Monday, November 17, 2008


Ask the Administrator: Selling the New Dean on an Idea

A new correspondent writes:

I'm a postdoc at a big research university.  We have a confluence of
events at our University which could really lead to something really
great happening, but no one seems much interested, and it would
frustrate me to no end if the opportunity is missed.   So I'd like to
know -- what's the best way to pitch a new idea to a new Dean?

Without spending a single dollar, our University is in the process of
having a large supercomputing center plopped down right in the middle
of it, and in fact the necessary flow of money will be adminstered by
the University.   This was arranged from start to finish by a small
number of fiefdoms, who quite reasonably feel very possessive about
it, and who control most of the new center and are gearing up and
ready to use this fantastic resource once it comes online.

As things stand now, this enormous resource will ensure that the
members of the founding fiefdoms do exactly what they have already
been doing, but bigger and better.   And that is surely all to the
good; but if this is *all* that happens, it's an enormous missed
opportunity.   This could be a boon to the entire University if
handled properly - if expertise were shared, both among the fiefdoms
and to other groups within the school who currently aren't huge users
of supercomputing for their research.   It could become a catalyst for
bringing together interdisciplinary work - first by sharing techniques
amongst researchers, then by sharing courses (why should every
department have to teach its graduate students their own separate
computational techniques classes?), and hopefully bringing people
across departmental divides to work on problems of overlapping
interests once people realize they're approaching similar problems.
It could improve connections with researchers within the same city and
with some access to this resource but outside the walls of the

But there's currently not much motivation for anyone to follow this
path.   The members of the small group of founding fiefs don't have
much incentive to encourage other groups to use the resource, and
other departments don't seem to be chomping at the bit - partly
because they're almost completely out of the loop as to what's going
on.   Even once they start realizing how to use this new resource on
their own, again it'll just be the same groups doing exactly the same
thing as before but scaled up.    Not a disaster, but not what could

Given that the resource exists, it seems like encouraging its use and
using its existence to encourage interdepartmental collaboration
through a campus-wide `computational science' program could be done
without spending a lot of dollars - it's not clear new faculty lines
necessarily need to be created, for instance.   The founding areas are
already ensuring that new hires are made with this new resource in
mind (at least one such person is already hired).    It's `just' a
matter of setting up institutional incentives for these new people to
talk to each other, and to find people in corners of other departments
willing to start using this resource for their own work.

The other part of this equation is that we have a brand new Dean -- a
very gifted administrator and one who has already successfully
shepherded collective projects across multiple administrative units,
but one whose background is such that the appeal of a campus wide push
for interdisciplinary computational science might not be self-evident.

I want to make sure that the new Dean knows (with lots of examples)
what could be with some administrative support.   I know that everyone
on campus has their own Great Ideas for the school and  there may well
be important reasons why this particular one may not work or can't be
a priority now, but I want to make sure that it at least gets real
consideration.   So what's the best way to do this?

There's a lot here. I'll start with the obvious.

In the short term, the political battles you're asking the new Dean to pick are substantial, and the payoff pretty theoretical. That's a tough sell. (This is especially true if the new Dean isn't a subject matter expert in this area. I've been in situations when I was pretty sure that faculty in a particular area were trying to exploit their greater expertise to paint a pet project as a mortal necessity. Go through that a few times, and the skepticism starts to come easy.) From this side of the desk, it looks like you're asking someone to do all the political dirty work for you, in service of your pet idea. Your pet idea may or may not be great – I have no way of knowing, and neither does he – but the dirty work is clearly difficult.

My suggestion would be to come at the Dean indirectly. Instead of just marching into his office with an idea and a plea, start by doing some of the political groundwork yourself. Flesh out some of the benefits of your pet idea, then target some of the faculty to whom those benefits would be the most attractive. In other words, gain allies. If you can do that in multiple fiefdoms, all the better. This will involve some basic shoe-leather diplomacy, as well as considerable patience and listening skills. It's not something that a few tossed-off emails will achieve.

That said, though, the upside is considerable. For one, you may learn from your prospective recruits that there are real obstacles of which you're currently unaware. Those obstacles may come from the rules set by whomever's paying for the thing, or from logistics, or from technological limits, or weird policy intersections, or departmental budgeting, or just about anything else. By finding out what those are, you will be able to develop work-arounds, or to know what work-arounds to ask to have constructed, or you may decide it simply can't be done in your current context. But whatever happens, you won't leave yourself vulnerable to devastating blindside attacks.

In addition to strengthening the merits of your proposal, you'll also be able – if you play your cards right and catch a break or two – to get around the objection that it's Your Idea. (This works particularly well if you're open to amendments.) The way this usually plays out in campus politics is that the proposal is reduced to the affiliation of the person proposing it. “Oh, that came from X department. We know what they're really up to.” If you have interdepartmental allies, that argument is much harder to sustain.

(If someone gets wind of what you're doing and raises an objection, ask to be recognized as a committee. Deaning 101 suggests that when confronted with conflict, the first and easiest path is to appoint a committee. You'll be pushing an open door.)

When you have interdepartmental sponsorship and an inherently stronger proposal, you'll have a much more compelling argument to take to the new Dean. Theoretical appeals are fine, but theoretical appeals that also have faculty champions across departments, and that have already been vetted for their practicality, are so much better.

The other thing I'd do is be prepared to achieve success gradually, rather than all at once. “Pilot project” can be a magic phrase, when properly deployed. The great virtue of presenting an idea as a pilot project is that it's much harder to argue against an experiment than against a wholesale change. If you can create enough of an opening to at least generate some of your proposed payoff on a pilot basis, then you can argue from actual success. Nothing succeeds like success, especially when that success has faculty champions from multiple areas at once. At that point, the momentum will be considerable, and the Dean will be both more sympathetic and better positioned to do something about it. (Insider Tip: we administrators get all excited whenever someone builds “outcomes assessment” into the design of the pilot. In other words, when pitching the pilot, tell us how you'd define and measure success. It takes a little more time upfront, but being able to say “we met or exceeded our expectations in key areas” gives you major credibility.) A pilot project that succeeded on its own terms becomes an incredibly powerful argument for a larger project.

Of course, you can also try the traditional and time-honored methods of brown-nosing, currying favor, etc. I have to admit that those sometimes work, depending on context, but they're pretty much independent of the merits of the proposal. The approach I'm suggesting could actually strengthen the merits of your proposal in demonstrable ways, such that you wouldn't have to worry overmuch about the usual political crap. Try the high road first; the low road will always be there anyway.

Good luck!

Wise and worldly readers – any thoughts?

Have a question? Ask the Administrator at deandad (at) gmail (dot) com.

You're a post-doc, so life may be too short for this.

Getting people to talk with each other can usually be done quite cheaply (i.e., by paying for lunch). What you may want to do is setup a large-scale computing seminar series (or if it already exists, then you may want to publicize it widely). Since this is cheap (especially without outside speakers) these "fiefdoms" (or someone else) might be willing to fund it.

Yeah - and worse, I'm a postdoc at the end of my term. But I just see a bunch of events coming together and just want, on my way out, to clearly and convincingly convey to the dean who may not realize it that ``this new resource could be the catalyst for a lot of cool cross-disciplinary stuff if you encourage its use that way.'' What I'm hearing is that it's probably not worth the effort.
DD and anon 1 didn't say it, so I will: Have you tried to get an appointment to meet with the Dean? I'd guess the odds of a audience with the dean of any large college in an R1 for a non-entity like a post-doc is vanishingly small unless you have a grant proposal to pitch. Otherwise, just send the Dean an e-mail stating your status ("I'm only passing through, but thought you should think about this opportunity") and idea.

Your idea also assumes there is any free time on that machine. Although possible, it is more likely that all of that time is fully devoted to producing the bazillions of papers promised by the people who "arranged" the grants to acquire this computer system.

Oh, and the reason different groups teach their own computational courses is similar to the reason many groups teach their own statistics classes: they don't trust any other department (particularly the computer science department) to teach the skills needed in their particular field.
This comment has been removed by the author.
Let me try again, a *little* more politely - this sort of effort, if successful, might help to avoid exactly the sort of parochialism that ccphysicist alludes to. I come from a physics background myself, and if I hear one more time about how physicists know more about numerical PDEs than people in applied math or CS, I just may scream.
Let me be perfectly clear "large computer" and at no cost are like "military intelligence". The cost of feeding (electricity) and cooling (air conditioning) the damn things easily exceeds any possible purchase price in a very short period of time and the room preparation that is needed would by your average small mansion in Beverly Hills BEFORE the recent meltdown (there are unceasing discussions of this on cluster mailing lists).

Threaten the fiefdoms that you will let the Dean and the VP for Physical Facilities know what baby will really cost if they don't share
My advice:

It's a great idea. Find someone else who thinks your idea is great, print this out, and leave, hoping that you've at least started the process.

DD is absolutely right about the process required for this; the choice isn't between going to the Dean with some chance for success and leaving -- it's between zero chance for success and someone else doing the sort of groundwork necessary.
This is utterly hopeless, unless you can convince the groups/departments that got the grant to support you. They've undoubtedly got a detailed plan for how to use this machine, and even if your plan is objectively better, there's simply no way you can force the issue. (No dean would support a postdoc to overrule faculty who just brought in a big grant. Even if you get a lot of other faculty on your side, it will still come down to who got the grant.)

So I'd forget entirely about the dean. The only chance I see is to go to the people who control the computer and to try to convince them that this is in their best interest. Make it clear that they won't give up anything (for example, that anybody else using the computer would be scheduled when they have nothing pressing - if such times even exist!), and try to explain the benefits of increased publicity and prestige, connections to experts in other areas who may be able to help them, etc. I don't think you'll succeed, but you never know.

Ultimately, you shouldn't expect that getting someone to relinquish control over a existing resource will be any easier than getting a grant to pay for such a resource in the first place.
Gee, Jon, now I wish I'd seen what you deleted! I would probably have agreed with most of it, except for the notion that the typical CS prof even knows what "PDE" stands for.

There are vast opportunities for collaboration between applied math and various fields (physics, engineering, biology, chemistry) that use high performance computing. Lots of cycles have been wasted with bad algorithms, not to mention bogus ones, but nothing quite like what gets wasted using pointers to manage large arrays when solving those PDEs in C++. But the real topic that CS doesn't address in its intro classes is that of code management. My experience included few CS faculty who dealt with 100,000 lines of code.

PS - Not to digress any further (I should blog about it), but I will mention my favorite chapter (Interlude: When not to computer) in Acton's "Numerical Methods that Work".
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?