How many (outside) letters are optimal?

A reader writes in:

I have a couple of questions about getting job letters from people outside of your graduate program. 

First, when is it a good time to ask someone for a job letter? How well does a potential letter writer need to know you to write a sufficient letter? 

Second, I will likely have at least two external job letters when I go on the market. Is this enough or would more be better? How many letters does one need on the market?

Excellent questions. Here is how I replied:

I don’t think there’s a simple answer to the first question, as in my experience different people can have different standards for how well they think they should know you in order to write a good letter. So, here’s my suggestion: if you have some working relationship with the person and any reason to think the person knows your work and may think well of you as a philosopher, it doesn’t hurt to ask. The worst that can happen is that they say ‘no’—but the best that can happen is that they say ‘yes’ and write a good letter. I’ve had both happen to me, and while it’s disappointing when someone says ‘no’, the only way to get some ‘yes’ answers is to ask.

On the second question, I think two outside letters is more than enough, particularly for someone just coming out of graduate school. Indeed, given that many job application portals limit the number of letters that you can submit, there is such a thing as having ‘too many letters.’ I think four or five letters total (2 or 3 from your grad program, and likewise from outside) is probably optimal.

But these are just my thoughts. What do you all think?

Comments

[Letters]

A side issue: resist the temptation to get many letters. Some people seem to think the more the merrier. But in fact, I think - due to a salience bias - people reading a file often remember a weaker letter. So if you have three good strong letters, and one okay letter, a committee member may focus on that one. They may even see it as "more honest". So do not be tempted into an arms race with numbers of letters.

[Timmy J]

A slightly different side issue: I think it’s wise to ask every single person who you think *can* write you a strong letter to do so. Not because you’ll use them all (Marcus and the previous commenter are right about this) but because it’s good to have good people think good thoughts about you and then write them down. Later when some such person is e.g. thinking about who to invite on their next edited volume, they’re more likely to think of you. There’s something at least very like the Franklin effect to think about here.

[William Vanderburgh]

Letters are not that useful to most committees, I think, so don't go overboard. Three is perfect; five is overboard. Have one that speaks to your research, another to your teaching, a third that's more general, and you've got it covered. Committees know that letter writers tend to oversell things, so they discount the praise somewhat and substitute their own judgement on your writing sample, teaching materials, and other evidence in the file--focus on making those things excellent. Having a letter from a well-known person who honestly says "best student I've had in a decade" certainly won't hurt, but even that isn't likely to be definitive in any given search.

If you use a credential service and have a (reliable, experienced) placement officer, try requesting five letters and then having your placement officer recommend which three to send (and to tell you if any of them should not be used).

If you have been out of grad school for a year or more, be sure to have the chair of the department where you have been teaching write you a letter (that's a sensible fourth letter). If they can't/won't, that probably tells you something important.

[P]

I think it depends what your AOSes are when you go on the market. For example, if you list ethics and political philosophy as AOSes but don't have a single political philosopher vouching for you then I think it's worthwhile to try to connect with a political philosopher who can do so. Likewise for other common AOS pairings like metaphysics and phil science or metaphysics and epistemology.