Climbing Mount Boredom
Two weeks ago, I argued that scientific papers are basically a waste of time. Today I’d like to generalize the results of that earlier post, by explaining why scientific talks are also a waste of time.
Let me set the scene for you. You arrive at the weekly colloquium eager to learn, like a cargo cult member who’s sure that this time the planes are going to land. But then, about fifteen minutes after the PowerPoint train has left the station, you start to get nervous: “Why are we stopping at all these unfamiliar little hamlets? Are we really headed for the place mentioned in the abstract?” You glance at your fellow passengers: are they as confused as you are? (You’d ask the guy sitting next to you, but he’s sound asleep.) Eventually the announcer comes on and … uh-oh! It seems the train is about to begin its long ascent up Mount Boredom, and you don’t have the prerequisites for this leg of the trip. Can you dodge the ticket collector? Too stressful! You get off, and the train roars past you, never to return.
Such was my experience again and again until three years ago, when I finally gave up on talks as a medium for scientific communication. These days, whenever I have to sit through one, I treat the speaker’s words as background music for my private fantasies and daydreams, unless the speaker chooses to interrupt with a novel idea.
But what about when I have to talk? To be honest, I haven’t intentionally perpetrated a research talk in years. Instead I do a stand-up comedy routine where you have to be a quantum computing expert to get the jokes. It’s like Seinfeld, except not that funny. So why does it work? Simple: because the crowd that expects to be bored is the easiest crowd on Earth.
Now one could argue that, by stuffing my talks with flying pigs and slide-eating black holes, I’ve been setting back the cause of scientific knowledge. But I don’t think so. See, the basic problem with talks is that they have no anti-boredom escape hatch. I mean, if you were chatting with a colleague who droned on for too long, you’d have several options:
- Change the subject.
- Say something like “yeah, I get it, but does this actually lead to a new lower bound?”
- Tap your fingers, study the wall patterns, etc.
- If all else fails, mention your immense workload, then excuse yourself and go back to reading weblogs.
The key point is that none of these tactics are inherently rude or insulting. All of us use them regularly; if we didn’t, it’d be impossible to tell when we were boring each other. Put differently, these tactics are part of the feedback and dialogue that’s essential to any healthy relationship:
“Was it good for you?”
“Could you maybe go a little faster?”
“Do you like it when I use this notation?”
The seminar speaker, by contrast, is a narcissist who verbally ravages his defenseless audience. Sure, it’s fine to interrupt with things like “Aren’t you missing an absolute value sign?,” or “How do you know A is Hermitian?” But have you ever raised your hand to say, “Excuse me, but would you mind skipping the next 20 slides and getting right to the meat?” Or: “This is boring. Would you please talk about a different result?”
(Incidentally, as my adviser Umesh Vazirani pointed out to me, when people get “lost” during a talk they think it means that the speaker is going too fast. But more often, the real problem is that the speaker is going too slow, and thereby letting the audience get mired in trivialities.)
So what’s the solution? (You knew there was going to be one, didn’t you?) My solution is to replace talks by “conversations” whenever possible. Here’s how the Aaronson system works: you get five minutes to tell your audience something unexpected. (Usually this will involve no slides, just a board.) Then, if people have questions, you answer them; if they want details, you provide them. At any time, anyone who’s no longer interested can get up and leave (and maybe come back later), without being considered a jerk. When there are no further questions, you sit down and give someone else a chance to surprise the audience.
If you don’t think this system would work, come visit our quantum algorithms lunch at Waterloo, Tuesdays at 11:30 in the BFG seminar room. Bring a result or open problem.
Follow
Comment #1 October 18th, 2005 at 2:33 am
Scott, the best (i.e. most enjoyable) talks I ever gave were at Hebrew university in the combinatorics seminar. It was fantastic–two hours of yelling, shouting, interruptions, side conversations, technical details, and philosophical meanderings. Some people were asleep, some were keenly interested, and some were overtly argumentative.
Contrary to popular belief, it’s really the audience that makes a talk soar (or, more often, flounder). Admittedly, it’s not for everyone; as the speaker, you have to be quick on your feet, and know your subject inside and out. But it’s the ultimate machismo ride–with audience members trying to dethrone you at every step, and trivialize every line of your proof.
You’re already sweating at the one hour break, and just when you think you can excuse yourself for a sip of water, three or four people corner you at the board, asking for more details about Lemma 7 and Proposition 22.
There’s nothing like 20 roaring Israelis to bring out the adrenaline junkie in you. 306 Soda, on the other hand… I’m sorry, did you say something?
Comment #2 October 18th, 2005 at 2:39 am
and while I’m on the subject…
your metric of “surprising the audience” is way off base. We can all bob our heads along to Britney Spears, but real music takes time and patience to appreciate. Like a good radiohead album, you have to endure some aural hardship before you’re smacked in the face with the beauty of it all.
Comment #3 October 18th, 2005 at 2:44 am
I completely agree with you. I am usually bored by trivialities and cannot catch the moment when it comes to more interesting things (and one minute later it is already too late). I can only make use of a talk if I continuously interrupt it with questions (which is customary at our discrete math seminar, but usually not welcomed at conference talks).
Comment #4 October 18th, 2005 at 4:25 am
I agree, and think your quantum algorithm lunches sound awesome, but if you rule out papers and talks, how do you recommend communicating results to a group of a few hundred people?
A dozen small talks? A marathon poster session?
I’m giving lectures to 60 sophomores now, and while of course the intellectual connection would be far better if I could tutor them one-on-one, it would also take me 60 times as long…
Comment #5 October 18th, 2005 at 9:29 am
Maybe the real point is that all means of communicating your results are a waste of time. You should just do research for your own sake, and demand that the system pay you.
Comment #6 October 18th, 2005 at 9:45 am
Greg and Aram: No! The solution is to have informal dialogues in front of a large audience (or dialogues that get recorded and written down). Although this idea is more than 2000 years old, I’ll post about it soon.
Comment #7 October 18th, 2005 at 10:23 am
Well, the dialectical style of presentation has its own advantages. But I don’t think you can say that one mode is clearly better or worse in all instances.
Writing papers for CS conferences may involve jumping through more hoops than necessary, but tweak the system a little bit, and posting to the arxiv can work well for everyone involved. Giving a boring powerpoint talk to 100 people may be a giant waste of talent and/or of a beautiful Sunday afternoon, but change the audience size to 15, or suppose that the talk is meant to be expository, or put enough Israelis in the audience, and all of a sudden it’s engaging for everyone.
Dialogues are great when there’s debate. That way you don’t just hear the opening argument, you find out how people can defend their arguments against criticism, and then respond to the criticism of their defenses and so on. But I can’t imagine replacing a well-structured calculus lecture with a Platonic dialogue.
(Heh – this is my way of drawing out your next post where you explain informal dialogues. Then I can attack that, and you can respond, developing the idea further….)
Comment #8 October 18th, 2005 at 11:02 am
“(Heh – this is my way of drawing out your next post where you explain informal dialogues. Then I can attack that, and you can respond, developing the idea further….)”
That’s exactly why I started a blog…
Comment #9 October 18th, 2005 at 11:03 am
Scott: Your position reminds me of Plato’s criticism of writing in general, as quoted by Andrew Odlyzko:
…this discovery of yours will create forgetfulness in the minds of those who learn to use it; they will not exercise their memories, but, trusting in external, foreign marks, they will not bring things to remembrance from within themselves. You have discovered a remedy not for memory, but for reminding. You offer your students the appearance of wisdom, not true wisdom. They will be hearers of many things and will have learned nothing; they will appear to be omniscient and will generally know nothing; they will be tiresome company, having the show of wisdom without the reality.
An informal talk before an audience of 1,000 people is a great idea. The only problem is that when I give informal talks, I usually get an audience of about 3 people. As Aram suggests, and as various data shows, the arXiv provides most authors with the largest serious audience that they can ever expect.
Comment #10 October 18th, 2005 at 11:05 am
But there is one continuing tradition of talking for hours, as informally as you please, to a committed audience. It’s called teaching.
Comment #11 October 18th, 2005 at 11:08 am
Anonymous (James?):
Yeah, I had a similar experience when I talked about the collision problem at Hebrew U. I started out by recalling the classical birthday paradox. Immediately Avi interrupted me: “There ought to be a sign on the door: let no one ignorant of the birthday paradox enter this room!” I took the hint, and quickly moved on to the actual results.
Comment #12 October 18th, 2005 at 11:45 am
Audience participation is key for a good talk. At a talk I gave at MIT, the students interrupted me with all manner of questions, such as “would this still work if you did it this other way?”, “is the third field strictly necessary?”, “aren’t you missing a special case?”… by the end of the talk one of the students had applied my idea to solve a problem he had been working on.
In contrast, in the average talk in most other places I’ve talked, the audience sits quietly bored rather than interrupting you and shifting your talk towards the interesting tid bits.
Also, imho, every good talk should have a general 15 minutes overview of the field, at a depth corresponding to the expected level of expertise of your audience. That way at the very least you walk out with a general idea of where “quantum algorithms for interactive proofs of on-line poker tournaments” are at, even if you don’t care much about the specific results of the “narcissistic” presenter.
It is true, however, that numerous interruptions wouldn’t be appropriate in 200 attendants conference talk. In that case I don’t know what is the alternative.
Alex Lopez-Ortiz
Comment #13 October 18th, 2005 at 12:43 pm
It seems to me that there is a difference between talks given to small groups where you are really trying to communicate the result to talks given to large crowds, say at conferences, where the purpose is more to expose others to your result without the details. While the first is more important, and deserves an audience which participates, it seems that the second would be hindered by this. I go to a conference not to learn results, but to learn what results I want to go back home and read up on.
Of course, you could think the second purpose is silly. As Greg says, just let the research speak for itself. There is certainly something to be said for this. Sadly, _the system_ doesn’t like this. The system whereby you get money to do your (and your students) work based not just on the quality of your research but upon your _reputation_. If you are enough of a star that your research shouts out to everyone who reads it, then you might be able to avoid _the system_ (as, for example, Feynman seemed to do ;)) But for the mortals…
And yeah, most talks suck and I agree that for me a good talk is one where I get some work done while I’m listening to the talk. But do you think the problem is really that the talks stink or that the research sticks?
Comment #14 October 18th, 2005 at 1:13 pm
“It is true, however, that numerous interruptions wouldn’t be appropriate in 200 attendants conference talk. In that case I don’t know what is the alternative.”
Maybe to have a few “designated interrupters”? 🙂
Comment #15 October 18th, 2005 at 1:20 pm
“But there is one continuing tradition of talking for hours, as informally as you please, to a committed audience. It’s called teaching.”
Yes, and I can’t wait to join that tradition! But I’ve heard there’s also an implacable enemy of the teacher, called the “syllabus.”
Comment #16 October 18th, 2005 at 1:27 pm
Dave: “But do you think the problem is really that the talks stink or that the research [stinks]?”
That’s a subtle question. Research that smells good when presented in five minutes, can start to stink if aired in public for an hour.
Comment #17 October 18th, 2005 at 2:47 pm
(The problem with blogs is that if you
are the 17th commentator, you will
have no impact.)
1) Math is MUCH WORSE than Comp Sci in
this regard. Talks are boring and
too hard. AND they all know this
but don’t care.
2) AGREE that audience interaction is great,
though really hard to do at big audiences
like FOCS or STOC. May work at Complexity.
DOES work in my 8-person complexity
seminar.
3) As for powerpoint:
THEY CAN HAVE MY CHALK WHEN THEY
PRY IT FROM MY COLD DEAD HANDS.
bill gasarch
Comment #18 October 18th, 2005 at 3:09 pm
The MIT cryptography seminar also sounds like something you’d like. (In form, not necessarily in content, since you don’t do cryptography. 🙂 Sounds similar to the combinatorics seminar mentioned by Anonymous #1, although I like to think the motivation of most people was to understand and apply the result rather than dethrone the speaker. Of course, more than one result was broken in seminar.
As for teaching calculus through discussion, people are trying. My high school did away with textbooks entirely, in favor of problem books. Each day, we’d be assigned a few problems, then we’d come in and present our attempts at a solution and discuss. At the time I hated it, but it much more closely mirrors the way research is done than lectures…
Comment #19 October 18th, 2005 at 5:43 pm
David M.: “My high school did away with textbooks entirely, in favor of problem books. Each day, we’d be assigned a few problems, then we’d come in and present our attempts at a solution and discuss. At the time I hated it, but it much more closely mirrors the way research is done than lectures…”
This *might* work, but isn’t the open-ended exploratory aspect a little contrived if the teacher knows the answer?
Comment #20 October 18th, 2005 at 8:16 pm
By the way, Scott, your blog would look a lot better in RSS and Atom if you titled your entries.
Comment #21 October 18th, 2005 at 8:46 pm
I rather like your solution. But consider my department’s alternative: do the talks at someone’s house (with many distractions) and serve wine, beer, and various types of alcohol. Works wonders, I tell you. AND leads to a very “lively” discussion afterwards.
Incidentally, my most recent bad experience was at a “Theories of the Mind” talk. The guybasically tried to tell us that only people in modern day Western Culture actually possess what we think of as a “mind.” Yikes… I ended up leaving the talk after 45 minutes of roasting the poor guy. But several others stayed to continue the slaying. I’m not sure if he made it out alive…
Comment #22 October 18th, 2005 at 9:12 pm
What’s your average # of talks for a typical conference? At a standard FOCS/STOC, I will go to about 5 talks, although I have been to as few as one, and as many as 15 (sometimes you get trapped in a session and have to sit through the whole thing).
But other people seem to take these damned things so seriously, and I don’t get it. Maybe they’re thinking “money is tight and I already paid the conference fee” or “if I’m taking off three days from my family, I had better attend some talks.” ???
Comment #23 October 18th, 2005 at 10:28 pm
“By the way, Scott, your blog would look a lot better in RSS and Atom if you titled your entries.”
Sorry about that! But you see, picking titles I’m happy with would probably take me longer than writing the actual entries — so if you want titles, I’m not gonna be able to post as often. 🙂
Comment #24 October 19th, 2005 at 12:44 am
Then I think that you should take some of your own advice about exposition and compose titles in a more relaxed way. 🙂