Microsoft Research Internships

One of the best things about working at MSR is the internship program. For a sense of what an MSR internship is like, I recommend this essay by Philip Guo.

In this post, I want to personally reflect on the MSR internship program and provide some context about how things have worked for me in the past. Let me stress that this reflects only my personal experience and may not be representative of other MSR researchers' experiences.

What I Love About the Program

Through this program, MSR researchers have the opportunity to work with the best students in the world. With our interns, we spend 12 weeks (usually during the summer) working hard on research. The internships are intense but always gratifying and I like to think that most MSR interns are really happy with their experience.

What I Hate About the Program

As much as I love the internship program, there is one aspect of it that I absolutely hate: namely, sending out rejections to the multitude of incredibly talented students that applied but did not receive an offer.

What bothers me so much about this is that I know that there is a huge disconnect between how students understand and interpret rejections and how rejections actually occur. After all, I was a student not that long ago.

More precisely, the problem is the following. Students think that the application process is a fair game, a meritocracy. That the "best" candidate wins and if a candidate doesn't get the offer, it was because there were other candidates that were "better"; for some vague, undefined notion of "better".

The application process, however, does not work this way. First of all, the set of all candidates is not a total order; that is, most candidates cannot be ranked in any objective sense. Counting papers is, frankly speaking, idiotic and focusing on publication venues is meaningless when there are so many biases in the publication system. Throw in the fact that some candidates have a huge disadvantage because they are from a smaller University or one that is less known in the US and you see that there is really no objective measure.

The other thing to realize is that simply hiring the "best" candidate (even if you could define such a metric) is often a poor strategy from our point of view. Often (though not always, of course) we have a sense of what projects/area we want to pursue over the summer. Given that we already know, roughly speaking, what we would like to do, our best strategy is to choose someone who has the required background to bring the project to completion. So being a good "fit" for the project is often the most important criteria; not some vague and meaningless notion of "being good".

Third, it is important to realize that we make hiring decisions based on a multitude of factors, some of which are completely outside of a student's control. Some years, we may have a high-priority project to work on and so the offer is made to the student with the most aligned background. Other years, we may make an offer to a student with whom we have an ongoing collaboration so that we can finish the collaboration. Some times, if we cannot decide between candidates then we may make the offer to the most senior since that student may be on the job market soon and the internship would be of greater value to him/her. These are just a handful of reasons but there are more.

What to do if You Didn't Get an Offer

The first thing is to not to take a rejection personally! 90\% of the time, it is not a reflection on the quality of your work or application. The most likely case is that the offer was made to someone else for one or more of the reasons outlined above.

The second thing to realize is that having submitted an application is good! How can it be good if you got no offer? There are at least two reasons.

The first is that often, when we evaluate a candidate that we conclude is not a good fit for the projects we plan on working on, we forward the application along to other groups who we think might be a better match. So, for example, if I come across your application and see that you are very strong in verifiable computation but I am not planning on working on verifiable computation this summer, then I will forward your application to Bryan Parno. This happens a lot! We spend a lot of time trying to find other options for people we don't make offers to and more than once this has worked in a candidate's favor.

Another reason is that even if you didn't get an offer this year, you will be on our radar and you'll most likely be considered the following year. I know of several cases where a candidate did not get an offer one year but got one the following year (even though they didn't re-apply!).

Does this mean that you will eventually get an offer. Of course not! There are no guarantees. However, the point is that applying is not a waste of time---even if you do not get an offer. Also, this clearly illustrates that you should not take a rejection personally. If we go through this much trouble to find you an alternative internship or to keep you in mind for the future, then clearly the rejection is not a reflection of your work.