Megan's post has two parts: it begins with a long list of complaints, many of which have nothing to do with the tenure system per se, and others of which seem to forget that academia is a tournament with known rules that people enter into voluntarily.
Consider this bit:
At the end of the process, most of the aspirants do not have tenure; they have dropped out, or been dropped, at some point along the way. Meanwhile, the system has ripped up their lives in other ways. They've invested their whole youth, and are back on the job market near entry level at an age when most of their peers have spent ten years building up marketable skills. Many of them will have seen relationships ripped apart by the difficulties of finding not one, but two tenure-track jobs in the same area. Others will have invested their early thirties in a college town with no other industry, forcing them to move elsewhere to restart both their careers and their social lives. Or perhaps they string along adjuncting at near-poverty wages, unable to quite leave the academy that has abused them for so long.The fact that most people who try to become tenured professors do not succeed is common to many professions that operate on tournament models, such as professional athletics or entertainment or architecture. This has nothing to do with tenure per se and would not change with its abolition. Such professions also consume your youth. Yep. And that would not change either. Indeed, if scholars do their best work when they are young, again like athletes, that is surely optimal. And people might have to move! On no. Never happens in any other field of work. And how about those folks who stay on in non-tenure track positions. They do so voluntarily. So what? Lots of people like to teach at college. That is why the pay is not very good for positions that involve teaching and no research.
The second part of Megan's post is a more standard set of arguments that focus on the "tenure encourages innovative research" line of argument. As noted below, I think that argument is mainly compelling outside the US.
Tyler, on the other hand, is on the right track, though his post adds on some bits of unnecessary and distracting (and rather dismal!) metaphysics. Think about the counterfactual. Most firms do not fire very many workers at all, either in the direct sense or (what they used to do at Farrell's because it was cheaper) in the sense of making them miserable until they go away on their own. Academia fires lots of people. So the issue is not whether or not weak workers are let go. The issue is the extent to which being a good faculty member is a relatively stable characteristic or whether it changes a lot over time. If it is stable, trying to figure it out and then making one round of cuts will work well. If it is time-varying, this will work less well.
Two other features of academic life are worth noting and are relevant to the question. First, there are still strong incentives to produce even after tenure. Even in a world with modest inflation, a decade of no raises hurts your real earnings. Of course, this incentive is strongest at places like Michigan where faculty do not get routine cost-of-living increases. Beyond pay, there is still promotion to full professor and the broader, and likely more powerful, pressure that comes from the expectations of one's colleagues.
Second, faculty do many different things, which are generally grouped into research, teaching and service. Faculty whose research output declines for whatever reason should be doing more of the other things that faculty do. One of my former chairs at Western Ontario used to say something along the lines of that his job was to make sure that everyone contributed to the department in some way, and in his regime faculty who were not research active did lots of teaching and administrative work.
So, my answer is that tenure does not matter much. If you get rid of tenure, the counterfactual will still likely be an "up or out" system in which a major decision is made after a few years, with contracts after that point generally renewed for almost everyone who makes it past that initial evaluation. This is like many jobs with probationary periods, but in academia the probationary period has to be long in order to get a reasonably precise measure of research quality and quantity. Getting rid of tenure probably would trim the tails of the distribution of views expressed in academic research. Valuing that is tough. This is less of an issue in the US, with its many private universities of varying politics, than it would be in more homogeneous state-run systems like those in Canada and most of Europe.
As an aside, the biggest pool of teachers with tenure is in government-run K-12 education. If you think the arguments for tenure are weak at the university level ...
Addendum: orgtheory jumps on Megan McArdle as well, but from a different angle.