Recently, I had a discussion whether physics is more in need of postdoc or faculty positions. Weird enough, the postdocs present suggested we need more postdocs. As far as I am concerned, the trend to run scientific research increasingly on postdoc positions, non-faculty short-term contracts that are in many cases supervised, is disastrous. I argued many times before short-term contracts favor short-term projects. As a result, ambitious, work-, and time-intensive projects suffer. Besides this, who wants to remain a postdoc forever? Thus, clearly more faculty positions are needed. But tenure?
Some faculty positions are nontenured. The Santa Fe Institute is a prominent example for fixed-term faculty. As you can read in Howard Burton's book, Perimeter Institute initially also didn't have tenure. Meanwhile we do have. The prime reason, I think, is that the sample of top-scientists who will be attracted by a non-tenured position is very limited. While that is a pragmatic reason to offer tenure, it is not an argument in principle.
So lets look at the arguments in principle. Science needs room to breathe. Tenure offers the necessary safety for researchers to work on controversial, risky, or unpopular topics. It gives them the time to run into a dead-end and start over again, without being immediately discarded as failure. It gives them the peace of mind to not worry about their peer's opinion. Safety is, without doubt, essential.
On the downside, safety invites idleness. The impossibility of getting fired doesn't improve neither responsibility nor quality of management or teaching. It is however in my experience the non-research duties that suffer. Peer pressure is sufficient to guarantee profs don't start twiddling thumbs once they are tenured. While tenure would give them the possibility to ignore their peer's ridicule, most take it very seriously. That's no surprise. In expert communities colleagues' approval is important.
In addition, as one of my fellow postdocs put it, a problem with tenure is you'll get stuck with the old guys after they've reached their expiration date. Considering the increasing number of grey hairs on my head, I find this argument borderline to inhuman. Yes, we age. We all do. Yes, with that we lose some abilities, though we might gain others. Throwing out people when they age to improve the output of a workplace creates, frankly, a system I don't want to be part of. The young and the old, they are part of our lives. Ruling out prescribed execution at 42, every one of us goes through these stages. The way to deal with it efficiently is to integrate people according to the stage of their lives.
What remains then is the question how much security is necessary and what time-scales are appropriate to judge on a research program. It is somewhat a mystery to me why academia fails to establish a functioning yet human job system. Let's take Stefan as an example. He decided not to take a postdoc position after his PhD and now works for a scientific publisher. After 2 years with that employer, his contract became permanent. Needless to say, that's not a contract of the do-what-you-want-we-can't-fire-you-anyway sort. It's just a job that can be continued as long as one does it well.
In a recent blogpost at TheScientist Is tenure worth saving?, a commenter summarized his/her problem
As a young researcher, I spend more and more of my time thinking on why I got into academia in the first place, and sometimes wishes there was a real alternative so I could get out. The problem is that I am now over 30, I haven't had the means (or the 'geographic security') to invest in a house or any of the other things that people I went to school with did 10-15 years ago. As a consequence I am highly educated, but have a very low financial status. This is a cause for stress. At the same time, I will be working on short time contracts for a long while yet, longer even if the administrators get their way. If there is no light at the end of the tunnel, all the smart people will disappear from academia and the universities will end up being schools. This is extremely worrying.
Letting people go after a couple of years without any particular reason as to their performance might increase the flux of ideas but it destroys continuity (and, for what it's worth, loyalty). In addition it favors people who are willing to postpone their life possibly until their late thirties or early forties. Which, needless to say, favors men.
In summary, the security tenure offers is an essential ingredient for scientific research. It is not necessarily the only option though. More long- but fixed-term positions, and renewable contracts have similar benefits and would help make academic research a more human career path.