Skip to content

Life After Tenure

I wanted to dequeue one more post before the semester gets going for real. This is a collection of random observations about the contrast between being an untenured and tenured professor.

Tenure is an up-or-out process: either you get it or you lose your job. This is stressful for everyone. Getting tenure is excellent in that this source of stress is removed, but it is also a little bit confusing since we are suddenly forced to ask what exactly it is that we are doing with our lives now that—at long last—we aren’t living up to some random collection of external expectations. You don’t have to look too hard to find examples of people who have voluntarily left academia after getting tenure; my guess is that this effect is sometimes a factor.

Basically everyone who gets tenure, including me, finds him/herself continuing to get busier every year. Not only do the demands on our time increase, but we can no longer fall back on the convenient selfish excuse “sorry—can’t possibly do that time consuming task while I’m on the tenure track.” The constantly overloaded TODO list gets old, and it also makes it hard to get actual research done. It’s no coincidence that a lot of tenured professors rely on students to do all of the technical work associated with their research programs. Untenured professors not only have more time to do research but also may be reluctant to risk their careers by putting students on too many critical paths.

I used to knock off work between 11 and midnight, whereas I now generally quit between 10 and 11. Additionally, during many of my pre-tenure years I got very little exercise, which resulted in high blood pressure and various other problems; now I spend some time staying in shape. On the other hand, I don’t take as many days off now, probably due to more stuff going on in general and the kids being on a school schedule. Even so, I probably work fewer hours per day now than I used to.

I’m trying to write fewer papers now, and to make each paper a really solid one. This blog is part of that effort: it siphons off a good deal of writing energy. At some point leading up to tenure I realized that writing a mediocre paper makes no sense; it is almost as much work as writing a good one and it dilutes whatever reputation for myself that I may be managing to build up. A few of my more recent papers, most notably the Csmith one, are the result of a large amount of effort that would simply not have made sense to allocate to a single paper before tenure. I could have turned Csmith into a series of papers, but why? I can’t see that as being useful to anyone. There’s an avalanche of crappy papers and I would love to not be part of it. It is possible that I’m doing my PhD students a disservice by publishing less; I try to deal with this by being up-front with them about the academic track: if they want to get on it, they need to learn to write fairly rapidly and extremely well. I’m happy to work with them on that project, but I’m not going to push. The impetus needs to come from them.

When I got tenure, one of my plans was to avoid letting my teaching be compromised by research pressures. I’m afraid that I have not accomplished this goal. I do a pretty good job teaching, but no better than I used to. I do, however, feel more free to take risks in the classroom by trying teaching techniques that aren’t totally tried and true and by throwing out old assignments more often. This causes me to take a hit on teaching evaluations sometimes, which is OK.

Finally, I’d like to say that I now take more time to think and read, but I’m not sure that this is the case. Probably the main thing that I do differently along these lines is write blog posts, which is certainly useful as a way to motivate thinking. I read relatively few books and papers related to my research, except on occasional bursts where I need to learn some new research thread or subfield.

{ 10 } Comments

  1. LarsBerg | January 7, 2013 at 3:37 pm | Permalink

    My advisor (John Reppy) also is a full professor and had a similar conversation with me that he prefers to just publish occasional, major results in either a top conference or journal and that the onus would be on me – since I expressed an interest early on for an academic career – to push out more papers from the group in order to get enough pubs to hit the minimum bar to not be auto-tossed out by a hiring committee filter.

    That paper issue is a unifying thread I keep hearing over lunches with senior faculty and research lab members who come through for distinguished speaker talks. I’ve even had corporate lab members tell me they find themselves in the novel position of having to nudge the pre-tenure faculty they’re working with to think about the long-term goals of the project instead of just riding the “>= 2-per-year” incremental paper train.

    And that’s not even to get started on the number of systems papers that describe incremental results that _aren’t even actually integrated_ with the project, but just occurred as a piece of hacked test work in a branch, generated some numbers, and then were abandoned as complete once the paper mill ran dry.

  2. regehr | January 7, 2013 at 5:22 pm | Permalink

    Hi Lars, for the record I’m just an associate professor– so maybe I have a hoop left to jump through after all. It’s great to hear that the paper issue is a unifying thread among people you’ve talked to!

    I think it’s fine for papers to stem from code that doesn’t make it into a code base, there are plenty of research results whose time has not yet come for more mainstream adoption. On the other hand, of course if people are doing that consciously as part of a paper mill, that’s no good.

    Doing research is tricky. I think every year I become less sure about what research is (or maybe just about what good research is).

  3. LarsBerg | January 7, 2013 at 6:10 pm | Permalink

    Sorry if I came across as overly negative! Possibly a side effect of being in the final push for numbers (and working code…) around my thesis. My only complaint is when optimizations or implementation strategies (particularly in compilers) are presented as a superior approach based on a bunch of benchmark results but the paper doesn’t mention they don’t handle the whole language or aren’t robust w.r.t. the full test suite of the compiler, and so are not actually implemented except in a dead branch. It’s very common and it makes it hard, as a fellow compiler-writer, to know if there’s a real result there or there’s still more research to be done before it’s reusable. Especially since so much work in a compiler and runtime is finding good heuristics and tuning parameters.

    I also don’t know what good research is, but I try to make sure that at least one thing I’m working on is something that people who would _never_ cite me still find interesting to talk about. That probably doesn’t make it good, but at least it’s not just narrowly interesting to my little “statically typed functional programming languages” echo chamber.

  4. Daniel Lemire | January 7, 2013 at 11:00 pm | Permalink

    I’m trying to write fewer papers now, and to make each paper a really solid one.

    I’m similarly motivated, but it is not about my reputation. I don’t think that writing lots of hasty papers necessarily harms your reputation… unless they are really bad. We are rather forgiving especially since your colleagues are unlikely to ever read you. Plus, there is some evidence that quantity and quality are not related in a naive way: it is not because I choose to write fewer papers that I’ll write better papers.

    However, I enjoy a lot of things about research beside writing the research paper itself.

    I feel that we lost sight of the reason why we do research…

    Another long term trend that I see in my research is that I tend to try to be more “useful” over time. This does not mean necessarily publishing at better place, maybe it can be the opposite actually… but I feel increasingly attractive the idea that I might publish something that an industry engineer might read and find interesting and useful.

    I don’t find enjoyable the “puzzle” aspect of research anymore… (e.g., prove that the problem is NP-hard…) Maybe because I don’t care as much as I used to about appearing to be smart and competent. If some people think I’m an idiot, then so what? I’ll gladly pay this price in exchange for the feeling of being useful.

  5. Adrien Lamarque | January 9, 2013 at 7:02 am | Permalink

    > I do a pretty good job teaching, but no better than I used to.

    I am surprised that you did not mention the Udacity course on software testing that you gave and that got me interested on the subject — it’s the reason why I am regular follower of your blog. You’ve contributed your share to this innovative movement that is massive online education, which is a pretty nice achievement in itself. Perhaps you’re not satisfied with the teaching on your “real life” students, but on behalf of them, I can say that your virtual students are quite satisfied.

  6. regehr | January 10, 2013 at 1:17 pm | Permalink

    Adrien, thanks for the note, I really love to hear this.

    I’m still pretty confused about how MOOCs and my regular teaching job interact, which is why I didn’t mention it.

    By the way a bunch of the content from my testing class is going into a book that I’m co-authoring that I’m pretty excited about. But it’s just getting started.

  7. Evan Jones | January 14, 2013 at 11:17 am | Permalink

    As a PhD graduate now outside of academia, I fully support fewer, better papers, and would love to see this change in academic CS. The emphasis on paper quantity is one of the reasons I didn’t find academia very attractive. Publishing a paper seems to be barely useful in my field (databases/systems), particularly when it looks at some micro-optimization that may only be useful in very peculiar circumstances (I have at least one paper like this). However, there are the occasional papers that are truly thought provoking or useful.

    The challenge is how can we work towards a system where this is the norm? Can we change the hiring/promotion system so that papers are not the primary metric?

  8. regehr | January 14, 2013 at 11:33 am | Permalink

    Hi Evan, I’m not sure the system can be easily fixed– the incentives are all wrong. The only reasonable way forward is for individuals to either decide they don’t like the system or else to sort of work within the system in minor ways. For example, I’m chairing one of our faculty search committees and I am making sure we don’t even interview people who like to crank out tons of incremental papers.

  9. Tyler Nichols | January 14, 2013 at 12:00 pm | Permalink

    > “I don’t find enjoyable the “puzzle” aspect of research anymore… (e.g., prove that the problem is NP-hard…) Maybe because I don’t care as much as I used to about appearing to be smart and competent. If some people think I’m an idiot, then so what? I’ll gladly pay this price in exchange for the feeling of being useful.”

    I find this comment precisely states a feeling that I’ve developed over the past few years. Although I’m an undergraduate and not in the same position in regards to research, I still find that much of the work force (and undergraduates) are continually fighting to prove their intellectual dominance through clever solutions to “puzzles”. We are continually trying to one up our colleagues. While I sometimes find “puzzles” interesting I find life is much more enjoyable when I can be useful. In addition being clever just for the sake of being clever has side effects that can negatively effect programs we write.

  10. Justin | January 15, 2013 at 4:38 am | Permalink

    Hi John, don’t know whether it’s worth posting here but Embedded in Academia has been mentioned in the list of programming and software blogs here-> http://www.talkora.com/technology/List-of-programming-and-software-development-blogs_105 (look for entry #23 in the list)