I was on sabbatical last year, working at IBM Watson, and also had the chance to make short visits to MSR New England, Google NY, and AT&T Labs. Having returned to the University of Maryland only recently, the topic is fresh in my mind. I want to stress that the comments below are based only on my experiences and should not be over-generalized; the comments about research labs reflect my limited exposure to the labs I visited (and the specific people I know there), while the comments about academia pertain, to some extent, only to UMD.
Let’s start with some of the advantages of industry. Perhaps the biggest advantage is that, believe it or not, you can actually devote more time to research working at a research lab. As a faculty member there are constant demands on your time, not just teaching (though that dominates) but also faculty meetings, internal committee work, and — let’s not forget — overhead due to applying and working on grants. Although I was just as busy working at IBM as I am at Maryland, I somehow found the atmosphere more relaxing, and less pressured, at IBM.
Another advantage of industry is that you will likely have more people working in your immediate research area. Given that you were hired at a lab, that lab probably has at least one other person in your area, if not an entire group. Most likely, this will not be true at the university where you are hired. In fact, there may not be anyone in the geographic area working on the same topic you are working on.
Ahh — but you can collaborate with people in other areas! Perhaps, but I have found such collaborations happen more frequently in industry than in academia.
Let’s talk about the environment. The physical surroundings of the labs I have visited range from pleasant (AT&T, IBM) to places I wouldn’t mind going on vacation (MSR, Google — and let me single out the cafeteria at Google for being nirvana). While there are some universities with fantastic facilities, Maryland is not one of them.
Salaries are generally higher in industry, though see below for another take on that. In particular, for me the pressure to get grants to pay my summer salary is enormous and all-consuming.
Finally, there is (to me, at least) something attractive about working on a problem and having someone actually care about the solution. The extent to which this is true depends on the lab (e.g., this is much more true for Google than for MSR), and of course some academics are able to tie into industry enough for this to happen for them as well.
So what are the upsides of academia? Tenure, I think, is not one of them. As Michael points out, research labs are remarkably stable in comparison to any job other than academia. Moreover, even when there is turnover in industry people tend to land a good job elsewhere. On the other side of things, tenure is not all it’s cracked up to be. Departments can effectively fire someone by not giving them a raise for 20 years, or by assigning them to teach 5 classes a year. And that’s not to mention the fact that departments, or even schools, can theoretically shut down.
One definite advantage of academia is that you really can work on whatever you want. Although some might claim this is also true in industry, it really isn’t. If I wanted to, I could start working on computer architecture or even English literature. In industry you can generally work on whatever problems you want in a particular area, but it would be hard to completely shift areas. (On the other hand, switching areas in academia might marginalize you in your department; see the previous paragraph.)
Another advantage I don’t usually hear about is that academia gives you an opportunity to do many different things. I am involved now in several consulting jobs that would not really be open to me if I worked at a research lab. I am applying to be a member of an advisory board, something that would again be difficult in industry. Writing a book would have been more difficult. And in industry I would never have had the opportunity to be involved in the CSSG.
Another potential advantage is geographical. If you want to work at a US-based research lab in TCS you are basically limited to the Bay area, NY, or Boston. Nothing wrong with any of those places, but some people prefer to live elsewhere. More to the point, all those places are very expensive. I contend — in fact, I’ve done the math — that one’s effective salary can be significantly higher in academia then in industry simply because of the location. (And one can supplement one’s income by doing consulting, something not really possible in industry.)