As I commented over on DM's post, I personally would not have moved to Prodigal U from National Lab without the possibility of tenure. I took a large paycut to come to Prodigal U, plus I had already passed the probational period in my Federal job. In my field, the pay goes as industry >> government > academia. In order to attract top scientists, universities need to offer some perks instead of monetary compensation (and research freedom alone won't cut it--it is possible to get that elsewhere with higher pay). Tenure also provides protections for faculty to research controversial topics, to be outspoken about university governance when necessary, and to be active in the local community without reprisals. The same type of job security is available at unionized workplaces and to government employees (at least at the Federal level), so it certainly isn't unique to academia.
FSP discusses another (and more interesting to me) question--should an excellent teacher but mediocre researcher get tenure at a research university? In my opinion probably not. In my comment to FSP's post, I was willing to consider the rest of the file, but we all (at a research school) know that research is the most important piece of research/teaching/tenure. How much should teaching count? Education is an important piece of the university's mission, so I strongly support denying tenure to bad teachers. Once a professor hits adequate, it should be enough to get tenure at a research university.
The situation of the mediocre researcher /excellent teacher at a research school generates a lot more discussion than the reverse, since it seems pretty clear (to me anyway) that an excellent researcher/mediocre teacher is likely to get tenure. There are a few possible backstories for our mediocre researcher /excellent teacher at a research school now going up for tenure:
1. MR/ET tried their best to reach the standards for tenure at their school and missed. It always sucks when someone falls short of the goal. However, this is the risk anyone accepting a TT position takes. It is bad for the person applying for tenure (they will be denied and have to move on) and bad for the department (they lose the investment in time and money in the person denied tenure). My university can't really afford to spend 6-7 figures in startup every 6 years, so we try to be very picky at the time of hire, and the tenure denial rate is low. This is something I investigated before I accepted my position, so I would have some understanding of how the standards are applied. As much as it sucks to be denied tenure, there are many people who go on to good careers both in and out academia after a denial, so it isn't the end of the world for anyone concerned (though it might feel like it).
2. MR/ET cared more about teaching than research, and this was reflected in their relative efforts and results. This one I don't really understand. We don't get to rewrite our job descriptions after accepting the position. There are many types of colleges and universities. If someone is not happy with the mix of responsibilities at one, they should look for a position someplace where there is a better fit. There are plenty of people at research universities who also excel at teaching, but at a research school, research has to come first. This is something MR/ET should have known going in.
3. MR/ET got really bad advice pre-tenure. Not sure how much more I can say about that. It is really important to have more than one mentor. It is really important to see how tenure decisions work at your school, keeping in mind that the ultimate decision is not the departments'.
I think in many ways it is "easier" to quantify research output than teaching excellence. It seems that most teaching is evaluated primarily through student evaluations (this is the case at Prodigal U), with maybe a direct observation or two. This is a terrible way to evaluate teaching, since first, it can be gamed by a professor catering to student whims without actually teaching anything (student evaluations are strongly correlated with grades), and second, students don't always appreciate a good teacher during a course, especially for difficult or required courses. Sometimes, it is only after time has passed (and the student uses knowledge from the course) that they realize the prof was a good teacher after all. Even worse, there are some studies (can't find the links) that show that students are biased in evaluations against women, non-native English speakers, disabled professors, and professors from visible minority groups. Student evaluations should be a part of the process, but clearly not the only factor.
I've heard that non-research universities are counting research productivity more and more. I am not sure if that is true (it is out of my direct experience for one thing), but I suspect that "quantitative" measurement is at least part of the issue.