Letters of recommendation

Tue Oct 3 09:44:48 EDT 2017

Every year I am asked by two or three dozen people to write letters of recommendation for them; I expect that most faculty members get a similar number of requests. Since doing a decent job is a lot of work, you can help us to do it better. Here are some suggestions.

First, be sure you're asking the right person. If you did really well in someone's class or had a good experience doing independent work with them or if you have some other reasonably close relationship, that's a good start. But it's hard to write a useful letter if the writer doesn't know you and your work pretty well. People who read letters of recommendation for a living are exceedingly good at seeing that the writer doesn't really know the person they are writing about, and such letters may well do more harm than good.

If you didn't do well in a class, there has to be something compelling to explain that; otherwise, the writer won't mention your work and the reader will wonder why grades and standing are not mentioned.

If you didn't show up in class very often, that's probably even less good; even if the writer remembers you, it may not be an entirely positive memory.

And of course in these days of very large CS classes, it's really hard for the writer to know enough about most of the people in the class. You will generally do better with whoever is supervising your JP or thesis than with the person who stood in front of a class of 200 or more.

Second, help the writer speak about you personally, not just in general terms like name, class rank and serial number. What did you do or learn or enjoy in the class? Is there some incident you remember with pleasure? Did you do a project or a paper that you can remind the writer about? (If nothing comes to mind, go back to step 1.) My memory for people and artifacts is far from perfect, so if I get more help, I can write better, and that's to your advantage.

Make sure you include an up to date resume so the basic facts about you are easy to find.

Tell me something about what you're applying for, especially if it's something less familiar than say graduate school in the US. Describe the program of study. Explain why you want to apply and what you think your qualifications are for the position. Include a copy of your "statement of purpose" or "research plans" or the like; even a draft version is a big help. It's not enough to say "I'm applying for a Fidget Foundation Fellowship. Here's the link."

Third, help the writer deal with the mechanics. You should provide a precise list of what has to go where and in what form by when. I can probably dig some of it out of web pages, but it's more likely to be accurate and on time if you spell it out in words of one syllable so I can't make any mistakes. There's a lot of wasted time and room for error if things are not set down precisely -- you don't want your writer guessing about what to do.

I personally find it easier if all the materials are collected in one place -- all the links on one page, the resume and related material neatly organized, and the like -- and all arriving at the same time. Ask the writer what form is most convenient, then provide it. Most recommendations these days are handled by email and/or logging in to special sites. Tell me what to expect from who so I can tell you if it doesn't arrive.

Fourth, don't leave things to the last minute. Don't ask on Friday afternoon for a letter that's due on Monday. I can usually write a letter within a week, but there are times when it's just not possible, especially when everyone else wants one at the same time.

I'll tell you when I have sent your letter, so you know for sure that it's on its way. For your part, when it's all over, tell me what happened. It's a bit surprising that for all those letters each year, no more than a couple of people tell me the outcome unless I ask. There's no need to recount things that didn't work out, but when you finally decide among Microsoft, Harvard and P55, I'd love to know.