A reader named Hernan asked me for my opinion of a well-known rant by Jonathan Katz of Washington University, about why young people shouldn’t go into academic science since there are so few jobs and the jobs that there are stink anyway. I posted my response in the comments section, but since it seems to be of general interest I thought I’d make a proper entry of it.

Katz is correct that opportunities in academic science (at least in the US) are much scarcer than they were during the Cold War; I think government shortsightedness deserves a huge part of the blame for that. On the other hand, countless would-be grad students have already followed the invisible hand and taken Katz’s advice, and are doing quite well in Wall Street, Silicon Valley, etc. So the ones going to grad school are mostly the ones willing to assume the (by now well-known) risks: if they weren’t, they wouldn’t be there.

My fundamental disagreement with Katz is that I think PhD work is increasingly excellent preparation for industry careers. Of course, in some cases (e.g. a quantum computing PhD going to work for an Internet startup), it’s hard to argue that the PhD provides much beyond general skills like analytical thinking, teamwork, project completion, etc., and that those skills couldn’t just as well be obtained elsewhere. But even in those cases, I think a PhD at least won’t hurt your chances in industry these days (notwithstanding Phil Greenspun’s PhD expunging service). So what the PhD does is to give many people an opportunity to spend six years advancing human knowledge and doing something they enjoy, before switching to something that’s actually rewarded by the economy. (One corollary is that, if you’re not enjoying grad school, then you shouldn’t be there. But this is just an instance of a general rule: don’t choose a career option that causes you years of suffering in the hope that the suffering will end later; it probably won’t.)

Furthermore, if there used to be a stigma attached to leaving grad school for industry, I think that’s basically vanished, and that now many PhD programs even see training students for industry as a fundamental part of their mission.

I can’t comment on the rest of Katz’s complaints (the need for conformity, the burden of writing grant proposals, etc.), except to say that so far, my own experience has been more positive. Maybe the worst is ahead!

Incidentally, my comments apply most clearly to computer science PhD programs, which are what I’m most familiar with, but I believe they also apply to physics and other sciences. As for humanities PhD’s … dude, you’re on your own.

88 Responses to “Sourkatz”

  1. rrtucci Says:

    Something missing from your philosophy? I think academics should be actively involved in starting businesses—submitting proposals not only to the NSF, but also to entrepeneurs.

  2. Scott Says:

    Yes, and I think those entrepreneurs should be actively involved in giving us money. 🙂 And indeed some of them are. But of course, from a CEO’s perspective, the fundamental problem with academic research is that when it succeeds at something, the results don’t necessarily benefit the company that paid — they merely benefit all of civilization.

  3. AC Says:

    I remember reading about how offensive it was to English students when they were told their degree was “broadening” by Elaine Showalter of the MLA. It was said that there was an implication that some types of PhD’s are “just another part of rich kids’ social initiation into the world of rich adults, like a Grand Tour, or tennis lessons.” I don’t think that science PhD’s would be seen in this way, of course.

  4. rrtucci Says:

    “Yes, and I think those entrepreneurs should be actively involved in giving us money. And indeed some of them are.”
    I think Money exchanged is a bad measure for speaking about the efficacy of partnership between academics and enterpeneurs/venture capitalists. The best measure here is number of stable jobs created.

  5. Greg Kuperberg Says:

    Jonathan Katz (like many others before him) puts forward the wrong solution to the right problem. It is certainly true that the academic job market in many areas of physics and mathematics is brutally competitive and that many good people are eventually pushed out of the system. It is also true that the postdoctoral circuit has largely replaced the tenure-track stage. The tenure decision is still there, but it is increasingly pro forma.

    But as Scott says here, graduate school can be excellent training for industry. Just look at Google, which prefers to hire PhDs in CS and mathematics and maybe physics too. Renaissance Technologies also hires that way and their success is nothing to sneeze at either.

    But what Scott’s post here doesn’t acknowledge, and what Katz’s post doesn’t entirely acknowledge, is that graduate school not only trains you to do basic research, it also trains you to want to do basic research. It makes you want tenure at a research university. In this context I fervently believe that informed adults are entitled to want what they want. They should not be subjected to sermons that their hopes are unrealistic. At the same time, a sound graduate program should also prepare graduate students for a “Plan B” in industry, since Katz is correct that the job market is a pyramid. If some graduate students instead think of industry as “Plan A”, that’s fine too.

    The UC Davis math department has had to consider this issue for a long time. In math more than in CS, students tend to think of teaching-only positions, or teaching-mainly positions, as their “Plan B”. But for some of them it’s industry.

    I have four students right now. My guess is that they are in all four corners: Teaching-mainly as A, teaching as B, industry as A, and industry as B. I think that it’s important for them to be both informed and prepared on this issue and I think that they have weighed their options wisely. But again, they are adults. If some of them want to pin all hope on research faculty positions, then they still deserve that chance.

  6. Robin Blume-Kohout Says:


    Nice commentary. However, I don’t think that physics has yet followed CS in ditching the “stigma attached to leaving grad school for industry.” My own experience, supported by others’ anecdotes, is that there’s a distinct undercurrent of disdain for that track — at least if “industry” is taken to mean finance, software, etc.

    Then again, I seem to recall that Soda Hall is a hell of a lot newer and nicer than LeConte… which might indicate that there are multiple benefits to cultivating a town-gown relationship.

  7. anonymous Says:

    I think there is still a great deal of stigma associated with leaving academia for industry after graduate school, even in computer science, particularly in theoretical computer science. In particular, professors are typically clueless about industry and have very little advice to offer to students who do not wish to be in academia; moreover, they are very discouraging of those who wish to do so.

  8. Jonathan Katz Says:

    For the record, Jonathan Katz is not me

  9. Scott Says:

    Greg, I also dislike sermonizing about why people shouldn’t want what they want, and was careful not to engage in it here. If people want to try out for something as competitive as the NBA but without all the distractions like money, fame, and groupies, they should go for it.

  10. Luca Says:

    This other essay by Geenspun brings up the interesting issue of selection bias. Tenure-track faculty jobs are appealing to graduate students because they interact with people (their teachers/advisors) who are, generally, very happy with their choice of pursuing a tenure-track faculty job. The career path of the average professor at school X, however, is not going to be the same as the career path of the average student at school X, for the simple reason that school X (for a typical X) grants PhDs at about twenty times the rate at which it hires faculty.

  11. Jay Gischer Says:

    Well, Luca said that better than I could. As did Greenspun. A former department chair said to me that back when he was in a math department, they wanted to add a Ph.D. program. The dean said, sure, you can start producing Ph.D.’s as long as afterward, you promise to shoot them.

    When I left academia, 300 people applied for my job. And it wasn’t at a top-tier school, either. But I don’t regret my career path, so what the heck.

  12. Greg Kuperberg Says:

    Greg, I also dislike sermonizing about why people shouldn’t want what they want, and was careful not to engage in it here.

    Understood, and I didn’t mean to accuse you on that point, even if it may have sounded that way. I just had in mind Katz’s piece and similar.

    If people want to try out for something as competitive as the NBA but without all the distractions like money, fame, and groupies, they should go for it.

    Yes, the numbers bear out that the competition is something like the NBA. But let’s not be too self-congratulatory about this. The academic system has a significant random factor, and it also favors those who simply retain their determination to become king of the molehill. (Katz warns that it can take a very long time to get a ladder faculty position. This warning has merit.)

  13. Scott Says:

    anonymous: I believe you that such attitudes exist; personally I find them suicidal. Our own survival as an academic field depends on our ability to produce graduates that industry wants to hire. On top of that, something like half my officemates from Berkeley are now working at Google, and every time I’ve visited them I’ve felt envious (not entirely because of the free food).

  14. John Sidles Says:

    It was Dirac who said “A Golden Era allows ordinary people to make extraordinary contributions.”

    At least, that’s what Valentine Telegdi used to tell graduate students, back when I was a student in the late 1970s. The nearest verifiable Dirac quotation I have found is appended. It seems that Prof. Telegdi improved it considerably!

    The Dirac-Telegdi point remains valid today, in the sense that the smartest thing that an average student can do, is to foresee which fields are about to enter into Golden Eras. Then join that field, and have fun!

    After all, on purely probabilistic grounds, there are (almost surely) several academic fields that just now are entering into their Golden Eras, and which therefore provide ideal academic habitat to young researchers.

    What, precisely, are the emerging Golden Fields? Your guess is as good as mine! 🙂


    @inProceedings{Dirac:75, editor = {H. Hora and J. R. Shepanski}, booktitle = {Directions in Physics}, author = {P. A. M. Dirac}, title = {The Development of Quantum Mechanics}, chapter = 1, publisher = {Wiley-Interscience, New York}, year = 1978, pages = {6}, jasnote = {Lectures delivered during a 1975 visit to Australia and New Zealand. “[In the early days of quantum mechanics\ldots ] It was a good description to say that it was a game, a very interesting game one could play. Whenever one solved one of the little problems, one could write a paper about it. It was very easy in those days for any second-rate physicist to do first-rate work. There has not been such a glorious time since. It is very difficult now for a first-rate physicist to do second-rate work.”}, }

  15. anonymous Says:

    J. Sidles: Math is the one field I can think of that’s in a perpetual golden age.

  16. James Says:

    I’m not such a fan of Katz’s piece, but I doubt he literally means that no one should get a PhD in science. He’s probably just trying to scare off all but the die-hards.

    But I do see where he’s coming from. However, I think the right approach is just to give students all the data and let them make up their own minds. What I have in mind is something like this: At grad school A, they should tell their students that about B% of students finish after 4 years, C% finish after 5 years, …. Of those that finish, D% get post-docs at top-5 places (I’m thinking in mathematics), E% get post-docs at rank 6-10 places, … Of those that get jobs at top-5 places, F% get tenure-track jobs at PhD-granting places within 5 years after PhD, …

    Another important statistic that grad students often don’t think about is the employment success rate of individual professors’ students. Often the other professors know that being a student of Prof G’s is the kiss of death. But when you’re a 2nd year grad student, it’s hard to know what fields have been dead for 20 years, and the other professors often won’t say anything because it’s not very collegial. Students should also have access to the career information of individual professors’ students.

    I don’t have much sympathy for Katz’s point of view on grants — if you don’t like the handouts, stop asking for them. But the job-search grind is awful. The whole thing is fundamentally a pyramid scheme, and the least we can do is let everyone know what the odds against them are.

  17. Fernando Pereira Says:

    Katz’s claims about other careers are silly. There is no magical career path. Claiming that physics students should switch to other fields for practical reasons ignores the fact that commitment and expertise are not easily purchased. Sure, some 27-year-old CS PhDs get very good jobs — the very good ones. I’ve been involved in hiring many people in industry and academia. There are many ways to fumble an interview, but one of the most effective is to seem mercenary about one’s chosen field.

  18. Paul Beame Says:

    In CS, the history is quite interesting as you can see from
    the annual Taulbee Employment Survey done by the ACM.

    Clearly, the history is not consistent with the complaints in the rant, though the absolute numbers you can see if you go back to the rest of the survey have recently grown rather dramatically. There could easily be a problem if the trends continue. On the other hand, much of the current growth in production may also be due to the large numbers of students who switched/returned to graduate school at the time of the bust. Certainly, the pipeline of undergraduates is not as robust as it was then, but it has been leveling off in aggregate terms, and has been rebounding at many places.

  19. Todd Says:

    I have always wondered if there are any jobs out there between industry and academics. I completed my masters in EE and decided to go no further. Other than a few labs around the country, it seems that there are very few, if any, jobs where you can work on research projects without having the responsibilities of a professor.

  20. anonymous Says:

    I believe you that such attitudes exist; personally I find them suicidal. Our own survival as an academic field depends on our ability to produce graduates that industry wants to hire.

    Is it suicidal for tenured professors?

  21. Greg Kuperberg Says:

    Clearly, the history is not consistent with the complaints in the rant,

    But to be fair, CS is a substantially different case from math and physics. The path from a PhD to industry has always been much wider in CS.

  22. John Sidles Says:

    With respect to all who have contributed to this very interesting thread, IMHO the responses so far are pretty dull. CalTech provost David Goodstein said all the same things, way back in 1994, in his essay The Big Crunch:

    … The era of exponential growth in science is already over. The number of journals is one measure, but all others tend to agree. In particular, it applies to the number of scientists around.

    Just for fun, here is an anti-“Goodsteinian” post. Let’s imagine a planet with ten billion people on it (which is surely coming), and let’s imagine that the planetary economy is heavily vested in informatics (which is hugely optimistic … war, famine, plague and death being the main alternatives).

    We’ll imagine that on this planet (which is soon to be our planet) that one person in 1000 has a career in informatic math, science, and engineering research. And finally, let’s imagine that each researcher publishes one good article every ten years.

    The result is an academic literature in which one million good information-theory articles appear every year.

    Without arguing the point in detail—because it would make for a long post—I will assert that this future world makes very good sense from every point of view: economically, mathematically, scientifically, politically, and even morally.

    One reason is, Goodstein’s essay failed to foresee the coming decade of exponential growth in scientific databases (sequences, structures, and stars). His essay also failed to foresee exponential gains in system simulation capability. And we are not near to any known fundamental limits (in math, science, or resources) to their continued exponential growth. So the era of exponential growth is IMHO definitely not over.

    IMHO, the folks at Google have grasped these realities more thoroughly than academia. That is why Google is working hard to create this new world. Perhaps that is why Scott (and me too) sometimes feels envious of his friends who work there? 🙂

  23. cody Says:

    maybe im overlooking something, but doesnt the idea that the funding agencies are over-funding graduate research run contrary to the idea that grant proposals are highly competitive?

  24. anonymous Says:

    I think the problem with poor graduate school advice has been alleviated recently.

    In the past, strong students were not encouraged much to pursue their own startups. Graduate school seemed like the only way to go.

    This is no longer the case today. Everyone is very well aware of startup opportunities especially on the web.

  25. harrison Says:

    Hm…as an undergrad interested in math and theoretical CS (where there are, like, fully no research jobs outside of academia) this is something I’ve thought about a lot. And I think Greenspun makes better points than Katz; after all, I don’t think most undergrads and science grad students have the illusion that they’ll make $150k a year at MIT or Yale by the time they’re 32.
    But people’s priorities change, and by the time you realize you might actually want a house and kids, instead of (maybe) choose-one, it’ll be too late.
    Incidentally, Scott, you’ve been very…pun-happy of late. (“The Nerderer?” Seriously?) Just an observation.

  26. Greg Kuperberg Says:

    I think the problem with poor graduate school advice has been alleviated recently.

    I started graduate school in a year when the math job market looked pretty bleak, worse than in many of the intervening years. It was trivial for the department to adapt its advice to students. In fact the entering math class at Berkeley (about 75 people) was given a quite pessimistic welcoming speech right when we came. In physics it was even worse. At one job-advice session that my wife attended, some physics professor included plumbing as an honorable industry career for physics PhDs.

    However, training speaks louder than sermons. If you train students to classify smooth 4-manifolds, then they will want a career in which they can classify smooth 4-manifolds. There is nothing wrong with that even though it’s esoteric and the job market is a pyramid. But students should also get other training to prepare for and enjoy other careers. The essential point is that this other training is entirely compatible with training in pure research — it doesn’t even have to be applied mathematics. As I said, two of my students want industry as either their first or second option. They are programming in Python as well as proving theorems.

  27. John Sidles Says:

    Just to keep this pot stirred, Scott’s starting post used the economist’s favorite phrase invisible hand as follows:

    On the other hand, countless would-be grad students have already followed the invisible hand …

    Let’s suppose that the invisible hand means an academic marketplace that is rational, informed, and free.

    Let’s think about this. What is the evidence that the academic marketplace is in fact rational? Well, peer review does a surprising good job of enforcing rationality at the level of individual publications. But globally speaking, and especially when it comes to academic politics, academists are no more rational than most folks. Which is to say, professors are not particularly rational.

    Is the academic marketplace informed? Heck no. If for no other reason than this: calculating market-efficient strategies is NP-hard, even when complete information is available. So rationality is not much help, if the competition has a bigger computer (which presumably is one of the main reasons why financial derivatives were invented).

    Is the academic marketplace free? Definitely not! Instead Matthew’s rule applies: “To them that have, more shall be given!”

    The point of these observations is simply to assert that the phrase “The invisible hand made us do it” is pretty much a content-free statement that serves to short-circuit our thinking, and thereby makes us feel a little bit better about circumstances that that we are either disinclined to change, or that we feel cannot be changed.

    In consequence (and speaking only IMHO), human beings have far more power to shape the future, both collectively and individually, than is generally acknowledged. We definitely are not ruled by any kind of invisible hand … except perhaps, the invisible constraints that we place upon our own reasoning and imagination.

  28. Scott Says:

    John, in a different context I said “the invisible hand has palsy and four missing fingers” — and I think that’s very often true. But in the specific case of CS students deciding whether to go to grad school or industry, my impression is that they do respond at least partially to changing incentives.

  29. Sean Says:

    Scott, your argument is extremely weak.

    “My fundamental disagreement with Katz is that I think PhD work is increasingly excellent preparation for industry careers. Of course, in some cases (e.g. a quantum computing PhD going to work for an Internet startup), it’s hard to argue that the PhD provides much beyond general skills like analytical thinking, teamwork, project completion, etc., and that those skills couldn’t just as well be obtained elsewhere. But even in those cases, I think a PhD at least won’t hurt your chances in industry these days (notwithstanding Phil Greenspun’s PhD expunging service).”

    The best you can say is that a PhD “at least won’t hurt” your chances?! This is hardly a compelling rebuttal to Katz’s article. The opportunity cost of five to seven years in a Ph.D. program is huge, and “at least won’t hurt” isn’t good enough.

    Most of Katz’s article is devoted to pointing out facts that you can’t rebut. You claim that because of the “invisible hand” everybody already knows all these facts. Maybe so. (I’m skeptical– if knowledge is free, then why would we need scientists?) But if not, then by reading Katz’s article prospective graduate students would learn these facts. That seems fine to me. Do you have any better sources to recommend with more complete statistics?

    I am additionally skeptical because of your position as a tenure-track professor at a top university who only had a short postdoc. You aren’t the typical case, and therefore why should we trust your perspective? You claim that your comments apply mostly to CS, physics, and other sciences, but I can only be sure that they apply to future Scott Aaronsons :). (I see in the comments that you know people who left academia to work at Google, but upon what else are you basing your perspective? Katz bases his argument on his long personal experience as a scientist and on hiring committees, and includes several anecdotes.)

  30. Scott Says:

    Sean, I only offered an opinion because I was asked for one, and it’s only based on my experience and those of the people I’ve known, as Katz’s argument is based on his experience.

  31. Greg Kuperberg Says:

    I am additionally skeptical because of your position as a tenure-track professor at a top university who only had a short postdoc.

    I’m a tenured professor in mathematics at a good university, but not a “top” university in the view of most people. I had three postdoctoral positions over the course of five years, partly because of a two-body problem. I was warned about the grim job market when I was in graduate school, and I have observed our graduate program for more than ten years. I think that Scott is more correct than Jonathan Katz is.

  32. John Sidles Says:

    With respect to all (and speaking IMHO only), this discussion has mainly established that no one is “right”, so long as there are many more good people on the planet than there are good jobs.

  33. James Says:

    Greg, I entered Berkeley in math in the mid 90s, just when the really bad times were winding down, and they gave us no information at all about employment success rates of their graduates. I also think actual numbers is much more important than scare stories about plumbers. Things worked out all right for me, but very few of my classmates have jobs in PhD-granting departments.

  34. James Says:

    By the way, regarding some things written above, what does it mean for a finite set of data points to grow exponentially? I can’t think of it meaning anything stronger than that they just grow. Scientific people say things like this all the time, and I can never figure out why.

    Just to keep things interesting…

  35. asdf Says:

    James, say the points are p(1),p(2),p(3), … Exponential growth simply means that there’s a positive number c and a number N, so that if x is greater than n, then p(x) is greater than exp(c*x).

    See the wikipedia article

    for more detail.

  36. asdf Says:

    I’m a former wannabe academic (burnt-out ex-math major). I did the startup thing a couple times after graduation. Didn’t hit the jackpot, but it was ok while it lasted. Now I work at a non-profit making about half what I made a couple years ago. I like it better than working for some huckster shop and I find I don’t care about the money that much.

  37. Greg Kuperberg Says:

    they gave us no information at all about employment success rates of their graduates.

    I don’t remember that they threw any such statistics at us either. They may or may not have done that. But they did give us a welcome speech, which as best I remember had some “just to warn you” comments in addition to the positives.

    Maybe my recollection of this isn’t fair. It would be strange to talk about the job market in the welcome speech for entering graduate students. Maybe I am conflating different incidents. Anyway, somehow people knew that the job market was bad. I would still say that supplementary or dual-purpose training is a better response to that reality than even the most masterful presentation of hiring statistics.

    By the way, regarding some things written above, what does it mean for a finite set of data points to grow exponentially?

    (As opposed to an infinite set, where the asymptotics are defined rigorously.) It means that an exponential curve is a parsimonious explanation of the data. Or in some cases it means merely that the logarithm of the data is a better presentation than the data verbatim. For instance, people often say that inflation is exponential, even though of course the data set is finite. That pronouncement can mean either of the above.

  38. Sean Says:

    I realize that you are giving your opinion, Scott, so are all of us. I hope the tone of my post didn’t offend you, since that wasn’t my intent.

    I think that I was hoping more for a rebuttal of Katz’s arguments, but I read your argument as saying that Katz is basically right but by the invisible hand it is okay. (And Katz himself concludes, “The result is that the best young people, who should go into science, sensibly refuse to do so, and the graduate schools are filled with weak American students and with foreigners lured by the American student visa,” so he is well-aware of the invisible hand.) You know better than I who your audience is, but if it is prospective graduate students then appealing to the invisible hand is bad advice.

    I haven’t read the comments through, but the best advice for graduate students I think would be to find the statistics. Like James, despite the invisible hand, I have no idea what they are. I’d guess maybe 10% of PhDs end up with tenure at PhD-granting institutions, and maybe 20% of postdocs? Is there a way of determining this information on a per-school or per-advisor basis?

  39. Scott Says:

    Sean, thanks for the clarification! I’m starting to regret my “invisible hand” comment; people are reading more into it than I intended. My advice to a prospective grad student would be the following: “Would you be happy to spend 6-7 years working on basic research questions, and after that to transition to a career in industry? If not, would you be happy to become a professor at a primarily teaching-oriented institution? If not, are you willing to throw your entire life into becoming (in Greg’s phrase) a ‘King of the Molehill’? If not, then there’s a serious possibility that grad school is not the right choice for you.”

  40. ms Says:

    While on this topic, says criticism is sometimes overrated 😉

  41. Greg Kuperberg Says:

    The best advice for graduate students I think would be to find the statistics.

    Statistics are fine, as long as the point isn’t to determine as precisely as possible how pissed you should be that the job market is bad.

  42. Pascal Koiran Says:

    On CS versus physics: I’ve heard from a department chair at a good state university that in the US, there are actually few postdoctoral positions in computer science (Scott would be one of the exceptions).
    This may be a good thing for CS PhD’s: they do not have to wait until they are 35 or 37 (like in Katz’s story) to get a tenure-track position.

  43. cody Says:

    im in it for the molehill.

  44. James Says:

    Greg: The point about using statistics as an excuse to get pissed is a good one. (Amusingly, the word ‘pissed’ here can be correctly understood in the US or non-US sense….) I hadn’t thought of that. Also I guess the point of exponential growth of finite data sets was really just my way of complaining that when people talk about some a data set being exponential, then usually actually have a lower bound on the growth rate but just don’t say it. So for instance, the statement “my bank balance increased exponentially each month last year” means only that my balance increased each month, but the statement “my balance increased at least 10% each month last year” means much more, and I wish people would just say all of what they mean rather than just part of it.

  45. Simon Says:

    I think it is a sky is falling argument… People choose PhDs for whatever reason. The primary motivator may be different for Americans and foreigners. But the market responds to the availabilty and grows. Look at Genetech, Amgen, Google, Monsanto..

    I’m of the philosophy that you can’t have too many PhDs as long as they are interdisciplinary, and that academia is not the final destination.

    Academic growth also responds to developing markets, and population growth. So, soon, as China develops and perhaps as English becomes more pervasive (even Europe), we should see increased job growth.

    Science is also hard, by definition we are exploring the unknown. For sure, keep realistic expectations. How many kids want to become Michael Jordan, how many waiters in LA aspire to be actors? Look, see, and learn.

  46. Nick Says:

    I think employment perspectives vary widely between CS and physics, especially the theoretical variety. To my knowledge, there is no such thing as an industry job which substantively requires training in theoretical physics, period. (Please correct me if you can give a counterexample, and include a link to the job posting.. ).

    From this perspective, talk of preparing physics graduate students for jobs outside physics is almost meaningless. I see the problem more as an issue of truth in advertising. It should be made clear that only a fraction of them will end up doing what they are trained for (their PhD advisor’s job) before they join a program.

    There is a mis-perception in society about career prospects in science, extrapolating the idea that the more you stay in school the better your life will be. Applicative fields such as CS are indeed rewarding in a predictable way. On the other hand, things like theoretical physics are similar to music and competitive sports with a few outstanding and well publicized success stories, and many failures.

  47. John Sidles Says:

    Nick says: To my knowledge, there is no such thing as an industry job which substantively requires training in theoretical physics, period,

    Hmmm … Nick, I had to think three times before replying. (1) Your assertion is flat-out wrong. (2) But some folks might be skeptical. (3) But if I give the evidence, that dilutes the need for imagination and enterprise … which is what the folks who hire are *really* looking for.

    So I’ll compromise.

    Nick, here’s an solid job lead. But just to mention … this particular job lead is a little bit stale … one year ago it was an incredible job lead. A job lead that anyone on this forum could have found … with a modest amount of enterprise, imagination, and diligence.

    The folks who figured this out a year ago are … well .. they’re pretty much the folks who are running the show.

    Welcome to the 21st century, everyone! 🙂

  48. Scott Says:

    Let me reiterate what I think was my main point, but which I didn’t emphasize enough. If someone does important and original research for their PhD, and later switches to an unrelated career (e.g., in Silicon Valley), I don’t see the time they spent doing their PhD work as at all wasted. Even among people who do stay in academia, a huge fraction of them did their most important research as grad students. So there’s simply no need to see grad school as training for something else — you can instead see it as the “real game,” which most people sensibly leave once they reach a certain age, but which a few stragglers stick around in to coach the next set of youngsters. From this perspective, the fact that most PhD students won’t become professors is analogous to the fact that most 20-year-old Olympic swimmers won’t spend the rest of their lives in Olympic-swimming-related careers (except for the few who become swimming coaches).

  49. lylebot Says:

    It should be made clear that only a fraction of them will end up doing what they are trained for (their PhD advisor’s job) before they join a program.

    Most of us aren’t even trained to do our advisor’s job. Speaking for myself and my labmates, we haven’t gotten training on how to write grants, how to advise students, how to design and teach courses, how to run a research lab, etc, etc. We get training on how to do research, but my advisor doesn’t really do that much research anymore—that’s what he has students for.

  50. John Sidles Says:

    Let me agree with what Scott says above, and amplify it.

    It is arguably true today, as in the past, that the most exciting research is taking place outside of academia. As evidence, check-out this amazing animation of sequencing technology that is taken straight from the Pacific BioScience job ads that I linked to in my previous message.

    AFAICT, there is no field of intellectual enterprise that will escape the challenge of the developments that are shown in this animation … and that definitely includes complexity theory, information theory, and simulation theory, in both their classical and quantum embodiments.

    Enterprises of this scale and intellectual daring are the reason why Scott, and me too, sometimes feel jealous of our industrial colleagues.

    I once tried to recruit a scientist working at one of these cutting-edge technology companies to apply for a tenure-track faculty position. His/her response was as follows: “Let me get this straight. No stock options, I have to hustle my own research grants, *and* I have to teach undergraduate classes? Plus, I have to deal with academic politics? No thanks!”

    That is why academic careers don’t make sense unless a person has an irresistible passion for teaching … and this is true even at so-called research universities … in fact, it is especially true at research universities.

    Meanwhile, both inside and outside of universities, research opportunities are absolutely unbounded … enormously greater than in any previous historical epoch for sure.

    So are we living in a great century, or what? 🙂

  51. Nick Says:


    I agree that there are a lot of cool emerging fields, but they do not require a physics Ph.D. . I meant something that specifically calls for knowledge or training that is unique to the graduate theoretical physics curriculum, the same way Google or Microsoft look for CS Ph.D graduates. My challenge was to see a job ad where it says, for instance, “knowledge of advanced quantum field theory required”.

    I still claim that there is no such thing.

    Don’t get me wrong, I didn’t say one can not get an industry job with a physics degree. But just look at your example. They ask for “individuals in the fields of nanotechnology, biochemistry, optics, high speed digital design and bioinformatics”. Those are disciplines other than physics, so graduates from the respective programs will be preferred over physicists.

    All I meant was just that: no industry specifically wants a theoretical physicist, they may settle for one in the absence of a closer match. I simply pointed this out as a difference in comparison to CS.

  52. John Sidles Says:

    Nick, last time I checked, modern optics is field theory. And within the nanoscale, strongly-coupled environment in those wells, the quantum optical dynamics is nonlinear as heck … a noisy Jaynes-Cummings Hamiltonian, I would guess … with cavity QED effects most likely … which adds up to a pretty sophisticated problem in applied QIT to extract a “GCAT …” signal with an optimized ROC curve.

    So this particular biotechnology sure looks to me like a quantum physics problem along similar lines to (but far more advanced than) Julian Schwinger’s wartime job as a radar waveguide analyst.

    Which was not a bad experience for Schwinger … that job was where Schwinger learned field theory … pretty much all of Schwinger’s Green function and source methods were learned on-the-job in the context of radar engineering, and were applied to fundamental QED only several years later.

    And I will go so far as to assert most modern biotechnologyies are rapidly evolving into quantum biotechnologies … the sensing, dynamics, and simulation bits, for sure … and these three technologies are becoming “the tail that wags the dog.”

    Now, to be fair, when it comes to string theory, the folks who are dismayed about the lack of industrial jobs are perfectly justified … but they don’t receive too much sympathy from me! 🙂

  53. asdf Says:

    Maybe we’ll see fusion energy generation start to become real in coming years. At that point there will probably be a lot of demand for theoretical physicists in that industry. There is probably some demand like that in the existing nuclear industry, on both the civilian (power generation) and military contractor (mushroom cloud) sides.

    There is certainly lots of quantum theory used in developing semiconductor technology, but I guess not at the nuclear level.

  54. Joshua Schraiber Says:

    This is why I am going into mathematical, theoretical population genetics—no one wants to do it anymore 🙂

  55. Greg Kuperberg Says:

    Maybe we’ll see fusion energy generation start to become real in coming years.

    I hope that you mean tokomaks, in which case, don’t hold your breath. The history of controlled fusion is a cautionary tale for quantum computation.

    If you mean cold fusion, then you also shouldn’t hold your breath, but it isn’t a cautionary tale for quantum computation. (Except maybe for D-Wave.)

  56. sp Says:

    Why sciencists just can’t simulate nuclear fusion on supercomputer instead making expensive unsuccessful experiments (don’t enough computation power?)? I would say that even now nuclear fusion is more realistic than quantum computation.

  57. Alex Says:

    @Greg Kuperberg

    Being somewhat ignorant when it comes to the practical details of fusion, I am curious, why do you think tokomaks are a cautionary tale for QC but not cold fusion?

  58. John Sidles Says:

    Asdf says: Maybe we’ll see fusion energy generation start to become real in coming years.

    Asdf, the latest IBM Journal of Research and Development is a theme issue on large scale simulation, and it includes a nice article by Ethier et al. out of the Princeton Plasma Physics Laboratory, precisely on the topic of fusion power generation. So you might enjoy taking a look at this article.

    IMHO, this is a historic issue of the IBM Journal. There are numerous excellent articles in it that span a broad range of topics. No matter what your field, there are likely to be one or two articles that are of professional interest to you.

    This illustrates that pretty much every field of science and engineering is being impacted by the advent of large-scale simulation capability … and this definitely includes quantum science and engineering.

    In addition to their science and engineering purpose, these massive computational simulations are increasingly serving a vital social purpose: they bind together the global-scale enterprises that bring new technologies on-line.

    The laptop I am typing on has about 30,000 times more computational power than the computers I used as an undergraduate, and these IBM Blue Genes have about 30,000 times more computing power than my laptop. So IMHO, this kind of computing power will become routine during the professional lifetime of pretty much every graduate student who reads this forum.

    As somebody or other once said (possibly Stalin?): “Quantity has a quality all its own.” The advent of large-scale simulation capability is qualitatively altering the rules of science and technology.

    The reason is simple: these simulations have become the primary arena where fundamental science, practical engineering, and the social aspects of technology development, are first united.

    In other words, nowadays, these simulations are where good jobs are being born today. Which is good news.

  59. Xerxes Says:


    Hot fusion is really possible (and already has applications of the mushroom-cloud variety), but workers in the field been over-promising results for the last 50 years. Unlimited “too cheap to meter” fusion power has been “just 20 years away” since the 1950’s.

    Cold fusion is what happens when you have a seemingly favorable systematic error in your experiment and go to the press to gloat about it instead of quietly working it out with your peer reviewers.

    Both are cautionary tales for scientists; since QC is physically possible but practically difficult, it falls into the former category rather than the latter. The moral being that they should avoid promising more than they can deliver.

  60. rrtucci Says:

    Cold-Fusion: No real gold, in substantial quantities, has ever been found in this mountain. Furthermore, the miners that made the original claim are very prone to tall tales.

    Hot-Fusion: There appears to be real gold in this mountain. However, the miners (plasma physicists) that I know, think that ITER is an expensive boondoggle that will teach us little about how to build a practical machine. So here you have the strange situation of academic miners digging for FALSE gold, and aware of it. Also, miners who do not go along with this charade are not welcomed.

    D-Wave: This company was founded to pursue an idea that sounded like it could work. But it didn’t. So G.Rose gradually modified the original idea into a chimera. At every step of the way, he would bully anyone who disagreed with him. Unfortunately, G.Rose found that throwing large amounts of money at a crappy engineering idea rarely turns it into a good one. Initially, D-Wave bought a mine that sounded promising. But they found no real gold in it. However, they found that their mine is full of fake gold, and that there are numerous rich people out there that can be fooled into buying fake gold. So now D-wave is mining and selling fake gold.

  61. Paul Says:

    Speaking of statistics for physics, though they are a little out of date and only an aggregate, about 2 minutes of web search yields AIP employment statistics for physics PhDs surveyed in 2001. Note that for the most recent 5 years of PhD graduates they explicitly exclude postdocs from the stats.

  62. Greg Kuperberg Says:


    A central question of quantum computation is whether we will have quantum computers. At the moment we don’t of course. So instead, there has been a debate as to whether they are possible in principle. The optimists have had a ton of good ideas in this debate. Meanwhile the skeptics have had little to say that looks like good research, in fact not all that much that even looks like viable research. In particular, they have not found anything remotely like a fundamental obstruction to quantum computation. So the status quo suggests that we will eventually have quantum computers, even though it’s completely inappropriate to promise any kind of profit or timeline.

    As I said, controlled fusion is a cautionary tale. That group also had a tacit debate as to whether or not controlled fusion is possible. That discussion has led to exclude certain proposals that might have seemed plausible, such as muon-catalyzed fusion. Here as well, there is no strong argument that controlled fusion as a whole is a waste of time. The bad news is that even though no one identified an insurmountable obstruction, progress is stalled after 60 years of trying. The community understands that controlled fusion gradually gets easier as you scale it up. You can speculate that if humanity’s industrial plant were three or four orders of magnitude greater than it actually is — on some other planet to make that possible — then we’d have controlled fusion.

    This comparison came out of a discussion that I had with Gil Kalai. The discussion led to the concept of a tautological kibbitzer who, unlike a constructive skeptic, doesn’t need new research to support his view. Where as the constructive skeptic posits reasons that you can’t pull your car out of the ditch, the tautological kibbitzer only reiterates well-known reasons that your car is still in the ditch. The problem is that the tautological kibbitzer is sometimes the most correct, even though he’s not helpful.

    Anyway cold fusion is not in the same category; it’s just crackpot. It was at first crackpot at a higher level than, say, one-page proofs of Fermat’s Last Theorem. Even so, the obstructions are fairly fundamental, so it is crackpot to just blaze forward with experiments. Experiment without theory is as bad as theory without experiment.

  63. Scott Says:

    Greg, that was one of the best technological prognoses I ever read.

  64. John Preskill Says:

    Fascinating thread. The issues being discussed are quite complex and of course there are no simple answers. The views of any one of us will be filtered through our personal experiences and observations. Katz wrote a forcefully argued essay based on his own experiences. I am posting this merely to point out that my experiences have been different than his.

    Katz says: “Why am I (a tenured professor of physics) trying to discourage you from following a career path which was successful for me? Because times have changed (I received my Ph.D. in 1973, and tenure in 1976).”

    Despite the unavoidable demographic pressures highlighted by Goodstein (quoted above by John Sidles), I have not witnessed such a big change. The academic job market has been cyclical. When I was an undergraduate, around the time Katz received his Ph.D., I was warned repeatedly that it was foolish to plan on a career in theoretical physics because there would be no jobs. The outlook also seemed bleak in the early 1990’s. It seems better to me today.

    Katz says: “I became a scientist in order to have the freedom to work on problems which interest me. But you probably won’t get that freedom. As a postdoc you will work on someone else’s ideas, and may be treated as a technician rather than as an independent collaborator.”

    But I have witnessed many postdocs who have developed their own ideas and have succeeded brilliantly. I don’t think the same degree of intellectual freedom is at all common in the commercial world.

    Katz says “The result is that the best young people, who should go into science, sensibly refuse to do so, and the graduate schools are filled with weak American students and with foreigners lured by the American student visa.”

    I noticed that American Ph.D. students in theoretical physics were in short supply back in the 1980s, but to me this seems less true today. In any case, US universities have a long tradition of training foreign students who choose to remain in the US after their degrees and eventually become US citizens. I hope that continues.

    Katz says: “Now you spend your time writing proposals rather than doing research. Worse, because your proposals are judged by your competitors you cannot follow your curiosity, but must spend your effort and talents on anticipating and deflecting criticism rather than on solving the important scientific problems. They’re not the same thing: you cannot put your past successes in a proposal, because they are finished work, and your new ideas, however original and clever, are still unproven. It is proverbial that original ideas are the kiss of death for a proposal; because they have not yet been proved to work (after all, that is what you are proposing to do) they can be, and will be, rated poorly.”

    It’s true that senior scientists spend a lot of time on proposal writing, and that the review of proposals is imperfect in many ways. Certainly highly deserving proposals are sometimes overlooked. But on the other hand, many deserving scientists do get funding, and they manage to do highly original and valuable work.

    Katz says: “If you are in a position of leadership in science then you should try to persuade the funding agencies to train fewer Ph.D.s.”

    Maybe. I confess that I have been a big contributor to the problem, by having 40 Ph.D. students so far (and I hope I am not done). I cringe when I read David Goodstein implying that professors who train too many students are putting an intolerable strain on the system. But I don’t feel too guilty about it. Some of these students have landed permanent jobs doing science and some have not. As far as I am aware though (and admittedly there might be some exceptions I don’t know about), they are all to varying degrees living productive, fulfilling, and happy lives.

  65. Greg Kuperberg Says:

    I confess that I have been a big contributor to the problem, by having 40 Ph.D. students so far

    Pig. I have only had one PhD student. You owe me. 🙂

  66. cody Says:

    i dont like how Katz is ultimately discouraging people from the pursuit of science. it seems to me that if there are too few academic positions in science, we ought to encourage more people to know more about science, at all levels.

    i imagine the eventual result of more people knowing more science would be more academic resources dedicated to science.

    and at the very least, i would prefer a society full of scientific-minded failures than unscientific-minded successes, (no offense intended, i myself am probably in that category).

  67. Job Says:

    The idea of an unscientifc-minded successful person bothers me.

  68. Greg Kuperberg Says:

    Alas, my not very polite joke about John’s students gnaws at me, even though John has very politely not responded.

    Just in case anyone is confused as to what I really think, you can see John’s list of students here. My main impression is: wow, this list would make a very good physics department. Of course it is to John’s credit to have taught all of these people.

    In honesty, I can’t help but feel a bit jealous. It’s also true that Katz is a little bit correct: John has 15 or so students who can take PhD students themselves; of course it’s not numerically feasible for the average advisor to have that many. But I’m doing well enough, since I have (I think) 4 students in the pipeline, even though it’s true that I only graduated 1.

    On a related note, I begin to wonder whether Katz’s essay is in the tradition of “A Modest Proposal”.

  69. John Preskill Says:

    Oh, I’m not polite … I just couldn’t think of any response that was funny enough. Besides, I usually don’t answer messages that use “smiley” emoticons.

  70. Michael Bacon Says:


    Does that mean you do respond to other emoticons? 8o|

  71. Jack in Danville Says:

    Nuclear fusion is always explained to the unwashed masses as harnessing the power of the sun. Yet as a tautological kibitzer I notice the real power behind fusion in the sun is gravity. Is it any wonder that after 60 years there is still more energy going into controlled fusion experiments than coming out of it? There just is not enough mass on this planet. (Yeah, I know some experimentialists claim net energy for a few milliseconds or so.)

    Scott, thanks for the latest installment of Democritus! Looking forward to more.

  72. anonymous Says:

    Off topic, but you might find this interesting:

  73. rrtucci Says:

    Dang. Quantum minigolf is brilliant. Keep the good software coming.

  74. Greg Egan Says:

    There’s also quantum soccer.

  75. John Sidles Says:

    The quantum golf and quantum soccer animations are fun, but that kind of animation contributes to a widespread (but mistaken) impression in engineering students that quantum wave-functions are knowable in principle.

    This is of course not true of noisy system … which are the only systems that engineers ever get to work with.

    To correct this impression we show students an animation of the IBM single-spin MRFM experiment in the context of a QIT lecture entitled How does the Stern-Gerlach effect really work?

    You will note that the QIT lecture is centered upon “Peter Shor’s Principle” (as first articulated on this very blog!):

    Interpretations of quantum mechanics, unlike Gods, are not jealous, and thus it is safe to believe in more than one at the same time.

    This turns out to be an excellent informatic principle for engineering students to analyze. Furthermore (as the lecture shows) Shor’s Principle turns out to be as rigorously true in the classical domain as in the quantum domain.

    By the way, I’m sorry that the animation file is so large (16 MB), but hey, it’s a full-length animation of a 13-hour IBM experiment! Holy Andy Warhol! 🙂

  76. rrtucci Says:

    “a widespread (but mistaken) impression in engineering students that quantum wave-functions are knowable in principle.”
    No. For certain situations like an atom in a lasing medium, sure, because the atom is strongly coupled to its environment. But if the coupling is very weak, as is assumed for qubits in a QC, then we may assume the system is in a knowable pure state, at least at the beginning, before we start measuring it. Your software has almost nothing to do with quantum computing

  77. rrtucci Says:

    P.S. I think your “Peter Shor” principle is wrong too.

  78. John Sidles Says:

    Rtucci says: Your [simulation algorithms] have almost nothing to do with quantum computing.

    You are exactly right, rtucci … noise-free and/or error-corrected quantum computers are precisely the kind of quantum systems that we engineers don’t even try to classically simulate … because we know from Scott’s Scientific American article that we won’t succeed! 🙂

    But when it comes to (sufficiently) noisy quantum computers, well heck, these devices can be simulated classically, and there is no way to uniquely deduce from the inputs and outputs of these noisy computers, what the computers’ internal quantum trajectories might be.

    To put it a common-sense way, a thermal density matrix—conceived as the output of a disastrously noise-ridden quantum computer—can equally well be constructed from an ensemble of number states, or an ensemble of coherent states.

    Since no measurement can tell the difference, we might as well classically simulate noisy computers using the quantum states that are most efficient to compute. Not as any kind of statement about quantum reality, but simply as a matter of efficient engineering practice.

    The IBM MRFM experiment, from this point of view, can be conveniently conceived as an exceptionally noisy one-bit quantum computer. The sole experimental objective is to determine the spatial location of that qubit … basically to determine whether any qubit is present at all.

    This limited QIT objective is not as sexy as a high-level computation, but when we simulate this kind of noisy experiment, that same noise gives us tremendous freedom to exploit QIT principles in the design of efficient simulation algorithms.

    For at least the past forty years, a steadily larger community has been pursuing these ideas, in the simulation of steadily larger quantum systems (perhaps without always understanding these ideas in modern QIT terms). And this progress is still accelerating. It’s fun!


    @article{Buhrman:06, author = {H. Buhrman and R. Cleve and M. Laurent and N. Linden and A. Schrijver and F. Unger}, title = {New Limits on Fault-Tolerant Quantum Computation}, journal = {47Th Annual {IEEE} Symposium on Foundations of Computer Science ({FOCS}’06)}, year = 2006, pages = {411-419}, publisher = {{IEEE} Computer Society}}}

  79. Ralph Kelsey Says:

    Remarks on Goodstein’s essays and books.
    (1) “The big crunch”: Sure, good point, but an extreme take . I’m not sure he is entirely serious.
    (2) “States of Matter” is one of the best undergraduate physics books of all time. Easy to read, in fact downright funny at times. For example, a chapter on statistical mechanics (or maybe kinetic theory) starts something like this: “Boltzmann worked on … for decades, before dying by his own hand. Ehrenfest carried on the work for another dozen years before meeting a similar fate. We shall proceed cautiously.”. A wisenheimer. The books ends with several up to the minute (back then) big questions. I forgot what they were, but they sure sounded interesting. I have long hoped for a second edition so I could find out how they turned out. If anyone out there knows David, please forward my request for a postscript.
    (3) “Out of gas, the end of the age of oil” Popular Science/Bad News category.
    (4) Checking Amazon, I see he has a couple other books including a “Feynman’s lost lecture”

  80. Wilmer Gunsl Says:

    Boltzmann’s tragic suicide, by hanging himself from a curtain, cannot be trivialized. Ehrenfest shot his young son through the eye and then turned the gun on himself. In both cases, family members were emotionally damaged for life. These are not humorous matters and I question Mr. Weisenheimer’s taste and judgment.

  81. John Sidles Says:

    Ralph Kelsey says: [David Goodstein’s 1975 textbook “States of Matter”] ends with several up to the minute (back then) big questions. I forgot what they were, but they sure sounded interesting. I have long hoped for a second edition so I could find out how they turned out.

    That’s a great idea, Ralph! Here (from pages 487-8 of States of Matter) is a 1975 Goodstein passage that provides a starting point for answering your question.

    When a theory is first proposed, the exciting thing is to show that it it right. … and so, unless the idea is a bad one, there is a tendency at first to confirm it. As evidence in its favor accumulates, the idea becomes better established, and at this point, if its correctness is not yet proved beyond doubt, its importance certainly is. The idea becomes a part of the received wisdom. Now an important inversion occurs. It becomes more exciting to show that it is wrong. Further verification would just be adding points to a well-known curve, but now falsification amounts to combating prejudice that stands in the way of new insights. All those factors that at first tended to ensure verification of the idea now operate in reverse, allowing one to doubt the original interpretation, allowing scope for an attack on the entrenched idea. At this point, with clever, resourceful, passionate advocates operating on both sides of the issue, one as reason to hope that some approximation to reality will emerge.

    Now it’s thirty-three years later. What fields of science have undergone (or are undergoing) the “Goodstein Cycle”?

    Example 1: Is biology experiment-driven, or is it theory-driven? This dilemma was much-discussed in the decades 1970-1990, and recent developments have resolved it by triangulation. Disciplines like biology, astronomy, geophysics, etc., are neither experimental nor theoretical, but rather, are observational. The Human Genome Project and the Sloan Sky Survey were the first examples of observational Big Science … but they will not be the last.

    Example 2: Has science’s “Big Crunch” ended? Here by “Big Crunch” we mean Goodstein’s 1994 assertion that “The expansion of science … was guaranteed to come to an end.” Without belaboring the point, I will offer for reflection the idea that the Big Crunch has ended; that the exponential growth of science has resumed; and that this new epoch of exponential growth is being driven by a new kind of “Big Science” whose foundations are largely in observation and simulation, with a healthy admixture of theory and experiment.

    Example 3: Are quantum systems hard or easy to simulate? This question comes closest to matching the Goodstein Cycle. For several decades the received wisdom has been that quantum systems are infeasible to simulate with classical resources … and this is surely true of systems that have zero noise and/or are error-corrected. But in the last few years, the appreciation has grown that this received wisdom is logically consistent with a world in which quantum systems that are noisy and not error-corrected are generically feasible to simulate. This latter case includes all systems that exist in nature, and all existing instruments that observe these systems, and all computing engines that simulate these systems … so it is a pretty important practical case (that IMHO turns out to have its own mathematical charms 🙂 ).

    Examples 1-3 dovetail nicely, and they collectively suggest a well-posed topic for debate.

    Affirmed: The Big Crunch of science and technology is over, and a new epoch of exponential growth is underway.

    This exponential growth is being realized mainly in the context of a new generation of “Big Science and Engineering” projects, that are founded largely upon observation and simulation, with a healthy admixture of traditional theory and experiment, and strong elements of engineering design too.

    The field marks of this emerging paradigm of “Bigger Science” are massive observational databases, whose raw data is supplied by quantum-limited instruments, which are queried via well-validated simulation algorithms running on energy-efficient computers. These “Bigger Science” projects represent a new approach to coupling fundamental science to urgent societal needs (that’s where the money comes from).

    Your mileage may differ, but the main point of this post is simply to note that if we assume that Bohr’s Principle “The opposite of a great truth is another great truth” applies to Goodstein’s Big Crunch essay, then we are led to a set of conclusions that is very good news for young mathematicians, scientists, and engineers. 🙂

  82. Steve Says:


    Well this is a really important question to address.

    I love to teach as a Teaching Assistant, I am working on my phd in Physics at a reputed place. My question is:

    What are the odds of getting a position in a non phd granting college/master’s degree insitute.
    I love to teach, will be connected with my field and will have
    adequate freedom to pursue what I want to?

    Please address the above question as I think this would address huge fraction of people who love their phd subjects and want to keep pursuing it.


  83. John Sidles Says:

    Hi Steve! I was traveling, or I would have answered your post sooner. Here is one person’s answer to your important but difficult question:

    (1) Your dream is achievable.

    (2) It won’t happen spontaneously.

    (3) For practical help, read (e.g.) A PhD is not enough (your library will have a copy.

    (4) Also read E. O. Wilson’s autobiography Naturalist.

    (5) Then talk to three people (or more) who have succeeded in the arena where you wish to succeed. Respectfully ask them how they did it. Then, do the same.

    If you think this adds up to a mixture of common sense, careful planning, and hard work … yep, that’s the advice.

  84. Snake Pliskin Says:

    Steve –

    I have been wrestling with a long postdoctoral / non-TT stint exacerbated by a two-body problem. I am on the market right now and have interviewed at mix of PhD granting research departments, and some teaching colleges and “terminal master’s only” universities.

    I didn’t really enjoy my visits to the latter. You are lucky if you get one day a week for research at those places, and the classes that are taught tend to be lower-end intro and service classes. A professional master’s program increases the opportunities for more advanced classes, but the students are extremely vocational in mindset.

    Anyway, if that’s what you are into, I think that the reason that I got the interviews with these places was having a lot of teaching experience with good teaching evaluations.

    These places want instructors that can reach the students that are motivated but may not have the best background, and they want instructors that can teach across the departmental curriculum.

    The more courses from the core curriculum that you’ve taught (as the instructor, not as a TA), and the better your teaching evaluations, the better your shot at appealing to a teaching oriented institution.

    You are better off racking up your teaching experience while a grad student than running around taking a lot of VAP, lecturer, or adjunct positions.

  85. Gil Kalai Says:

    Contrary to Katz’ view I think that lack of high quality supervision, mentoring, and feedback are more serious problems for junior scientists (especially theoretician) than lack of independence.

    It is very hard to give advice to people on what to do in the future. But it is somewhat easier to remark on what was done in the past. Very recently, Carlos Mochon (#37 in Preskill list or so) talked in our local quantum computation seminar about his wonderful solution giving a protocol for weak coin flipping with arbitrary small bias. It ain’t get much better than this!

  86. Darren Says:

    The average IQ of the commenters in this thread (myself excluded) is probably around 200, but I see an insufficient appreciation of the concept of opportunity cost. A professor who has supervised many PhD candidates says: “As far as I am aware though … (my former students)are all to varying degrees living productive, fulfilling, and happy lives.” However, physics PhDs are smart people, and smart people often find a way to lead fulfilling lives, with or without PhDs. The question is – did a physics PhD contribute more to that fulfillment than some other alternative use of their time would have?

    Our host in the OP said something like: “a PhD won’t hurt”. As someone else has already pointed out, that isn’t good enough. A PhD takes several years of a very smart person’s time, and the true cost of the PhD is forgoing whatever else you could have done in that time, which might represent several hundred thousand dollars in earnings. There had better be a payoff (professional or spiritual) that you couldn’t have gotten otherwise.

    PS: love the active preview feature – neat!

  87. Scott Says:

    Darren: Presumably, the reason most people choose to go to grad school in the first place is that they expect to enjoy it more than a “real” job. In any case, I trust people to know their own preferences enough to be able to make that judgment for themselves. What they know less about is what their job prospects will be after grad school, so that’s what I tried to address in the post.

  88. John Sidles Says:

    When we teach our medical residents how to critically read the outcomes literature, we emphasize that the weakest outcome studies are those done by a single highly-skilled physician whose conclusion boils down to “All my patients do great.”

    These outcome studies are gravely flawed for several interlocking reasons. (1) The physician (often famous) is exceptionally skilled. (2) The patients (who have sought out this famous physician) are exceptionally motivated to report a good result. (3) The indications for treatment are murky … which allows the physician to select patients that (s)he intuitively expects will do well. (4) The criteria for medical success are determined ex post facto.

    The predictive value of such studies is nil. These flaws are summarized in the surgical aphorism “The best way to obtain reliably good results is to operate upon patients who don’t need surgery.”

    This is not to say that graduate study has no value … my own experience is that graduate study often has great value. Roughly speaking, this value seems to derive 1/3 from the student, 1/3 from the teacher(s), and 1/3 from the community within which the students and teachers are embedded. This is true not just at the graduate level, but at all levels of education.