Letters of recommendation and letters of reference

I was thinking about posting on this subject already, and now Matt at The Consternation of Philosophy has done so, so I'll repost my comment there and say a bit more about it. Matt writes:

It's common knowledge that letters of recommendation for academic positions are generally inflated. This gives rise to some concern about how the letters are to be interpreted, as one must "discount" the praise for a candidate in some way in order to arrive at a more accurate assessment of his or her abilities. There is less concern, however, about the morality of this practice in general.

In the comments I respond:

Deception is certainly bad, I agree, but I don't know how much actual deception there is in academic letters of reference. I've read a lot of such letters recently and they appear to divide into two types. One, which I'll call letters of reference, describes the person in what seems to be an objective way, noting both good points and bad. The other, which I'll call letters of recommendation, only mentions good things. I don't see this as being dishonest, but it is very hard to compare two candidates when they have different types of letters being written about them. At the initial screening stage I think many search committees are looking for reasons to exclude a candidate from further consideration, so any negative comment might do real harm. When so many people are writing letters of recommendation, I think it's arguably immoral for someone to choose to write a letter of reference instead.

I guess I'm not really convinced that Matt's premise is right (although it might be). Or at least I wouldn't put things as he does. His analogy with grade inflation is useful, as is the implicit analogy with inflation generally. When prices go up this might be bad, but it isn't (necessarily) dishonest. There is a sense in which (some) goods just are worth their market price. A price is low if it is lower than the normal price and high if it is higher than the normal price, but there is no non-relative "true" price that somehow ought to be charged for something like a tomato or a can of beans. The monetary value of a can of beans just is what you can expect to get for it on the open market. (Compare "meaning is use.") And much the same goes for grades, I think. There just is no such thing as a true C that, because of grade inflation, would generally be given a B. If it's the kind of thing that would generally get a B, then it is a B. It is neither more honest nor helpfully trust-inspiring to give Cs to work that others would grade as B. The most helpful thing is if everyone applies the same standards. Or so I think.

I'm not bothered at all by letters that are liberal with words of praise. I expect that. If there is no specificity to go with it then it's pretty meaningless anyway. I can't quote examples, but a letter that just said, "This candidate is great. Really, really great," would do the candidate no good. One that gives examples of the candidate's greatness, though, or specifies some kind of ranking (e.g. the best graduate student I have worked with in the last x years at y university) might well help. If the writer is just lying then that's obviously bad, but otherwise this all seems well and good.

What I think is bad is the refusal to write a letter of recommendation (as opposed to one of reference). Why bring up bad points if they are not relevant? Sometimes there is a reason, of course. Perhaps the candidate has something odd about their CV that needs to be explained. But otherwise I think letters should contain nothing but praise. Then the least praised candidates will (on the whole) be the weakest. In my opinion, given that so many letters are purely letters of why the candidate is recommended (rather than what the candidate is like, good and bad), letter-writers ought to leave out irrelevant concerns (e.g. "the candidate's disability, which is otherwise not mentioned at all in the application, is barely noticeable and does affect her performance at all"), out-of-date concerns ("when I first saw her teach she did a terrible job, but she's much better now"), and outright criticism ("one thing he does badly is z"). Admittedly this last category is one I have mixed feelings about, but so few letters mention such things that I think it's unfair to the candidate to include them.