What a Letter of Recommendation Is Actually For
I write roughly fifty college recommendation letters a year. Each one means sitting down and describing, in specific, evidenced detail, a version of a student’s scientific promise that they usually haven’t fully articulated to themselves yet. I don’t think that’s a special gift. It is the same skill the book spends seventeen chapters trying to teach a reader to turn on themselves: look at the actual evidence in front of you, not the story you’ve already decided is true, and let the evidence revise the story. The student who will ask for a letter and the teacher who will write one both start months before the deadline.
What a strong letter is made of
A strong letter is specific, evidenced testimony. It names what the student did, in what room, against what difficulty, and what that conduct predicts — not adjectives, and not a transcript restated in prose. The raw material is built over months, in the small moments a teacher happens to be watching: the question a student asked after the unit moved on, the result they refused to discard before asking “what if it isn’t a mistake?”, the week they kept a notebook honest when nobody was checking. A letter manufactured in the two weeks before a deadline has none of that to draw on, and admissions readers can tell.
What a student can do about it, starting now
You cannot write your own letter, but you can make sure there is something to write. Sit in office hours or stay after class for one or two teachers, not to ask for anything, just to become a real person to them before a favor or a letter is ever needed. Ask to be pressed harder on the questions a course raises. Finish the thing that didn’t work and say so. When you do ask, ask early — six weeks is respectful, two is not — and hand the teacher what a letter needs: the program, the deadline, and two or three specific moments you think showed who you are as a thinker, so the letter can be evidence rather than praise.
What a teacher can do about it
Keep a running file, one line at a time, of the moments worth remembering — the same idea journal the book recommends for a scientist’s own half-formed questions, kept for students instead. By the time a request arrives, the letter is mostly written; the sitting down is assembly, not invention. And write the version of the student that is true and not yet visible to them. The best letters I have written told a student something about themselves they had not put into words, which is also, not by coincidence, what the best teaching does.
- Evidenced testimony
- A claim about a person backed by a specific, checkable instance of their conduct — the same claim-evidence-reasoning structure a lab report uses, aimed at a student instead of a result.
- Idea journal
- A running record of moments that don’t sit right or don’t fit; for a scientist, the seed of a question; for a teacher, the seed of a letter.
Check your understanding
- Rewrite one adjective you would want in your own letter (“hard-working,” “curious”) as a single specific moment a teacher could have watched.
- Why does a letter written from months of small observations carry more weight with a reader than one written in the two weeks before a deadline?
- What is the difference between a letter that praises a student and one that gives evidence about them? Which would you rather have written about you, and why?
Drawn from the Epilogue and Chapter 15 of Everybody’s Science (the recommendation-letter passage moved from the chapter to this site in the August 2026 revision) and from the author’s own practice. No external sources; nothing here is a claim about admissions policy at any institution.