I will be heading off later this week to the Academic Sponsors’ Day at MSRI, going as Shmuel’s proxy for uchicago. I’m not sure to what extent (if any) there is for me to make policy suggestions, but any comments you would like me to pass on to MSRI management would be appreciated. Post them below!

There is one program at MSRI that (from my mostly second hand observations) seems as though it could do with some improvement. MSRI regularly holds a number of summer schools. For example, this summer, Kevin Buzzard is giving an introduction to the theory of automorphic forms. I would have loved to been able to go to such talks when I started out as an graduate student, and I really wanted to send some of my current students to this. The problem? Firstly, there are a severely limited number of places. I’m not entirely sure I understand this, but I can imagine a few reasons. However, MSRI (apparently) goes to extreme lengths to be as “equitable” as possible in admitting people from as many different sponsor schools as possible. The result? I am told (internally) that the University of Chicago will be at most able to send two or three students in total to the seven or so different programs available. The result is that the only people uchicago sends to these programs are basically advanced students who are about to graduate and aren’t in any sense the target audience. On the other hand, these programs also tend to admit students from schools with much weaker backgrounds who aren’t even comfortable with basic concepts from algebraic number theory. This seems to be a very stupid way to choose participants for any program, even if it is purely in the name of fairness. What does a Harvard number theory student have to do to be able to attend an introductory course on automorphic forms — prove the Sato-Tate conjecture?

Apparently at least one lecturer plans to give their 20 courses in the order 1,11,2,12,3,13,etc in order to try to please at least half the people half the time.

One wag suggested that schools from weaker programs should consider it in their best interest to only admit students from places like Harvard, since then at least the programs would be training their future professors ([Caveat: I heard this second hand, so it may well have been in jest]). Presumably the reason behind MSRI’s policy is that the spoils of MSRI programs should go equally to (students of) universities that fund MSRI (sssuming they contribute a similar amount.) However, there seems to me to be a very natural alternative solution to this problem, namely, to continue having “advanced” summer courses but also introduce some deliberately introductory courses tailored to people from schools with less background. Students from schools with more advanced programs could be barred from the lower level introductory programs (since they could be more easily reproduced locally) and then, when it comes to the more advanced topics, there wouldn’t be the restriction to admit at most one student from each school. In particular, the selection criteria should concentrate (in part) on who stands to get the most out of each program.

Please add any further suggestions or complaints below

“The result is that the only people uchicago sends to these programs are basically advanced students who are about to graduate and aren’t in any sense the target audience.”

I don’t quite get it. Chicago has great first-year students; why not send them, if people at that stage are the target audience for Kevin’s course?

Chicago is asked to send a ranked list of applicants to MSRI. When I explained to the relevant [internal] person how my students were perfect for this program exactly because they didn’t yet have the background in this area, the response was:

In this case I’m afraid he will not be supported by MSRI, as it is very competitive. However, they say that additional students may be considered if they pay an attendance fee of $1700 and also cover the travel expenses. This is far from being sure, but a possibility.

It may indeed be the case that the exact mechanism whereby we nominate students is non-optimal, of course. However, I have the impression that even if *I* was teaching the course, I wouldn’t be able to get my own graduate students into the program.

I’m giving a 45-minute talk about Sage and SageMathCloud at this upcoming MSRI day – let me know if there is something you want me to address or mention…

This is not exactly a response to your question, but… whenever I try to find something in magma using google, I invariably arrive at the high page rank site “overview of magma,” which never ever contains any useful information about how to actually use particular functions. It’s always a little frustrating. Not that there can be examples of every single possible computation in the help files, of course.

Is that meant to be magma or sage?

I think that I can see both sides of the issue there, as someone who co-organized such a summer school recently (Summer 2015, on gaps between primes). MSRI did try to tell us that the students came from different backgrounds, asked us to submit a questionnaire to assess the average knowledge of the audience, and asked us to suggest reading lists, but nevertheless, I think that the range of previous exposure to any analytic number theory ideas was much greater than I expected, so some of the later lectures were certainly quite hard for many students. (And schools involving much more formalism are probably in even tougher positions).

On the other hand, I am absolutely sure that 99% or more of the students did benefit from being there, and the proportional effect is probably greater for those (properly motivated) with weaker prior experience. The solution you suggest smells a bit to me of “equal but separate” (with the choice, would I prefer to teach the advanced school for students who already have Brun’s sieve at their fingertips, or the basic one where one would prove the Prime Number Theory after two weeks?). I think there is no shortage of schools and conferences of various kinds accessible to the best students from top universities — but many students at our MSRI summer school would probably never have had any other opportunity to see Sound lecture on the Bombieri-Vinogradov theorem, or Maynard on Maynard’s Theorem…

There are key starting points in many subjects that students find tricky to pick up easily — and covering one of these topics would make for an excellent summer school. Given the number of easily available good expositions of the prime number theorem (for example), I think it would a waste of resources to have you teach such a course on that rather than about (say) sieving. Similarly, KB should be explaining how to think about admissible representations of GL_2 rather than p-adic numbers.

The alternate answer that “there are lots of other great workshops available for stronger students” argument might also be reasonable, but if MSRI wants to go that way, they need be much more explicit about it to everyone (including the lecturers). But I’m also not even sure it’s true. I think there are *advanced* courses for graduate students, but not so many intermediate courses. The Arizona Winter School is a notable exception, but that’s in part why the AWS is so popular. (I think Chicago has something in the order of 15 graduate students going to AWS this year…)

Back when I was involved in organising the AWS (and most other commenters above were students participating), we’d typically get five or six applicants from Harvard (and similar places, on occasion including UT), all well-qualified so we’d write to the Harvard graduate advisor that this was the case but our funds were limited. We’d offer to fund two or three (and not try to distinguish which ones) and ask Harvard to fund the rest, which they would. I’m sure something similar is still being done.

Indeed; I believe that I am supporting the travel of about 15 of our students (not all mine, of course).

I strongly agree with your complaint about the way MSRI handles summer schools. I was a graduate student participant at such a summer school on arithmetic geometry a couple of years ago . It was very hard to secure a spot for the reason you mention: top schools have many interested students but only two or three spots across *all* summer schools. After a couple of days at MSRI it became clear that a significant portion of the audience were graduate students in analytic number theory (typically, from schools with weaker background) who didn’t know the first thing about arithmetic geometry (nor schemes for that matter). Naturally, the lectures were well over their heads and some of them didn’t even show up for the second week! What an utter waste of resources, when plenty of well-qualified and motivated students from top schools were denied the opportunity to attend!

In your fourth paragraph, “sssuming they contribute a similar amount” should be “assuming they contribute a similar amount.”

But of course if I correct it now it would make this comment silly

@galoisrepresentations: I *am* teaching an MSRI course this summer, and I *know* that my graduate students can’t attend, because my institution does not support MSRI. I was told this very early on and of course was already aware of this when I was told (TG and RLT had explained how it worked to a certain extent, and I was quick to ask MSRI for details). I was also told that there would be <= 2 students per supporting institution, although more recently I was told by MSRI that they could now strengthen the result: there will now be <= 1 student per supporting institution. I am well aware that this means the standard deviation of what the students know will be large. It will be a challenge to deliver material which is deemed appropriate by all I guess — although Emmanuel's comments do fill me with hope, and of course one can ask hard questions about easy material to keep the Chicago student happy while I'm explaining some of the basic structure of the absolute Galois group of the p-adic numbers…

@JSE: once this situation was explained to me (about who the audience would be) I was asked to submit prerequisites. I said "the p-adic numbers and a first course in modular forms". I am similarly bewildered about the concept that I might end up with 5th year grad students. This is supposed to be a first course in what I think the local and global Langlands conjectures for GL_n *say*, with essentially no details about how one might prove anything beyond GL_1.

Pingback: Report from MSRI and Berkeley | Persiflage

Pingback: MSRI Now | Persiflage