Saturday, October 13, 2007

The Year of Ignorant Living

This is a slightly modified version of an article I originally wrote for Conduit, our departmental newsletter.

Alumni might wonder about the charmed lives faculty lead on sabbatical. To be sure it is tough to return to civilian life, but not for what might seem to be obvious reasons (in fact, I've greatly missed the teaching!). Instead, this year has been terrific for me mainly because of what it's meant: a return to a state of ignorance.

For all our talk that research is an activity of constantly confronting ignorance, that's not what we really do. Research is more typically a man, a plan, canal panama (women sensibly leave absurd canals out of the picture). We may not know what precise result we're going to get—or even trying to get—but in the big picture we don't flail around very much.

I hadn't planned to spend this past year flailing. Now, I regard tenure less as a reward for past activity and more as a recognition of future promise; so the best way to honor it is to do something new, to view the freedom to take risks as an obligation to do so. Anyway, that's the theory; this runs headlong into (a) having established programs of work in place, (b) not knowing how to achieve ignorance (it's easy to decide to not publish papers or write grants, as Kathi and I did, but harder to decide what to do in its place), and (c) terror.

Proceeding with routine, I spent the summer and early fall working closely with Leo Meyerovich, Greg Cooper, Michael Greenberg, and Alex Bromfield on our new programming language, Flapjax. We finally released it formally in the middle of October to quite a bit of press coverage. In less than a year the experience of disseminating Flapjax has coughed up several surprises (press coverage for a programming language?—must be slow news days...), some negative in a curious fashion (as a result of which we've come to think of Flapjax not as a language but as a library), some surprisingly positive (such as its use at Berkeley). Those are all subjects for a different article.

We worked overtime on Flapjax last summer in part to have it out before I began my sabbatical travels. Kathi and I had been planning these trips for ages, carefully synchronizing the places we visited to be of mutual interest (since a sabbatical is also meant to be a time to recharge personally). Even before we left Providence, however, my carefully-laid plans were destroyed by a decision by the Brown administration that demonstrated a staggering lack of wisdom (needless to say, that won't be the subject of a different article). In a way, though, it was strangely liberating: if Brown didn't want me to accomplish what I'd set out to do on sabbatical, then I was free to do other things. So I did.

Our first stop was Edinburgh. Kathi was there to visit Keith Stenning, a cognitive scientist she knew from her work on diagrammatic reasoning, while I was there to visit Phil Wadler, one of the designers of Haskell and a pioneer of many programming language concepts. I was, however, also looking forward to talking to the seemingly dozens of other researchers Edinburgh has in programming languages, verification, and other parts of applied logic and in which Brown is desperately lacking. When it came to picking an office space, Phil told us that, by coincidence, he and Keith had adjacent offices and the one across the hallway from them was empty; would Kathi and I be willing to share it? It's been a long time since I've had an officemate but Kathi and I figured we could (just about) survive each others' company, and this way we could reduce our space footprint on their department.

What we didn't learn, until our first day in Edinburgh, is that our office neighbors in Edinburgh were Keith, Phil...and nobody else. Where I'd envisioned a long hallway with logicians in every direction you look, we were in rooms of a small tenment, whose door was locked to the world at large. Nobody was ever going to find us here, nor were we going to find anybody else! (Phil did arrange for me to have another, exclusive, office in the King's Buildings, but distance from home—more than any anti-royalist tendencies—made me use it only rarely. There I would have been near all those logicians, but still in a bit of an odd corner of the world.)

Geography is destiny, they say, and it couldn't be more true here. Stenning, it transpired, was no longer working actively on visual reasoning per se; instead he was understanding the logical models behind how people reason. His focus, with his collaborator van Lambalgen of Amsterdam, was on the famous Wason experiments in cognitive psychology, which are a kind of card trick that ask the subject to arrive at conclusions and measure how closely they hew to the entailment relation of classical logic; very poorly, it turns out. This has led some to conclude that logic itself is a poor way to study how people reason. (I hear the hallelujah's from Brown's cognitive scientists.) In contrast, Stenning and van Lambalgen, and others, had revisited the issue with much more detailed studies and found that there were parameterized families of logics that perfectly well explained how the subjects reasoned, and furthermore environmental characteristics—such as how the prompts were stated—predicted how people set the parameters.

Well! Kathi and I have been spending a lot of effort on the reasoning that goes into access-control security policies; but we've always known that what we're studying is tool support without reference to the underlying cognitive models. I had been nagged for a while now that properly executing this work demanded an understanding of these human factors, but I had no idea where to start. And now Stenning had accidentally shown us the world we were looking for. Understanding the consequences of this—and learning how to supress the repressed memories of my college psychology coursework experiences—has taken up a great deal of our effort since November, and will become an even stronger focus in the future. (There's one experiment I'd love to report on here, but can't yet. Yet.)

From Edinburgh we went to Oxford and Lausanne for PC meetings, thence to Paris to fly out to India. I've written at length about returning home after such a long time. After India came Australia (for a conference, followed by a personal vacation), about which, too, my notes will eventually show up here—for now, even nine months later, the memories of that continent are too vivid for words. This was the infamous left-right-left-right period of my life.

In late-January I attended a Dagstuhl event on Web programming, in which the main thing I learned is confirmation of my opinion that the Semantic Web folks are hopelessly out of touch with reality (perhaps it's a stealth marketing strategy). I was back in Deutschland ten days later at universities in Berlin (see blog), Tübingen, and Darmstadt, a well as another Dagstuhl, this one on end-user software engineering. Coming as it did after my Damascene conversion to thinking about user-interfaces this was a fantastic opportunity to revel in ignorance and soak up knowledge from the likes of Brad Myers, Mary Shaw, Margaret Burnett, Alan Blackwell, and Stephen Clarke (a UI designer at Microsoft).

In the early spring we visited the programming languages, security, and verification people at Penn, having several enlightening conversations with Insup Lee's group on obligations as a complement to access-control. We were originally due to spend all of spring at UT Austin; given all this other travel, however, we instead made just two very focused trips to UT (which too has a wonderful mix of applied logicians of numerous stripes). UT recently had the wisdom to hire Brown alum William Cook, who is surely one of the smartest and most tasteful researchers in programming languages; only Will can make even a topic like meta-modeling sound interesting. So a week spent primarily with Will and Don Batory was heavenly.

There were other trips scattered around, but the summer was a good time to consolidate and move forward. Usually I spend much of the school year planning for the summer (and hiring students for that purpose), but this year was obviously exceptional. So it was essentially pure luck that I stumbled upon two of the best students I've worked with at Brown, Jacob Baskin and Brendan Hickey, who continue in the tradition of Brown undergrads taking me in new directions (not least of all Brendan, thanks to whom I'm talking to vice-presidents and lawyers). Combined with two students elsewhere whom I'm co-advising, and my current PhD students—Arjun, who has made strong progress on a very interesting security technique, and Jay, who is feeding me doses of the Coq theorem prover when he's not busy getting married (congrats, Jay!)—it's hard not to realize that sabbatical is over and I'm back.

The end of sabbatical doesn't mean I've stopped plumbing the depths of my ignorance. In August, Spike got me excited about graphics for the first time, and I've been programming sporadically in Matlab since. Indeed, for the first time in my life I wrote a one-use, throw-away script that actually used trignometry. This has gotten me interested in research questions related to both the images and Matlab. I can only hope that if I lie down for long enough the feeling will pass.

I've also taken the plunge on a few other fronts:

  • I've long been skeptical of blogs, which associate a false temporality to thoughts. Largely pushed by Brown alum and Blogger employee Pete Hopkins, I created this blog anyway. It will be obvious to readers that I don't “get” the medium, treating it as a repository for essays rather than a dumping ground for thoughts; whether that will change, I don't yet know. I felt obliged to use Blogger, but in retrospect I realize I should have used anything but: that would be the way to test whether Pete was merely trying to drive up Blogger usage or whether he actually cared about what I have to say (my bet, like yours, is not on the latter).
  • I finally decided to self-publish my programming languages text, and to put it in print using Lulu, who have been impressive. (I actually publish the book in three formats: for-pay paper, for-pay PDF and free PDF. The beauty of self-publishing is that you can perform any outrageous experiment you want!)
  • I dove into understanding Creative Commons licensing—something I've put off for far too long—and found that it offered just the right mix of options for my book. So now people who've been excerpting parts of it (a.k.a., “remixing”) can do so legally.
  • I've started negotiations with a publisher in India that may result in a low-cost Indian print version, which is the one of the main benefits of a formal publisher I've missed.
  • I finally learned to use an image-processing application, so I can stop asking my colleague Spike, and Brown grad Morgan McGuire, how to do what I think they find the equivalent of balancing parentheses (well, for me; I count parens like some sharks count cards).

It's also been a wonderful year personally: from the urban delight that is Edinburgh to the new world being created in real-time in Bangalore, from walking in awe of nature in Australia to biking in Lance's town in Texas, from seeing (from afar) the site of the Burgess Shale to lying on my back on the Scituate Reservoir dam to bask in the Perseids. I've seen, up close and (sometimes) personal, everything from rattlesnakes to kangaroos, from a platypus to both black and grizzly bears. And as my blog's name suggests, cricket hasn't been too far away, from following a good chunk of the World Cup to fulfilling every fan's dream: watching England play Australia at the Sydney Cricket Ground, even if that verb is a euphemism for the abject surrender of the Three Lions we witnessed that day. Over up!