Using rec letters for 'internal' job apps?

On Twitter, a philosopher by the name of 'Julia' (Lecturer, Toronto) asks:

Job application etiquette question: is it considered bad form to use a letter of recommendation written by a person who works at the institution to which one is applying?

I'd be very surprised if this were considered bad form, as I would have thought that recommendation letters are generally thought of as 'all purpose' documents (i.e. the recommender supporting your candidacy for jobs in general). But I've also been surprised many times at the kinds of things that people can object to, and I also tend to think that when any question arises in one's mind, it can't hurt to ask! What do you all think? Would a search committee have any problem with this? Should Julia ask the letter writer if they are okay with it?

Comments

[Julia]

Thanks for posting this, Marcus. I got about a dozen replies on Twitter and the responses were mixed, although the general consensus was that using a letter written by a person at the institution to which one is applying is perfectly ok.

It occurs to me that the context might matter. If the letter-writer is your doctoral supervisor, then search committees might be more likely to see this as a conflict of interest, since it's in the interest of supervisors for their advisees to succeed professionally. On the other hand, if the letter-writer is, say, someone you met at a conference and maintain a professional correspondence with, it's less obvious that there's a conflict of interest for the writer.

[Tom]

I applied to (and was hired at) an institution where one of my letter writers - and external dissertation readers - worked. This writer asked me not to use their letter, as they worried that it would decrease my chance of getting the job, by creating the impression that they were exerting undue influence on the search. They also reclused themselves from discussions until the finalists had been chosen. Not sure how much of this was due to the dynamics of the department or how generalizable this is.