What’s Wrong With Soliciting Letters of Recommendation?


A special guest post by Philosopher Michael Huemer (University of Colorado)

I think the practice of soliciting letters of recommendation for academic positions is both foolish and immoral.

Factual background:

(a) Most jobs and fellowships in the academic world include “letters of recommendation” as an application requirement (see https://philjobs.org/). Usually, you have to have three of these letters sent to the job search committee or fellowship selection committee.

(b) If you want to apply for that position, you call up three (or more) of your friends in the profession and ask them to write letters about you. Caveat: the letter writers must be professors, and if you yourself are a professor, then they must be of at least the same rank as yourself.

(c) The letter-writer then writes an inflated letter about how great you are, with the degree of inflation depending on how the writer feels about you, as well as how comfortable that particular writer is with dishonesty. Often, they write in vague generalities. When they give specifics, it usually consists of describing and praising some particular work of yours that the committee could simply read and evaluate for themselves.

(d) The selection committee then reads these letters and scrutinizes them for any sign of less-than-maximal enthusiasm (that being the standard for a “bad” letter). Sometimes the committee searches for ways of interpreting particular words of phrases as some sort of indirect speech act with a hidden, negative meaning.

(e) The committee also takes into account the level of fame and the general status in the profession of the letter-writer. A letter from a big shot at a top-ten school counts for much more than a letter from a low-status professor. Of course, you can only get that letter if you are friends with the big shot or studied with the big shot.

(f) The theory behind this practice is that the committee doesn’t have time to familiarize itself with the work of all the hundreds of candidates who are applying, so they rely on an outside person who is already familiar with that work. Thus, the letter writer is, in theory, acting as the *agent of* the committee, with the task of helping the committee select the best person. In reality, most letter writers conceive themselves as agents of *the candidate*, with the task of helping their friend (the candidate) get the position that the friend wants.

Now, here is what I think is wrong with this whole practice:

(1) The probative value of “letters of recommendation” is approximately zero, due to points (c) and (f) above. It’s basically like asking, “Tell us whom you’re friends with, and let’s see who can get their friends to say the most exaggerated things about them.”

(2) It contributes to a system in which the most coveted, scarce goods are distributed according to personal connections, rather than merit.

(3) It creates incentives for people to curry favor with the “big names” in the profession and thereby causes more of this to happen, due to point (e). For example, if one is at a conference, the incentive would be to chat up the high-status philosophers at the conference, rather than, say, talking to students or other low-status people.

(4) It unfairly rewards people who are unashamed about bothering their friends for favors, while systematically disadvantaging people who are considerate of their friends’ time, or who tend to make friends with low-status people, or who tend to have unusually honest friends.

(5) It is manipulative. It recruits outside people to do work for the committee (point (f)), but without paying them or giving them any benefit in return. Instead, the committee requires the candidate to presume upon the candidate’s personal relationships with others, to get these other people to do work for the committee. This is also why the resulting letters are evidentially worthless (point (f)).

(6) It imposes an utterly unreasonable cost on the profession. Take the case of a fellowship (e.g., the NEH, ACLS, or NHC fellowships), for which there will generally be hundreds of applications, each with 3 letters, and each letter-writer is supposed to write a letter specific to that fellowship, for that year. In this case, the agency giving out the fellowship is imposing hundreds or *thousands of man-hours of labor* on the profession.

Even if this improved the committee’s decision-making somewhat, this cost would not be worth it. But it is actually extremely doubtful that it improves their selection at all, rather than worsening it. It introduces great scope for irrelevant factors to affect the outcome; if the other information in the applications is reasonably probative, then introducing this large influence from irrelevant factors *worsens* the selection of candidates.


9 responses to “What’s Wrong With Soliciting Letters of Recommendation?

  1. This is a compelling case. However, one wants to count the costs of alternatives before reaching a conclusion. But the costs outlined here are real and serious.


  2. This argument makes some good points about insiders and power, but not everyone who participates in the system is such an insider. It completely misses the point for most departments and candidates. Addressing point (c) specifically: Letters should come from people expert in the candidate’s area. Often departments are hiring for someone who adds an area to their curriculum, and often there is no one in the department expert in that area to do the sort of evaluation that the letter writers are asked to do. The inflated ego-driven nonsense letters you are talking about would carry no weight at all in hiring at my department at U Mass Boston. If we were looking for a specialist in an area beyond our expertise, we definitely looked for what more expert people would say. On more general professional questions, of course each can make their own decisions.


  3. I’ve made a similar case at my institution for many years. No success so far, because we are part of a state’s public university system that has some systemwide requirements. Applications for graduate programs should also not require letters. I believe they should be used infrequently, if ever, for the workload consequences alone. Many of us spend scarce research time in Summer writing these letters, which contribute almost nothing to a search process. I stopped reading them years ago, on the grounds that I was perfectly capable of reading and assessing the materials, and reading letters takes away time from that.


  4. These letters are also yet another way of disempowering early career academics, particularly graduate students who must ask advisors for letters that they don’t always deliver. I’m sure I’m not the only one who has been disqualified–multiple times–from fellowships for not having recommenders pull through, even despite doing everything “right” (asking well in advance of the deadline, sending painstakingly diplomatic reminders, providing simplified instructions and pre-filled forms, delivering application components for their optional review, only applying to the most feasible of competitions, etc. etc.).


  5. This rings true and the details of the case are nicely presented: the cost-benefit argument is surely unanswerable. Yet the conventional argument is not that such letters of support are a substitute for the committee’s own assessment, but that they provide opportunity for firsthand evidence about qualities or abilities not yet or not yet fully reflected in e.g. a c-v, gpa or test scores, to be added into the assessment. Second, in the case of job applicants, applicants to graduate programs, and undergraduate scholarship applicants, “letters of support” allow the reader to get some sense of the applicant’s mentors, and so allow a committee to make the necessary discounting of negatives or limitations in the applicant’s own work: given their independence, or different mentors and colleagues, even parrots and peacocks, one romantically hopes, might turn out to be larks, owls, nightingales, or eagles. Both these uses for letters seem to me real, even if neither meets the cost-benefit argument. What is absolutely clear is that letters of reference require skills in reading that too few committee members have, and that if the committee has such skills they would be better focused on the applicant’s own writing.


    • “they provide opportunity for firsthand evidence about qualities or abilities not yet or not yet fully reflected in e.g. a c-v, gpa or test scores”
      Judging from your comment, you are agreeing with the post in that recommendation letters should not be **mandatory**, right?


  6. One valuable aspect of LORs, the opposite of Patrick’s point here, is that there are people with GREAT CVs, who are terrible to work with. You might or might not care about this. Admittedly I have no suggestions for how to keep that part and not bring along the rest of the problems. Maybe give the phone numbers of 3 people who are willing to have a few 5 minute conversations on your behalf (possibly less hassle than the letters, harder to lie convincingly in a live back-and-forth?) Also, maybe start calling recommenders after a year for some sense of accountability? As it stands now, probably not too much thought given a year later to an overenthusiastic recommendation to get someone out of your lab, but if you know in a year your phone will ring and you’ll hear either “Hey, she’s working out great, thanks” or “You lying sack, why did you do this to me?”


  7. If they want to stamp out discrimination, they need to stop the whole job application process as it works today. Interviews rely on (often job-irrelevant) social self-presentation and instantly reveal race, gender presentation, and (in some cases) disability and sexuality, meaning people can quietly discriminate with plausible deniability (same problem if someone discloses, which is pretty much necessary to get “reasonable adjustments”). Some bosses also use potentially discriminatory performance tests and utterly pseudoscientific methods like handwriting analysis. I’ve even heard of things like nonconsensually causing unexpected stress to test if someone flips out or not. And of course they can perform discrimination based on name or postcode before they even interview. I reckon these are the main reasons that (gender, race, disability) discrimination legislation has been so ineffective. It wouldn’t be too hard to make them appoint the best-qualified person, or the person chosen by a third party like the jobs agency, or make them design job-specific performance tests which can be audited by third parties. This would also cut out a fair amount of the casting couch #metoo-ing by powerful men. There might have to be exceptions for a few sectors, maybe personal services and very small businesses, but I don’t see why it couldn’t be done for the vast majority of industrial, service, and professional jobs (except that the bosses wouldn’t like it and the government won’t do stuff the bosses don’t like). In cases like academia, there’s also the problem that it’s a bundled job and incapacity in any of the components overrides merit in any of the others (e.g. research/writing, teaching, admin, grant bids, impact). That’s also a subtle kind of discrimination, particularly in relation to disabilities and personality types (and in the case of bids and impact, political dissent).

    Illich also makes strong arguments that reliance on general qualifications is discriminatory because they track class and hidden curriculum conformity rather than sectoral ability (this seems an even bigger problem with vocational qualifications, where groups like Travellers are suffering from the growing demand for papers rather than ability). People can end up without school qualifications for a whole range of reasons such as disability, “behaviour”, class/race (going to a “bad” school), having a medical/family emergency at exam age, being bad at exams, and in cases with school leaving certificates or necessary specific exams (Gaelic to go to Irish unis, philosophy in Norway, science, English and maths in the UK…) then incapacity in one area blocks progression paths in lots of other areas. According to Illich, growing use of qualifications leads to a process of qualification inflation where class markers aren’t eliminated, but pushed “upward” (from O-level to A-level to uni to postgrad etc) at great public expense and increased exclusion of the “unschooled”. He prefers reputation and apprenticeship systems I think. I have to say, dark net markets are far less open to discrimination than any other system I’ve encountered, because sellers are always anonymous and rated solely by performance. The buyer never finds out if they’re black, white, disabled, foreign, female, gay, fat, Muslim, unless they’re caught and taken off the market. It’s possible to rig bad ratings and there’s literacy and cultural markers in listings, and the seller needs internet connectivity and ability to use the appropriate anonymity softwares (which may be class-linked) but that’s about it. I could imagine an anonymised job market functioning a bit like a dark web market making discrimination rather difficult, particularly if the traits listed were restricted.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s