Classic Edition: You Talkin' to Me?

This post dates from all the way back in July of 2002, and contains a bunch of thoughts on the preparation of different types of scientific presentations. I've re-covered some of this ground in the previous post, but there's enough different material to justify a separate Classic Edition post.

Since posting this, I've given several more Public Lectures, and they're a lot of fun. I've also gotten a lot better at ad-libbing physics lectures with minimal notes, which may or may not be a Bad Thing.

The text of my 2002 post is below the fold.

Having been tiresomely political for the past few days, I feel like I ought to provide some actual physics content. Happily, I'm supposed to be giving a lecture to a bunch of high school kids on Saturday, as part of a program to encourage poor and minority students to attend college, so I can use this web log to help pull some of my scattered thoughts together into a moderately coherent public lecture. Of course, being an inveterate procrastinator, I'm first going to babble a bit about the different kinds of talks you get to give as a physicist.

One of the biggest misconceptions about science and engineering is that scientists and engineers don't need to be able to write well, or speak well. The popular image of a scientist is a sort of socially retarded obsessive, thoroughly enraptured by odd details of science, but shy and mumbling and inarticulate when talking to other people. There's a little bit of truth to this, mostly in the "obsessive" part, but the reality is that communication skills are at least as important in science as in other disciplines. You can have Nobel Prize-worthy data, but if you can't explain the results, in print and in person, well enough to convince other people of their worth, you'll never shake hands with the King of Sweden. There's a lot of writing involved in science, and a lot of public speaking, though not the same sort of public speaking done by people giving oral reports to their high-school English class.

First off, it's worth noting a cultural difference between disciplines. In the humanities, when somebody "presents a paper," that's exactly what they do. They write a paper on the topic of their research, and then stand up in front of an audience, and read the paper to them. Visual aids are rare, as far as I can tell, and while some people write and speak in a lively enough manner for this to be bearable, it seems generally to be about as awful as you'd expect from this description.

In the sciences, on the other hand, visual aids are mandatory-- if you're ever asked to give a presentation to scientists, show some pictures. The pictures needn't be all that relevant, but scientists are used to getting pictures to look at when we listen to a talk, and without some sort of visual element, we get cranky. It's also considered very bad form for a scientist to prepare detailed remarks in advance-- exceptions are made for people talking in a language that's not their best, but reading from index cards, or even sounding like you've memorized the text of the talk, will start dark muttering among a scientific audience. You're expected to be able to present a series of data slides, and talk about them in a more or less extemporaneous manner-- you can have bullet points or snatches of text on the slides as a reminder to yourself of what goes where, but on a sentence-by-sentence, word-by-word level, you're supposed to put it together as you go along. I've never heard an explicit statement of why that's the rule (you never do, with these "community standards" sorts of things), but I think the idea is that you ought to know your work well enough to be able to ad-lib a coherent explanation of it. Figuring out how to cope with these restrictions is something of an art, especially since they're so different from the standard public speaking tips and techniques you get in school, which are aimed at someone who's planning to give the Gettysburg Address, not a scientific presentation.

(Historically, there's also been a rather rigid division between scientific disciplines as to what visual medium they use. Doctors and biologists inevitably used slides, while physicists always used overhead projectors and transparencies (often hand-written). It used to be that you could walk into a room at a conference, and make a good guess at the background of the speaker based on how he was presenting his data. These days, though, more and more people are doing PowerPoint slide shows, in all disciplines.)

Within those constraints, there are three rough categories of talks a physicist can be asked to give: Research Talks, Classes, and Public Lectures. There's some overlap between categories (you can teach a class about your research, for example), and there are subdivisions within the categories, but those pretty much span the full range of possibilities, unless you include things like commencement addresses and admissions pep talks, which I'm not going to concern myself with, since I don't often get asked to do those.

Research Talks are talks given to other scientists, though not necessarily people in the same field. This category contains the two extremes of the difficulty scale, in terms of preparing a talk.

The easiest sort of talk to put together is, somewhat surprisingly, a long research seminar. These generally run something like forty-five minutes to an hour, and are given to an audience of other physicists in approximately the same sub-field you work in. These can be a gruelling experience to deliver, depending on how many questions are asked during the talk itself (I gave a talk to my old research group at NIST a couple of years ago, and brought slides for the thirty-minute version of my research talk (I knew the group well, you see...). An hour and twenty minutes in, we called a halt to let people go to lunch, and I skipped the last five slides in my stack. It wasn't hostile questioning-- on the contrary, that's one of the most enjoyable talks I've given-- but they weren't letting any details pass unquestioned). They're easy to prepare, though, because if you're any good at all, you know the research inside and out, backwards and forwards, upside down and sideways, and could talk about it for two hours standing on your head underwater. The biggest challenge here is realizing that the stuff you spend 95% of your time working on holds absolutely no interest for the audience-- nobody cares about debugging code or aligning optics, no matter how clever you were about doing those things. As long as you stick to the essential physics, though, these are easy to do, and the format is open enough to allow digressions and extensions if necessary.

The hardest sort of talk to prepare, also in the research category, is the contributed paper at a conference. Not coincidentally, these are also the hardest sort to listen to, as so many are done badly. Contributed talks are generally given a ten or twelve minute slot, with a couple of minutes at the end for questions. There can be almost a dozen of these in a session at a conference, so the time limits are fairly strictly enforced. These usually take at least three drafts before they're really presentable, and even the best of them are little more than a teaser for the real work, a plea for members of the audience to ask questions after the session is over, when they can be answered in detail.

The problem with contributed talks is that it's nearly impossible to condense actual new research down into ten minutes, and have the result be comprehensible even to other physicists. You need to ruthlessly trim out all sorts of material, provide just enough of a sketch of the background problem to put the work in context, and present a single new result in the most concise manner possible. This is exceptionally difficult to do-- worse yet, these talks mostly fall to grad students and junior postdocs, who often don't have much speaking experience, so the information transfer from listening to the average ten-minute talk is near zero. Unless the session is on work that's directly related to something I'm doing at the moment, or something I've done in the past, I tend to skip sessions consisting of contributed talks, and either seek out sessions with longer invited talks, or gossip in the hall with other people who are also bagging the contributed sessions.

Somewhere in between these extremes are the half-hour invited talk at a conference (generally given to senior post-docs or established researchers with important new results. I've been lucky enough to give a fair number of these, and they're nice-- not as easy to put together as a full seminar, but much more relaxed than a ten-minute talk), and the general colloquium (generally an hour-long talk, given to students and faculty in a variety of fields, which may not overlap your own. These take a bit more work, because you need to provide more context for people who don't know the details of the field; on the other hand, you're not likely to get asked really difficult questions in one of these, and it can be really rewarding if done well. I like to think I'm good at this sort of talk, and that's what got me my current job, but I could be kidding myself).

Classes are a different sort of thing altogether. When you're teaching a class, there's generally a very tightly prescribed amount of material you have to cover, and at the end of it, the students need to understand it well enough to answer questions about it later. That means there's no fudging on the math or glossing over the messy technical details (as there can be in a research talk, where the idea is to get the general impression of the key results across). On the bright side, though, it also means you're allowed to have notes-- indeed, it's a terrible idea to try to teach a class without notes. The three or four worst classes I've ever sat through, an the very worst class I've ever taught, were done without the benefit of lecture notes.

The requirement that students leave the class with some understanding of the material also means that for every new concept you introduce, you need to have three or four different explanations (there's always at least one student who doesn't get the first explanation, but will be enlightened by a different approach), preferably with concrete examples or at least analogies drawn from everyday life, to drive the point home. Demos are a good idea, too. Happily, physics is a field with lots of quality demos-- I can't begin to think how I would go about teaching a math class.

The biggest problem in prepping a class often turns out to be finding the enthusiasm for the material-- when you're talking about research, you're necessarily talking about something that you find interesting enough that you've chosen to dedicate at least part of your life to studying it in detail. I can babble for hours about my research, without getting tired of it. I can't work up the same enthusiasm for a lecture on, say, rotational kinematics or vector multiplication. But however dull you may find the material personally, you still need to find a way to put it across to the students well enough that you can use it as a jumping-off point for more interesting material later. Even if you're bored stupid by the whole concept of rotational kinematics, you need to pretend it's fascinating material in order to get students to remember it later. It's a challenge, and it doesn't always come off, but it's an interesting sort of problem, and fun when it works (sadly, it also tends to leave the students with the impression that you're absolutely the biggest dork ever to stalk the groves of academe, but there's little that can be done about that).

The last category is the Public Lecture, which is basically what I have to put together for Saturday. This is a talk about physics to an audience who knows nothing about the subject (or at least an audience who can't reliably be expected to know anything about the subject). These are also amazingly difficult-- probably just below the ten-minute talk, and just above the class lecture. (Happily, I've already got some slides for a freshman-level talk about laser cooling, and the sterling example of Bill Phillips's public talks to crib from guide me.) The challenge here is to distill the material down to the central, really cool essence, without sacrificing too much, and without actually lying to the audience. You're not trying to teach people the full details of what you do, but you're trying to get across the basic idea, and give some hint of the vast and exciting possibilities. Again, demos and real-life examples are a good idea, the more the better.

I haven't done a lot of these-- most of the public-type things I've done have been presented to people who already have a reasonable background in the sciences, which makes it easier. In a lot of ways, though, this is the most rewarding sort of talk to do. To get the basic message across, you really need to recall what it was about the field you're in or the problem you're working on that drew you in in the first place-- you just don't get to the Ph.D. level in a science without thinking, on some level, that the field you're in is just the absolute coolest endeavor ever conceived since our many-times-great-grandparents first rubbed two sticks together and set fire to the savannah. It's that excitement and enthusiasm that you need to convey, even in the smallest part, to accomplish your purpose.

Reminding yourself of that excitement is also a wonderfully re-energizing thing-- recapturing even a small part of whatever spark drove you into and through grad school is enough to wipe away a week's worth of aggravation from, say, tracking lost purchase orders and fixing stupid little parts. The gleam in the eye of someone who heard you talk and Got It is reward enough to wipe out the effects of years of toil and drudgery. That's why I volunteer to do talks like I'll give on Saturday, that's why I got a job teaching undergraduates, and it's a part of why I'm doing this weblog.

Wow. That ran a lot longer than I expected. Next time out, I'll talk about laser cooling. For now, I have talk notes to prepare.