…even if some people write useless letters, as they do.  But the arguments here are not sensible (for example, "prestige bias" is a canard, as we've noted before).  As a senior philosopher elsewhere wrote to me last week:

In many years on hiring committees I have found these letters extremely valuable in distinguishing the A candidates from the A- candidates.  Maybe they aren’t so helpful though in distinguishing the B+s from the Bs?  I can’t think of another explanation for why someone would make such a bizarre anti-meritocratic suggestion.  Another oddity is that she seems to think that the ostensible purpose of these letters is to “contextualize” the candidate.  I have no idea what that means.

I concur.  In winnowing down an applicant pool, I look at three things:  (1) where was the applicant trained; (2) is the applicant a good fit for the position (based on training, interests etc.); and (3) the letters of recommendation.   I'm sure I am not alone in such an approach to figuring out which applicants warrant closer scrutiny of their writing samples.   Although, as we have noted before, too many letters are written in code, with time and experience, one can learn to decipher the code (and one also learn which letter writers are utterly unreliable).  And, happily, letters increasingly provide comparative assessments, which are far and away the most meaningful data point.