Jun 012014

[spoiler alert:] It was a two-body problem/opportunity. But many people have asked me about it and so I thought I’d pitch in my 2 cents about academia vs. industry (which will hopefully not be too biased since really I like both and the decision is mostly independent of their merits). Of course, opinions are strictly my own.

[expand title="First, let me tell you how I got to academia and back (click to expand)" startwrap="" endwrap="" ]
I worked on my PhD in Applied Physics at Stanford from 1997 to 2001. I was only on campus for classes, however, as I was an advisee of Bernardo Huberman at Xerox PARC (Bernardo is a consulting professor for Stanford). As I was about to graduate, Bernardo offered me a job at HP Labs, where he was moving to, and I stayed working in this group until 2005, when Mark Newman forwarded me a job posting from the University of Michigan. I wrote back ‘There is no way I could be a professor’. He wrote back that he thought I’d be good at it. At the time I only knew of Mark (or rather ‘Newman’ as I called him then) through his work, which I thought was brilliant, and so I believed what he said (since getting to know him better, I know his work is impeccable but that he can be wrong about other matters). Regardless, I applied, I got the job offer, but then spoke to my husband TJ and said ‘But there is no way we could move to Michigan’. Being the great guy that he is, he said, we’ll work it out. We did. TJ took a research job at Ford and I started my tenure track job at Michigan (School of Information/Center for the Study of Complex Systems/and eventually EECS). 6 years later we were back in California for my sabbatical (I was splitting my time between Berkeley and Facebook). When my sabbatical was over, UofM permitted me to extend my leave and I stayed on at Facebook while TJ first led Ford’s Silicon Valley Research Lab and then switched to a startup. This month my leave expires and I will transition to being an Adjunct, which will permit me to keep teaching for UofM via Coursera, and will continue my full-time job at Facebook. And that is the story of how I became a post-tenure drop-out.[/expand]

Here’s what I think about academia:
[expand title="You will have great colleagues but you will have to work hard to talk to them or work with them." startwrap="" endwrap=""]
(Update: After seeing this post, many of my SI colleagues chimed in to correct that the social environment at SI is thriving, and I believe them (and also already mentioned that the new digs do a lot to help this). So I stand corrected — if you are at SI you’ll have lots of opportunities to talk and work with your colleagues!).
At PARC, HP Labs, and now Facebook, if you need(ed) to talk to someone you worked with regularly, you walk(ed) over to their office/cubicle/desk. Most days you can even gather up a small group to go to lunch together. In my naïveté I expected something similar at UofM, especially after having such nice long chats with everyone when I first interviewed and then on my return visit. However, with some exceptions, most faculty were in hiding, either trying to get work done at home or in their “other” office, or working behind firmly closed doors. A few weeks in I finally spotted the nice colleague across the hall in his office. I rushed over to say hello. He blocked the doorway, hurriedly said his welcome and then explained that he had to do some work. Around the same time I got feedback via email from the chair of the curriculum committee on a new course I had proposed. Since I had never taught (not even been a TA) I asked to meet in person to understand the feedback better. He said he would have time in 3 weeks. Once I started teaching the following semester, I behaved exactly as they had, keeping my office door firmly shut. Puzzlingly, some of the most prolific faculty, e.g. the aforementioned Mark Newman, leave their door regularly ajar and if you knock will ask you to come in and chat. How they accomplish that, I do not know.

Over time I gained some regular collaborators, and enjoyed working with them immensely. But others I would only chat with e.g. when we were on the same thesis committee. At some point I realized that I most cherished the 10 minutes when the student has defended and is asked the leave the room. Then the committee discusses what the student could have done (but didn’t) and these great ideas start flying around. But then the student is called back in before they freak out, and you realize that the next time you’ll be in a room with these people having such a great time discussing research will be the next thesis defense. I should say that especially the School of Information (SI) @ UofM was very cognizant of these issues, and the offices + cubicles in the new building were designed to facilitate collaboration and interaction. But I visited many other universities when giving guest lectures and for the most part saw closed faculty doors, or was regularly told by the people I was visiting that they came to campus just to hear my talk or meet with me. Actually, outside visitors (guest speakers or job candidates) ironically (if i’m using the term correctly) present a great opportunity to talk to people in your *own* school/dept.[/expand]

[expand title="Teaching is the most rewarding thing, if you can control yourself, and even if you can't" startwrap="" endwrap=""]
Near the end of my stay at HP Labs I was getting a bit depressed. Even though I was writing the papers that would eventually be my most cited, the days seemed to drag on. I wasn’t sure about the worth of my work, what exactly had I accomplished? Once I started teaching, all of that was dispelled. During every lecture/lab, I knew from the feedback that I was imparting some knowledge on others, knowledge that I was excited about and felt was important. It also doesn’t hurt that students are often eager to tell you when they go on to use those skills elsewhere. It just feels good, very good, both in the short and long run.
But it can also consume all of your time, and not everyone will be grateful (understatement). I knew that the most beloved courses were often ones where the instructor puts up some slides, discusses some concepts excitedly with the class, and then grades easily. But this wasn’t how I learned. In my undergrad days I learned very little in lecture. I learned by solving problem sets. So for my classes I would spend hours and hours cleaning data sets or setting up agent based models or writing data-fetching scripts the students could modify. As a result, in my stats class they could figure out the relationship between braces/math team/cheerleading and having a boyfriend, the relationships between age and happiness/sleep, etc. etc. In my social network analysis class they could simulate many of the processes at the slide of a slider and the click of a button. In the agent based modeling class, I wrote shells for the students’ agents to compete either with random payoff matrices or in a simulated stock trading market. It was really fun (especially for me), but it was also a lot of work for me, and for the students, which of course wasn’t always appreciated. And I had very little self control. Even though I taught stats several years in a row, when a new data set came out, e.g. a Pew survey on attitudes toward internet privacy, I couldn’t help but make a new assignment. Or I’d ask colleagues to give me data from their studies to make the course content more relevant. What was worse, because I was most motivated to do all of this the night before, the material was less organized than it ideally should have been..
In any case, I’m grateful that UofM will let me continue to teach for them through Coursera, because I do miss teaching.

There is a different kind of teaching, namely mentoring graduate students. This can be rewarding (as you see the student mature and find their own footing in the research world and also many of them are also just so damn bright and a joy to work with) and frustrating as well (as you find yourself taking the role of copy editor rather than getting to do the fun parts of the research yourself). The responsibility of it can be frightening. Here is a person whose future career trajectory depends on you: not just whether they will finish their PhD, but whether you will help them write the kinds of papers and will write the kinds of letters for them that will get them an academic job.[/expand]

[expand title="You get to do whatever work you want, on evenings and weekends" startwrap="" endwrap=""]
It’s unsurprising that you can fill all your time with teaching. But that is only part of it. There are a myriad tasks that will happily fill your academic day, very few of them having to do with actual research. There is the committee work, but it is bounded. However, once your work is out there enough, you will get constant requests to review both papers and proposals. The one year I counted, I figured out that I had reviewed 80 papers + proposals. 80!! Even turning down that volume of requests is a chore. Whenever you think you finally have an hour or two to do some work you’d like to do, more than likely something like an overdue review or recommendation letter will stand in your way. Funny story: a colleague at Facebook had asked a well-established academic to collaborate. Reportedly, the academic laughed and said: ‘Do you know how long I’ve been in this business? All I do all day is write recommendation letters’.

Now, probably for most people, recommendation letters are no big deal. I don’t know if you’ve ever read a recommendation letter by Jon Kleinberg. Jon, as you probably know, is a recognized genius, beloved teacher, and (as I now know) a generous collaborator. You might think that next to all these achievements, he would have little time to write eloquent, 3-4 page letters, that not only describe all the best qualities of the students and their papers, but even give better summaries of the students’ independent work than the student gave. But no, he does write those letters. I, on the other hand, tend to sweat over these. I procrastinate as long as possible, then spend an entire evening on one, only to have to increase the font size so that it spills over to page 2. Don’t even ask about tenure letters, those take considerably longer. Almost-funny story: at some point UofM changed their student travel grant form to an online format. Whereas before, there were two lines on the paper where you needed to scribble why the student should get to travel to the conference, e.g. “has paper, needs to present”, now you were asked to attach a PDF of the recommendation letter. I spent some quality time writing these letters, until I finally checked that it’s ok to attach a PDF saying “has paper, needs to present”. The time I really realized how time consuming letter writing is was shortly after my son’s birth one October. In November and December the requests flooded in, and the only work I managed to get done was writing the letters.[/expand]

[expand title="You are your own agent" startwrap="" endwrap=""]
As a professor, sure, you have a boss. They might ask you to do things, such as teach particular courses, get more funding, join committees. But once you do those things, you pretty much rule your own world. You get teaching ratings, and your course proposals are reviewed by a committee, but once you’re in the classroom, you run the show. You pick the research you work on, and the recognition you seek is primarily external, since this is the thing that ultimately counts. You publish what you like. You spend your grant money as you please. In contrast, in industry, you have a boss. You check in regularly with your boss. Your boss might be very cool and very supportive (as has been the case with all of mine), but you are always thinking what they’ll think of your work and whether they’ll approve. You also have to get your publications approved (Facebook is particularly cool about this, but it is not necessarily the case everywhere in industry and at all times). Though in research labs, no one is checking your hours and you can work from home, in academia you can disappear for days, and as long as you’ve clued in your students and other people who depend on you, people will think it perfectly normal.

It’s all about you and your work
Whatever you do in academia, it’s about you and your work. You travel around, you talk about it. You write about it. Even grant writing I think is wonderfully straightforward: you apply to do some work you want to do, (eventually) you get the money, you do it. Yes, you are part of a this larger thing, your university, and you represent them when you go around, but the affiliation is secondary. You’d still be doing the same work if you were elsewhere. In contrast, when you are part of a company, you are a part of it. You represent it at conferences and you represent it to friends and to strangers. When I worked at HP Labs I would invariably get the same question after every talk: “Why does HP care about social networks?” (this was back in ~2003, of course everyone these days cares about social networks, right?). In academia the external recognition is also mostly a direct consequence of your efforts: if you write a good paper you will get cited, or you might receive awards or more offers of invited lectures. If other researchers at your university do something really cool, you might feel some pride, but ultimately it has not that much to do with you. In industry, the internal recognition mechanisms are typically very good and rather rewarding. But externally, you are somehow responsible not just for your own work, but for whatever experiences other people have with aspects of the product(s) you have never worked on. It’s a bit confusing.

On the plus side for industry, you are party of a big, mostly-efficient machine. There is a chance that your efforts will be for naught, but also there’s a chance that because of the existing machinery and good will of your coworkers, your efforts will multiply and have considerable impact that will be experienced by many people. That is, getting something “out there” for real is easier (though of course in academia many people license their technology or spin out startups and make the world a better place in these and other ways, e.g. my friend Amy Herr).[/expand]

[expand title=" It's not just the things you don't want to do that are the problem, it's the overabundance of things you do want to do" startwrap="" endwrap=""]
Seminars: Back in my days at Xerox PARC and HP Labs (and now at Facebook), I’d go to 1-2 seminars a week. On special occasions (e.g. Stephen Hawking or Persi Diaconis giving a talk), you go over to Stanford and make an event of it. The rest of the time you go about your work. At UofM, there are not only departmental seminars in a great many top-tier departments, but also many relevant interdisciplinary seminars, all somewhat relevant to your area of research. If you are unlucky enough to work in something like “Network Science”, there’s a good chance there are relevant talks during the same week in the CS, complex systems, econ, sociology, political science, public health and some interdisciplinary seminars, including the one you were encouraged to run in your first year as faculty (which is a major time sink, and if you, like me, are awful at recruiting student and staff resources to manage these for you and also forgot to ask for teaching credit for it). This doesn’t even get into the hiring season, when even more speakers come and you are asked (or ask) to meet with them.

Multi-university collaborations: There are some findings correlating authors being from multiple universities and the impact of the work. I’m still not convinced that it is actually the multi-university aspect that matters. I think it’s more that if you have exceptional scientists (who are more likely to be in a high energy state) and who have good ideas (which are more likely to be appealing outside of their university), you naturally end up with high impact work bridging universities. However, many of the funding agencies took this to mean that it’s very important to bring 5 universities together at once all working on the same thing. Although I like my colleagues at other universities very much, in practice this generated more in-person-travel-requiring meetings, and weekly or bi-weekly conference calls, and still mostly single-university output. At times fun, but I’m not sure the extra coordination expense is worth it.

During my time at SI, a lot of exciting developments were going on. First there were specialization “clusters”, then specializations, then different specializations, then a joint undergrad minor/specialization, then an undergrad major, then a joint program with public health, then lots of new courses (because as you can imagine the field is very rapidly evolving). Not to mention hiring to keep up with all these developments. Going back to teaching, this also meant that most courses did not yet have text books and that the material was continuously evolving (a social media course is certainly outdated just a year later). This isn’t the situation where you prepare some lecture notes from a textbook, assign problems from the textbook (adding a few of your own) and then teach the same thing for many years (which is what I imagine my undergrad physics professors did for the most part).

Academic output is out of control. I think Higgs was quoted as saying that he would never have gotten tenure by today’s standards because he didn’t publish enough papers. I think I wouldn’t have gotten a PhD by today’s standards either, or maybe even gotten into grad school. I did some research as an undergrad, but none that resulted in publications (or maybe just one). I published ~4 papers during my 4-year PhD (and these weren’t slave-away-in-a-lab-for-2-years-to-report-some-new-experimental-results kinds of papers, either). These days I’m not sure how a student would fare with such a poor publication record! We are writing more and more, which of course then produces more content to review as well.[/expand]

[expand title="Travel is fun until it's not" startwrap="" endwrap=""]
I have the good fortune of working in a blossoming field(s). As a result there are many, many meetings to go to. First, the conferences: WWW, Collective Intelligence, ICWSM, NetSci, WebSci, CSCW, several workshops on complex networks, “schools”, workshops on social media, crowdsourcing, economics and networks, computational social science, social computing, etc., etc. Then there are invited talks at universities, which are appealing because of the aforementioned opportunity to get some 1:1 time with interesting people. Then there are the NSF panels, at least 1-2 year because it would seem unwise to say no to people who hold the purse strings, and besides, there is just now a new interdisciplinary funding area where you should definitely have a say. Then the in-person meetings for the multi-university funded projects. All of this can easily add up to 1-2 trips/month, and some faculty I know actually do more. At first this was all very exciting, exploring new places, meeting new people. But then I stopped sleeping so well. It wasn’t just jet lag. No matter where I traveled, I’d wake up at 4 in the morning and go through meetings half-asleep. I’ve always been horrible about both booking travel and getting receipts reimbursed. I imagine that in some non-existent prior times one would have staff to help out with this, but since it’s all now online, we all do it ourselves, more or less (my case) competently.

I’ve been told about the virtues of saying “no”. You don’t have to do all this travel. However, as I mentioned, I was organizing a seminar and also invited people to do all sorts of things, e.g. join a program committee, keynote at conferences, etc. And you know what? Almost none of them say “no”. So this “no” thing might be a myth we like to believe and aspire to. However, in industry, it’s relatively easier to say no, you have other things to do.

Then there is the question of whether any of the travel is necessary. Someone told me about 10 years ago that Steven Strogatz doesn’t travel. Indeed I don’t think I’ve ever run into him at a meeting. However, now looking at his webpage, I see a calendar of upcoming speaking engagements and contact info if one would like to book him for one. Hmmm. Regardless, up to about 5 minutes ago, I thought that he had it figured out. He stayed home, wrote influential papers and books, and got at least some of the recognition that was due. At the same time, I spent about a year going around different places trying to shop a couple of papers I was very into to (one on trust, the other on focus and productivity). People clapped and said ‘interesting talk’, but the work didn’t get cited. Perhaps like Strogatz I should have stayed home. Perhaps then I’d be contributing to a New York Times column and making guest appearances on 99% Invisible, and making millions of people care more about math than they had before. Just kidding. [/expand]

[expand title="Technology is not (yet) your friend" startwrap="" endwrap=""]
Email is an enemy in so many ways: students emailing for help, random people emailing for help or with their papers, foreign funding agencies asking you to evaluate proposals for them, Israeli faculty asking for another committee member for their students, freelance journalists asking you to write articles for them that they may or may not get published, startups asking for advice etc. etc. At the same time office hours are fairly sparsely populated. I’d poll students to find the best time to have office hours and still many office hours went by with hardly anyone coming by. But I got lots of email. I tried to deflect most of it toward Google Groups where students could resolve most things on their own, but it only helped so much. Email is just too easy to send. If people had to write, call, or visit in person, the requests would go down to a small trickle.

I tried to use new teaching technology as much as possible: online forums, LectureTools (featuring such awesome features as saving each of your slides as an image, then manually shuffling them on the site, then inserting a quiz, which was supposed to be engaging for students, all in under 1000 mouse clicks or less),etc. But each of these was more work, not less. Just once I prepared a lecture by writing out notes by hand and then writing things out on the blackboard. It was a bit nerve-wracking to be so exposed and not hide behind polished slides, but it took a fraction of the time it normally took when I had to prepare slides, print them (though the printing was just a temporary phase, now I think no one wants paper any more), link them on the syllabus, upload them to their corresponding online folder, after also generating a pdf version, etc.[/expand]

[expand title="Balancing work and life is difficult, unless work is life" startwrap="" endwrap=""]
You may have noticed from the above rants that life in academia is a lot of work, nay, that it is work. I’ve also contemplated the required hours and the superhumans who do them. There were so many weekends where I had made plans and had to sheepishly renege once I realized how much grading I had to do, or that there was a paper deadline, or that I hadn’t prepped any of the coursework for the following week ahead of a trip out of town. Eventually I just didn’t make any weekend getaway plans during the school year. And I’m certain it wasn’t just me. A few of us female faculty would every once in a while try to coordinate a super-efficient shopping trip: we’d all pile into one car and go shopping for cheap veggies and cheap clothes and be back in a couple of hours. But we’d have to plan weeks and weeks out — once some of the teaching, grant proposal deadlines, and out-of-town travel were out of the way. And then still new things would crop up and only some of us could make it.

When I worked at HP Labs I had hobbies. I picked up welding and machining at nearby community colleges. I went sailing, kayaking, and wind surfing. Many weekends I’d go on hikes. I gardened. Now at Facebook I’ve gotten into woodworking, and many weekends I go camping with family. When I started my faculty job, I did nothing but work (except for some gardening in the summer). Speaking of summers, I imagined that I’d be able to go for months to Croatia. Go to the beach in the mornings, work in the afternoons and evenings. I was going to get away with something that most Americans do not — summers on the beach just as I had experienced them in my childhood. The reality was rather different. There were many conferences/meetings in the summer, and I tried to squeeze in the “vacation” between different trips. It had to be around the 4th of July because of Ford’s mandatory week off then. I had maybe 3 weeks I could spend in Croatia, so I then resolved to actually fully take vacation then. Unfortunately, this always coincided with some NSF deadline (e.g. the CAREER or CISE application), and of course every graduating student just had to defend right then, despite the fact that they really can do that any other time of year too. Sure, I’d go to the beach, but I knew I had hundreds of pages to review, or some paper to revise, or something else hanging over my head. It wasn’t restful, it was stressful.

And speaking of stress, I’m not a very fun person when stressed. You know how there are happy drunks and mean drunks? And there are some people who are really good with stress. Take Michael Bernstein for example. You might meet him after he has had 3 hours of sleep in the past 3 days because of multiple deadlines, and he’ll still talk to you excitedly about more research to do. But I’m a grumpy stressed-out person. I’m impatient, morose, and running to the bathroom often. On the plus side, I found myself not having too many existential crises, there were always more proximate crises/deadlines to worry about. And so the years passed, arguably some of my prime years, and they held a sense of excitement: publishing papers, interacting with the top people in my field both at UofM and at conferences, teaching hundreds of students. But when I look back my fondest memories aren’t really of those academic accomplishments, they are of spending a couple of lazy days with friends on a lake or river. The same is true of my years at Caltech. I am glad I kept up a crazy schedule of a max course-load and two majors, and I feel quite proud of having survived that and will point it out to anyone who will listen (wait, am I bragging again about Caltech? Oops). But the fondest memories I have of that period are of the pre-academic year backpacking trips, sitting around in the lounge, participating in some of the pranks and traditions, and roads trips across California, Arizona and Utah.[/expand]

[expand title="Being an academic with kids offers flexibility, but the time constraints do not go away" startwrap="" endwrap=""]
I heard from multiple faculty that having kids is easier in some ways if you’re an academic. Because you don’t have to clock in, you can participate in your child’s activities. One professor regularly coached his kids’ after-school sports games, another said she had been able to attend most of her son’s basketball games. At the same time, you can bet that those same individuals would either wake up early to get work done before their kids awoke, or worked late into the night after their kids went to sleep. Just try and set up a conference call with someone with kids about a paper you are writing together. I bet it will be 9 or 10 pm their time, which they will find to be very convenient because their kids will have gone to bed.

There is also just the sheer exhaustion of a demanding job and childcare. Many faculty hire help to alleviate some of these demands, but aside form full-time daycare, we didn’t quite figure out how to take advantage of these. Many hours at work were followed by many hours of complete attention to child. In addition, because I was released from teaching for one semester (my maternity leave), I lost two of the courses I had designed from scratch, and had to prep two new courses when my kid was less than 1 year old, while dealing with both his and my colds/sniffles which we were getting thanks to his exposure at daycare. I of course ended up overdoing it as far as prep was concerned (see here), and even though I enjoyed the new courses, let’s just say that the sabbatical came not a month too soon. Someone mentioned to me once that he prefers a department to hire women who already have kids, because then you know that they can handle it. I’m sure it was meant in a very positive way, i.e. women who have kids and are applying for tenure-track jobs have already demonstrated great skill in balancing work and childcare. But to me this suggested something else entirely: perhaps many women actually can’t handle both kids and a tenure-track job, and maybe I was one of them? (Edit: I think we can do whatever we want to do, and I never had too many doubts… ))[/expand]

In industry there is certainly both the opportunity and the incentive to work hard. I’ve seen many parents’ and non-parents’ careers thrive in both environments. I think in both academia and industry much of the work and over-work is self-inflicted. We are just those kinds of people. We love to work, we are passionate about the questions we pursue and we always want to do better.
So how does one decide between the two career tracks? The choices are rather different. And it is possible, but not easy, to switch between the two. So if you are still trying to decide, good luck. If you have already decided, then carry on :) .

© 2011 ladamic's blog Suffusion theme by Sayontan Sinha