think you can talk about slavery in the postdoc US case
    (talk with some people in germany about their hellish academic hiring process).

    The very valid point is about precariousness which is a very hard thing to require from applicants: 5 years in grade school, 36 months of postdocs and then your application will be taken seriously... and for people willing to get married or a life for that matters, it's impossible, most people break up several times during the doc+postdoc stage..

    Who would enjoy living with all his belongs in two suitcases and a laptop bag and ready to move any time soon like this guy in "Heat" ?

    it's some kind of recursive descent misery: everyone has suffered the same so why do the newcommers complain. I think it's cruel, especially when it comes from people who got hired before they finished their own thesis....

    I'm really pessimistic about all this circus waiting to crush under its own weigth..


  25. I think the postdoc system allows would-be researchers to enlarge their group of collaborators and "academic friends" as well broaden their horizons a bit.
    Moreover, it smoothens the transition from being a graduate student to being a regular faculty.

    Yes, it has personal costs -- but who said that scientific career is an easy one. This last aspect is probably not so clear when one is a graduate student -- which is understandable since one rarely sees the personal lives of the people that one is supposed to emulate. So probably a bit of warning about this is a good idea from the advisor to a would be grad student.

  26. Postdoc salaries are not that low. I'm a postdoc and I am paid quite a bit. By all statistics I've seen, the average CS theory postdoc makes above $50k in the US. Sorry, this ain't slavery.

    And don't forget that profs have many other distractions, with teaching, committees, managing students/postdocs, etc. Postdocs are mostly free from this stuff, and at this point in my career, I'll take the pay cut in exchange for more research time.

    All these postdocs are great for the collective research output, but maybe bad for those who end up doing something different anyway. Or maybe not. Maybe those people still value the extra time they have been given to think about theory. I do.

  27. Yes, posdoc is slavery. In mathematics and computer science too, but especially in experimental sciences like chemistry or physics.

    The real problem in US is that graduate students and posdocs are used as bench slaves, getting little credit (as fourth/fifth authors who actually did all the dirty bench work), but being used and reused with little mercy. Professors sign papers and take away credit, and hold their posdocs and students in dependent positions; as years pass, stress wears posdoc health down, and eventually, many end up jobless, and some may even get essentially deported with their families and young children as many are foreign imported students to begin with.

    The root of this evil is economic and political logic of exploitation. Graduate students are a workforce required for experimental work, and they are payed measly, at near subsistence level, posdocs are payed a bit better but are worth more to the all-paper-signing advisors=ruthless exploiters.

    Hiring this cheap labor on graduate student level means granting many more PhD s, as giving PhD degree as part of a package is much cheaper than hiring technicians to do the dirty bench work (in Germany's Max Planck institute, every graduate student has a bench technician, the point can thus indeed be education, not exploitation). This saturates the market and drives down hiring prices of posdocs, main slave laborers in chemical/bio labs at universities. For a faculty member in experimental disciplines, quality of graduate students/posdocs is essential, as they do most of the work, and this makes the main difference between universities, as only some will be able to attract quality students to be efficiently exploited and extensively destroyed in many cases.

    In theoretical disciplines, this blatant exploitation is not of this sort. However, there is so called "profzi scheme", a pyramidal scheme of graduate students who rise the importance of their advisors in subsequent quotations and spreading of often intrinsically dubious fields. In any case, every professor wants many graduate students, and hence, many more PhDs are granted than are called for. This in turn saturates the market, and makes many losers who would be better off not going to do grad school in the first place. In mathematics, there are 1000+ PhDs granted every year in US, and there are 100-200 positions available. TCS follows this scheme.

    In the end, this extensive approach creates misery, stress and lack of flexibility in the market, that destroys the potential of the community, favoring crude politicizing and a mindset of slave labor and exploitation, where influence means everything, and science is reduced to subordinance to fashion quirks and whims of the tenured mob.

  28. The facts asserted by Anonymous in his/her post "postdoc is slavery" are mainly correct, and yet they are far from the whole story.

    My observation is that students in systems and synthetic biology—at their standing-room-only seminars—have by-and-large embraced a dry-eyed assessment that:

    (1) Wet-bench experiments are for robots.

    (2) NIH-funded careers are non-viable.

    (3) Tenure track jobs make sense iff one has an absolute passion for teaching ... and one pays a heavy price for embracing this passion.

    (4) Hybrid careers *do* make sense ... these blend public/private service; they require strong team-building and team-leading personal skills.

    Nowadays, successful academic careers also require that elusive quality called "risk aptitude." Computational resources and creative skills in applying them are an essential element of risk aptitude.

    Increasingly, even at junior levels, meaningful participation in any large enterprise (in finance, engineering, physics, biology, chemistry, or medicine) requires creative computational skills ... and this has created a huge opportunity for CS/CC/QIT.

    So, have "The Rules" for successful research careers changed? Undoubtedly.

    Have "The Rules" changed for the worse ... or for the better? That is open to debate.

    My own assessment is, the new version of "The Rules" is better ... potentially it is exponentially better ... and this is good news for young people in every branch of math, science, engineering, and medicine.

  29. I am currently a post-doc. I had to move for this job and will almost certainly have to move again when this position is over. I am willing to do this IF I get a decent academic post in the end. However, if I do not, then I wish I had just been spit out into industry after graduate school. I would start laying down roots in one place and claiming the corporate ladder.

    Also, I have a nice post-doc. If anything, I have too much freedom and get paid too well. People around me are very supportive. I realize not all post-docs are in this position. But even being in this position, I would go to industry now if I knew there were no research jobs for me after my postdoc.

    My opinion is that people can go to graduate school and then leave for industry if there are no good positions. They learn a lot, have some fun, and have something to offer industry (even if they are not paid super well). However, I feel that doing a post-doc and then going to industry is a bit more of a waste and should be avoided if possible. Do other people agree? Or disagree?

  30. I believe people are overreacting with respect to the number of postdocs in our community.

    Seriously, how many theory postdocs in the top-30 schools in the US do you know who are underpaid and / or have no chances for an academic position?

    There are virtually no theory postdocs outside top-schools AFAIK. These positions are also quite prestigious, and these people do make substantial contributions to our field and sit in program committees: Should we slow down the development of CS theory and force some of these people out of academia? I don't think so.

    Also, don't forget that for non-US PhD graduates a postdoc in the US is a de-facto necessary condition in order for them to obtain faculty positions in their home countries.

    In contrast, if you want to see slavery, you should rather look into chemistry, biology, and such, where tons of postdocs live on $38k/year salaries (this is the NIH standard) to do lab work and will never find an academic job by the pigeonhole principle.

    As long as CS people will consider industry jobs and faculty positions at top schools as the only viable options, not everyone will find academic jobs. The only possible way out I see is to understand that in the long term even excellent candidates will have to take positions at lower-ranked universities if they want to stay in academia, and that this is not necessarily a bad thing. It already happened in other disciplines, and CS is not yet there.

  31. Don't the CI fellows get paid at least 75K plus travel, relocation and expenses such as a new computer? And this is not the highest paid postdoc in theory, which is probably microsoft. Why are people complaining?

  32. It isn't about the money --- although pushing a faculty or industry position a few years back in favor of postdocs is easily a hundred-thousand dollar hit, plus benefits.

    It is mostly about the stress, the uncertainty, the inability to stay in one place for more than a year or possibly two, the additional randomness --- that makes it hard to live life, to set down roots, to manage relationships, to start a family.

    Although biology postdocs are much more poorly paid, they are also often for longer terms, which brings sanity to it all.

  33. The average graduate program takes 5 or 6 years, and thus a new graduate will be roughly 27. Go to one of the salary databases (,,, etc) and see what a 27 year old is making at Goldman, Morgan, or McKinsey. If you graduated from a top school and were recruited into one of the respective training programs you'll be bring in a minimum of $150k. Top talent will reach as high as the mid six figures. The very top talent can make partner/MD at an elite hedgefund, bank or consultancy and see compensation reaching into the seven figures (this is rare but not unheard of, I'd say you are statistically more likely to make partner/MD at one of these firms before 33 than you are of getting tenure at a top research university by that age). In fact, if you graduate from a top school you know people (who likely weren't as smart as you!) who are bring home these kind of checks. These numbers are for elite talent, of course, but elite talent is also the people who have a chance of landing a top academic gig. The skills (focus, drive, intelligence, etc) aren't the same, but they are similar.

    When evaluating a postdoc, don't forget to evaluate the opportunity cost of not spending those years doing something more profitable. A prized NSF postdoc paying $55k a year really weighs in as costing $100-200k a year, consider the lost income from not pursuing a more lucrative career. Also keep in mind that the firms mentioned above do not recruit tenure losers and it is very hard for a 30-somethingacademic (with "this is my second choice job" tattoo-ed to his head) to land one of the jobs mentioned above. Most likely you'll end up teaching service courses at You-Never-Heard-Of University for $70k a year (if your lucky). Sure you can do research in your free time (but you can also with most any other job!).

    It's time that academia started to pay real world wages! A Berkley professor gets paid less than 1/10th of the price of a four bedroom house in the area. I'm glad to see the research labs hiring away top talent from universities, because the universities don't deserve them (and hopefully this is helping drive up faculty salaries, at least in CS, as they try to compete). I can hear many of you saying "but academia isn't for profit, they can't pay industry wages." Do you know who manages your school's endowment money? Most likely a sizable chunk is in a hedge fund that charges outrageous fees (2% of assets 20% of profits), the very same hedge funds that are paying my former classmates who couldn't cut it in academia several hundred thousand a year to shuffle money. Do you know who issues debt for your university? Most likely my friends who work for investment banks. And of course, anyone who has ever looked into the finances of university administration won't see too much difference between the compensation structure there and in the private sector.

    Let's not forget the biggest budget buster in academia: big hideous buildings designed by highly overpaid architects. Universities have no problem spending $50-100 million on a fancy new building for department that is only marginally bigger than the department's current home. Administrators even like to demolish the existing structure to ensure no actual value is added to the university. This money could instead be used to endow chairs (say a $5m endowment per chair, which is a very conservative). At this figure a university could attract 5 to 10 of the very best people money could buy (with salaries at $300k a year) and really transform itself into a top program (in perpetuity!). Instead administrators prefer a fancy building, which they pay a big name architect millions of dollars to make look horrible. These have little impact on the research quality of the university.

    See also Greenspun on this topic:

  34. " ... and see what a 27 year old is making at Goldman, Morgan, or McKinsey. "

    Most people who enter a scientific career wouldn't consider a career at places like Goldman etc. even if they are paid a lot more. Nor would they be happy at such a job. Even those few academics I know who like to voice opinions about money matters mostly at lunch tables -- actually, leave all their money in checking accounts and TIAA-CREF in real life.

    Computer science is somewhat special in that it is concentrated mostly in the US. If one considers mathematics, which is more of a global community -- most mathematicians in the world would consider it absolutely ridiculous if someone compares their profession with whatever they do at Goldman Sachs (if they have at all heard of Goldman Sachs in the first place).

    We should all strive to get better better deals for *all* academics -- better job security, higher salaries etc. But comparisons with professions far removed from academia (in fact, those whose ethics are totally alien to academia) is not useful at all.

  35. "Computer science is somewhat special in that it is concentrated mostly in the US."

    Nice attitude, good to know how US CS people think of themselves. Good to know that say CS at Swiss ETH, German or UK universities, TCS at Weizmann institute or AI in Japan, or god forbid Indian Institute of technology with their presumably stupid algorithms (like PRIMES in P) amount to nothing in the eyes of their arrogant US colleagues.

  36. What anonymous says above is entirely true. Example: any of our department's trauma surgeons could immediately double their income, and cut their working hours by 30%, by leaving academia.

    But then they would not have awe-struck residents looking over their shoulder and whispering "Incredible ... I want to be a surgeon just like you."

    This give-and-take calls forth excellence in both the students and the teachers; this excellence is worth more than the sacrified income. Although I would be a bad surgeon myself, this dynamic is beautiful to watch in the operating room and clinic.

    It commonly happens that institutions seek to replace human rewards with exclusively financial rewards ... in medicine particularly, the results generically are discomfiting, morally repugnant, and in the long run, economically disastrous.

    So too in complexity theory, systems engineering, and all branches of academia in which teachers are role models for students.

  37. "It isn't about the money --- ...It is mostly about the stress, the uncertainty, the inability to stay in one place for more than a year or possibly two, the additional randomness --- that makes it hard to live life, to set down roots, to manage relationships, to start a family."


  38. All of anonymous' complaints are inarguably true ... and they are true of surgical residents too.

    But when our department recently forcibly reduced our resident's work-week to "only" 80 hours ... the residents complained bitterly ... they wanted to *continue* their 110-hour workweeks ... saying "It's professionally demeaning, and ethically wrong too, to send us home in the middle of crucial learning cases."

    These differences arise, I think, in the following daily realities: medical residents are entrusted with (and fully exercise) life-and-death authority; their progress is critically evaluated monthly; learning and teaching are open, public processes; residents are entirely confident that they are learning skills that are respected and valued; and most of all, senior residents are entrusted with the teaching of junior residents.

    What this thread is effectively about, therefore, is simply this "How might theory postdocs become more like medical residencies?"

    We'll know we've found good answers, when instead of watching Hugh Laurie play the physician "House", we can watch (say) Johnny Depp play the complexity theorist "Thurston" ... in the context of narratives that are similarly rich, and moral problems that are similarly weighty.

    I know that my wife would watch ... it is baffling to me, how much she likes Johnny Depp! :)

    By the way, our medical school has achieved full parity in male versus female enrollment ... largely in consequence of attention to the elements cited above.

    Does mathematics want to achieve the same? Then perhaps mathematics will have to do the same.

  39. @Anonymous 9:47 PM: "Most people who enter a scientific career wouldn't consider a career at places like Goldman etc." I know many people who started in science and ended up in finance. Ask anyone who graduate from an ivy in the past ten years, and they'll give you a list of names. In fact, I know a nontrivial number of people in academia who have done consulting for elite firms like those mentioned. Very few I know have turned down lucrative consulting opportunities because it would get in the way of their sacred pursuit of knowledge.

    So why are there so many talented students in academia? Consider the following: A student goes to Harvard/Mit/Princeton from a working class community or developing country. They hear stories of $400k salaries on Wallstreet and see brilliant academics making $150k salaries at their universities and they compare these side by side. Yes, they are willing to live on "just" $150k for the thrill of research (the "academic premium" that Slides talks about). What they fail to condition on is the probability of these payouts. There are loads of jobs in finance that pay $400k or more a year. You can't spite at noon in Wallstreet without hitting a ten-millionaire. There are perhaps 5 TT jobs a year leading to a salary of $150k a year (which even if you get, is still an uncertainty and years off). If you look at the EXPECTED income of each of these career tracks, which students seldom do, you start to see a more accurate picture.

    @Anonymous 9:47 PM: You seem to be quick to say the finance sector's "ethics are totally alien to academia." right after you ponder "whatever it is they do at Goldman Sachs". Like academia, finance/consulting has its problems but for the most part it's good people trying to do decent work and make money at it. As a person who has worked in both sectors, I will unequivocally say that my mentors in finance where much more interested in my success than my mentors in academia. In finance, I was treated as a professional, my boss and I would have lunch at least once a month and he would go out of his way to introduce me to senior people at other firms and get me exposure inside the firm. When I decided to leave to go back to academia, he even offered to help me get a job at a different firm if my issue was with the current firm. It was in everyone's interest for me to succeed.

    In contrast, my advisor in graduate school considered my time of negligible worth to his and showed up to scheduled meetings with probability seemingly near 1/2. I never ate a meal with him in my life. He viewed his job as throwing work at me, and never made an effort (beyond giving me a little travel money to attend conferences) to introduce me to potential collaborators or employers. When it was time to graduate he wrote a recommendation letter and considered his job done.

    If this is what you mean by "ethics are totally alien to academia" I agree.

  40. @Slides 7:22 AM: "This give-and-take calls forth excellence in both the students and the teachers; this excellence is worth more than the sacrified income." It's a lot less of a life difference to be a UCLA surgeon making $600k (compared to $1.2m in private practice) than it is to be a postdoc making $55k compared with the $300k he could be making in the the private sector. UCLA has 34 employees making over $500k, and a quick Google search reveals that nearly all of them are on the medical school staff. Also notice that the UCLA surgeon can easily pick up and go into private practice if she needs more money at vertically any age. A 35 year old theoretician can't just pack up and go into a career he could have had if he needs money.

    @Slides 7:22 AM: "This give-and-take calls forth excellence in both the students and the teachers; this excellence is worth more than the sacrified income." Also notice that the research labs and places like IAS, which don't have students, have no problem enticing top talent away from the thrill of teaching.

    @Slides 9:45 AM: "All of anonymous' complaints are inarguably true ... and they are true of surgical residents too." I don't agree that its the same for residents. By the time a surgical resident is 33, she is going to be raking in the cash. Hours worked as a resident are an investment that will almost surely pay huge dividends! You don't see too many underemployed surgeons. Hours worked as a theory graduate student is an investment that almost surely will break even or be a bust. I see tons of underemployed CS PhDs.

  41. Anonymous, I agree that there is a huge disparity in the upper-bound incomes of mathematicians relative to surgeons ... names like Shannon, Simons, Scholes, and Shaw come to mind ... to consider the letter "S" alone ... and the evidence clearly establishes that mathematical incomes are far greater. :)

    Seriously, I work with level-1 trauma surgeons .. of which no state needs more than a handful. These top-ranked surgeons earn their pay: level-1 trauma procedures like (say) a forearm replant are more arduous than can readily be conceived.

    With a foot in both worlds, I have acquired similarly great respect for mathematics and medicine. But when it comes to educational practices, it's inarguable that by almost any measure, medicine performs better.

  42. My 2 cents: If you are dedicated to research and your freedom is more important than money above a certain threshold, postdocs are an excellent idea. They enable you to build a track record apart from your dissertation advisor, and also allow you to "re-tool" with a change in focus, or an added area of expertise, if necessary. Working in an industrial research lab can also have this effect---it shouldn't be seen as such a fork in the road.

  43. Another question is whether postdocs are good for research. I think that they are, since a postdoc can focus entirely on research.*

    On the other hand, it limits the kinds of questions you can research. A postdoc cannot invest too heavily in learning a new area or in a risky line of research because to land a job he or she needs immediate results. The pressure is much higher than the tenure clock's pressure on tenure-track faculty, and is a bit higher than pressure from grants. This research pressure is probably more harmful to TCS researchers than to biologists or other fields with entrenched postdoc cultures.

    * And on submitting faculty applications, and on lining up another postdoc in case those applications fall through, and on worrying.

  44. In the first meeting with my advisor, he clearly told me that if my goal was to make money in life, I was making the wrong choice.

    The reason why academics are paid much less than in industry is the very same reason that keeps research going: Most of us are doing this because, ultimately, we have been having lots of fun doing this. And in the end, we are ready to take compromises.

    To anon, 11:39: Comparing a UCLA surgeon with an average theorist does not make much sense to me. Compare him / her instead with a medical postdoc on a $38K NIH fellowhip. People tend to forget that the NIH postdoc salary standards are used as guidelines for setting postdoc salaries overall. And guess who works at / with the NIH? Yeah, right, not too many CS people ...

  45. Perhaps $50k is less than theory postdocs "should" get or otherwise job conditions are imperfect, but it is more money than 99% of the people on the planet earn (see ) and working conditions are similarly better than most jobs in human history.

    Calling it "slavery" is offensive, given that there are many actual slaves and near-slaves in the world. This is a bit like the way people talk about being "raped" by being assigned difficult math problems in their homework. It reflects ignorance of the world, and also of the capabilities of the English language for expressing nuance.

  46. @Anonymous 1:08PM: " Comparing a UCLA surgeon with an average theorist does not make much sense to me. " If you reread my comments, you'll see that I didn't compare a UCLA surgeon with a theoretician. I was responding to John Slides comment that many surgeon's take a pay cut to work in academia. My point was that the even surgeons in academia make enough money that they start to see the diminishing utility of money. Even a moderately lavish lifestyle will not distinguish between an academic surgeon and a private practice surgeon. In contrast, there is a huge difference between a postdoc at $55k a year and a job at a bank at $200k a year, especially if you are starting a family and/or live in an expensive city. I'm not arguing the salaries should be the same, just closer together than they currently are.

    @Anonymous 1:08PM: "In the first meeting with my advisor, he clearly told me that if my goal was to make money in life, I was making the wrong choice." I never claimed that students are being duped into thinking the salaries are higher than they are. I do, however, think students are often mislead regarding the expected career path. Faculty members do a horrible job of conveying how hard it is to get a permanent research position, how long this will take, and the associated toll on one's personal life and relationships. Most graduate students think they'll get a job similar to their advisor's, and simple combinatorics shows this isn't near the truth.

    I have been to dozens of career panels throughout my academic career and never have heard a panelist speak candidly about the hardships of an academic career. On the other hand, almost every friend I have on this track has spent a good amount of time struggling with these issues discussed here. Part of this is sample bias (no one calls up the people who didn't make it to invite them to the panels) and part of it is that panelists are often (justifiably) insecure about discussing these personal struggles publicly. The result is that students go to these panels and here almost exclusively stories that end "so it was hard, but I stuck it out and I couldn't imagine doing anything else!".

    The "we do it for the advancement of humanity, not for the money" is a nice fairytale, but it is a facade. I have never once seen a faculty member turn down a raise and ask the money redirected to postdoc, student or research funding. The first priority with every NSF grant I have ever seen administered is the PI to pay himself the maximum summer salary permitted. When a faculty member gets an offer from a university he has no interested of moving to, what's the first thing he does? Decline politely and go about his day of research and discover? Nope! He uses it to try to get more money out of his current university. Often times this may include wasting a lot of peoples time at both universities. Are these selfless acts of discovery or free marketers maximizing their profits?

    Your advisor is telling a noble story, but it isn't today's reality.

  47. The complaints of the "anonymous" collective all are true and balid ... but how man of them are new?

    AFAICT, all of these points were well-covered in Richard Hamming's two great essays You and Your Research (1986) and Mathematics on a Distant Planet (1998).

    We are led to ask, given that little has changed in the past two decades, what grounds exist for optimism is there that substantial improvements will come in the next two decades?

    Blue-ribbon committees are unlikely to effect change, however well-qualified and well-intentioned their members are ... because these committees (and their various conclusions and recommendations) have not changed much over past decades and even past centuries.

    So what grounds *is* there for optimism ... particularly for young people? I doubt that any answer that I could give, would be as valuable as the advice that Richard Hamming gives, first at the end of his essay You and Your Research:

    "Some of the reasons why so many people who have greatness within their grasp don't succeed are: they don't work on important problems, they don't become emotionally involved, they don't try and change what is difficult to some other situation which is easily done but is still important, and they keep giving themselves alibis why they don't. They keep saying that it is a matter of luck. I've told you how easy it is; furthermore I've told you how to reform. Therefore, go forth and become great scientists!"

    and then at the end of his essay Mathematics on a Distant Planet:

    "As I regularly tell my students, 'In science and mathematics we do not appeal to authority, but rather you are responsible for what you believe.'"

    What has changed since 1986 and 1998, when Hamming gave this advice? Enormously much has changed ... young reseachers now have far more numerous tools ... tools that are far more powerful ... and far greater access to those tools.

    When I attend seminars in synthetic and systems biology, I see young people who have assimilated Hamming's radical message, and who are excited to participate in a math-enabled revolution that already is in-progress ... and is accelerating.

    This cheers me up very much! :)

  48. I had many friends in gradschool and beyond from fields with a strong "postdoc culture." My personal impression is that life in those fields is nasty, and I'm glad TCS is not there (yet).

    On the other hand, TCS luminaries have been day dreaming for a long time about turning our field into theoretical mathematics or some version of philosophy guaranteed to have no relevance in the real world. So what exactly did we expect to happen?

    On a more positive note,TCS postdocs are not too bad. In 90% of the cases, the host institution has no leverage over you, so you can have a good quality of life by simply ignoring the notion that you are currently employed as a postdoc. Just pretend you are a PhD student with unlimited discretionary time. Live 2 hours away just to make sure you preserve the independence.

    Some noble-sounding ideas, like using a postdoc to establish an independent track record, sound dangerously naive. If you want a job, you need politically powerful people on your side. An independent track record doesn't do much good in this field of ours.

  49. For once, I must say that I agree with Mihai Patrascu.

  50. The increasingly dismal statistics associated to academic jobs are familiar to all.

    That white papers from blue ribbon panels have accomplished little in the past to reverse these trends, and are likely to accomplish little in the future, is clear to all ... even to the members of those panels.

    That these problems do not originate in shortage of jobs, or even a shortage of research funding, similarly is clear to all. These problems are merely the (severe) symptoms of an underlying (severe) disorder, that disorder being a shortage of economically viable enterprises.

    Specifically, what are those new enterprises? And specifically, what capabilities from CS/CT are essential to creating and sustaining these enterprises?

    If there are no good answers to these questions ... then the CRA working paper can be short, confident, and valuable to all young STEM researchers:

    "No substantial alteration to present adverse trends is foreseen, and no concrete actions are contemplated."

    The new enterprises that are required to reach any other conclusion, are as readily conceived by junior members of the CS/CT research community, as by senior members.

    So what are these new enterprises?

    I pose this question because (IMHO) recent advances in CS/CT *do* offer some pretty good answers ... the question is, what are they?

  51. This is a totally serious question and I'd like an answer. Many commentators have said that tenured professors use postdocs to do the real research and then just add their name. I fully believe that happens in some but not all cases. But why would anyone do that? If you have tenure what is your motivation in padding your resume this way? It may lead to a very minor pay increase over time, but at a cost of a lot of effort. If you don't have a deep interest in research, why bother effectively doubling your workload for a 5% pay increase or something, to pull some numbers out of a hat? And, if your interest is in your own research, this practice still doesn't help...I have spoken with several professors that lament that they haven't done research in weeks because they have been busy with students and postdocs. So why do people do it? I get why untenured people do, but why once you have tenure?

  52. "Many commentators have said that tenured professors use postdocs to do the real research and then just add their name."

    This is an exaggeration, and you are compounding further exaggerations on top of it. Tenured professors still need to work, you know. :)

  53. @4:02 PM-I don't think I am exaggerating anywhere. The statement I said that many commentators made is one that I have heard from many commentators, including at least one above. The estimates on the increase in workload from advising students and postdocs are reasonable. Perhaps not doubling, but advising students AND postdocs AND getting grants to support them AND writing reference letters AND hiring them does add up to a large portion of time, at least 1/3. And I certainly am not exaggerating when I say that professors have complained that they haven't done research in weeks because they are advising students. So, where's the exaggeration, that it leads to only a 5% pay raise? Seriously, why do this?

  54. @9:31/4:38 I think you are compounding two *very* different claims. The first is senior faculty just putting their names on papers without doing any of the work. I have basically *never* seen this happen in our field. The second is of senior faculty putting so much of their time into writing grant proposals, reference letters, and all of the ancillary stuff that goes along with being a faculty member, that very little time is left for research. That I *have* seen. But I think that's mostly a combination of bad time-management, inevitable obligations, and a need for grants in order for departments to survive (most departments would be in big trouble without the overhead provided by grants). And I think these are also generally seasonal things. Overall I think most faculty find grad students and postdocs to be great partners in research, with teaching and learning a two-way street.

  55. It *does* happen. I know several blatant cases in fields ranging from chemistry to physics to mathematics. There are even more less serious cases, where role of advisor is little more than providing grant money with no direct supervision other than determining what the problem is. Such "advisors" sometimes write papers, that is at least OK but many times they don't get involved in any way. In most serious cases, that have repelled some people and basically destroyed their careers, advisors went so far to steal ideas and didn't even sign the ex students, powerless to fight back.

  56. From my experience, I would say that it does happen that advisers take undue credit. It is rare, but it does happen. What happens far more often though is the opposite, where advisers give undue credit to the student to help them along.

  57. It definitely does happen, I've seen it happen, and I know people who I trust and who later became succcessful scientists who said that their advisor did nothing. That's of course the most egregious behavior, and there is a wide spectrum of less egregious behavior too. My feeling is that the reason for this can be seen by looking at comment #41 on Read it and try to figure out why S would be interested this---I think the explanation is that he couldn't admit to himself what he had actually done and the conference allowed him a chance to pretend---it is so easy to fool yourself.

  58. There are even more less serious cases, where role of advisor is little more than providing grant money with no direct supervision other than determining what the problem is.

    You think finding a good problem that is interesting _and_ can be solved with a reasonable amount of effort is easy? It takes me a couple of months, much trial and error, and a ton of reading and preliminary work, to find problems of this nature! And it is only once I find a problem like this that I hand it over to my students and postdocs to work on, so they often don't see the process that goes on behind finding and formalizing the problem.

    In short, finding a good problem is a lot of work. The only people who think finding a good problem is easy are clueless students whose advisors have always handed them good problems.

    --An Anon Professor

  59. @An Anon Professor

    Do we have an admission - a professor who thinks providing a problem - a fundamental duty of an advisor - entitles him to sign a solution to a problem that he has not participated in?! Amazing.

    Certainly most advisors DO NOT sign papers where their contribution is limited to their basic advising duties. They are happy with acknowledgments. But hey, there are true advisors and true thieves, and some apparently feel they are entitled to their intellectual embezzlement.

  60. To 2:34pm Anon:

    What if you give someone a problem and you are not their advisor? So you don't have any obligation to fulfill the "fundamental duty of an advisor"? Why do you assume only advisors give students problems?

  61. "Do we have an admission - a professor who thinks providing a problem - a fundamental duty of an advisor - entitles him to sign a solution to a problem that he has not participated in?!"

    As has been pointed out, choosing the right problem is half the problem and is worth coauthorship. It does not matter whether the problem is from your advisor or from another colleague.

    I disagree that providing the problem is a fundamental duty of an advisor. Good students---and all postdocs---should be able to find their own problems. The advisor's role is to help guide students to find and attack the right problem.

    Some students of course need extra help, and need to be given a problem on a platter. If you are such a student and are resenting your advisor's name on your paper, then you should step back and think to the future. It is wise to be generous with collaborators, and not hold grudges over trivial issues. Your advisor no doubt has his or her reasons, probably including considerations as to how to fund your salary!

  62. "Some students of course need extra help, and need to be given a problem on a platter. If you are such a student and are resenting your advisor's name on your paper, then you should step back and think to the future. It is wise to be generous with collaborators, and not hold grudges over trivial issues. Your advisor no doubt has his or her reasons, probably including considerations as to how to fund your salary!"

    what makes you think you are talking with a student? your tone certainly reveals what kind of intimidation you would resort to in dealing with your students (and it seems you extend your arrogance way beyond that).

    it seems some vanities would be hurt very much if truth about their practices would be fully revealed. as i said, not all advisors are thieves who sign their students papers, and mine certainly was not. i got my phd way back and certainly do not steal from my students, who get their problems (and lots of help) without such expectations, i.e. signing a paper that they worked on without significant contribution from all those who are coauthors.

    but i have seen a lot of advisor abuse, from signing papers, to much worse things. however, that is not certainly always the case. just one example (and though it might tempt you to a new ad hominem attack, it has little to do with me or my field, though i know details quite well). a student/posdoc has a brilliant idea that will seed a sequence of papers in the field (say, combinatorial geometry). he tells advisor about this, and advisor not only signs papers, he completely excludes the guy from it. gives zero credit.

    ideas go both ways, with talented students often contributing both problem and solutions (in fact, without solution, student ideas are rarely credited at all).

    in normal circumstances, it is pretty clear who gets credit and for what, i.e. advisors do get signed on papers where they participated, and on those that they do not, they do not even ask to be signed. in one of my first papers, that rise to a degree of fame in the field, me and the other younger guy solved a longstanding problem, and a senior professor - not my advisor - was a coauthor (he didn't even provide a problem, that was learned about in the seminar), but in no way i considered this unfair - he provided guidance, we worked and wrote the paper together. that is not what i am talking about.

    what IS outrageous, and it does happen a lot (but only with certain type of people, who are NOT in majority), is when the advisor does not contribute to the work at all, but signs every paper. that is abuse, and i suspect the arrogant "anonymous professor" knows full well what i am talking about. this thing happens a lot in some experimental fields, and there could be some justification (especially if, as is often the practice, professor contributes at least to writing up the result), but there is no justification for signing papers in theoretical work, where problems are in fact pretty much in the open, yet some professors, whose "talent" is that of a manager, not of a scientist, chose to advance their petty careers on expense of the more talented students. I have seen this happen to several people in several fields, and it is this exploitation that any reasonable person with clear conscience would condemn. In some experimental fields abuse and exploitation is so extreme, that they drive students and posdocs to suicide - one ivy-league professor had 5 suicides over a decade in his lab, and only then he was banned from taking new "students" i.e. slaves who he so beneficially "supported".

  63. You are talking about extreme cases, often in other fields. I can't argue with them. I only argue with the student who was complaining earlier.

    Advisors deserve credit, including coauthorship, even for fulfilling their "fundamental duties," including suggesting problems to stuck students and gently monitoring their progress.

  64. Advisors deserve credit, including coauthorship, even for fulfilling their "fundamental duties," including suggesting problems to stuck students and gently monitoring their progress.

    It depends on what the problem is. If it requires original and deep insight to come up with a problem, then I would tend to agree that co-authorship is warranted. However, if it's a general suggestion to devise a different proof to someone's result, or to simplify some complicated-looking algorithm, or to extend a previous result, then an acknowledgement would be more appropriate. If coming up with such ideas is sufficient to be a co-author, then anybody can have a long publication list.