Better conference talks

I've been to a lot of conferences and seen a lot of talks and it's amazing to me how a bad presentation can get in the way of really exciting science. The recent Lunar and Planetary Science Conference wasn't unique in this regard but for some reason I felt pushed over the edge by it. This (very long) post is a response to my frustration about bad conference presentations. I do feel a little hesitant to set myself up as an expert on this because I know I have a lot of work to do, to improve my talks. Still, I think I have useful advice to offer.

I can summarize the whole thing in three words:

Respect your audience. All those people in that room in front of you: they are not you. But their time is as valuable as yours. Work to deliver them a presentation that is designed for them, to inform and interest them in your work, to leave them pleased that they spent that 5 or 10 or 50 minutes of their valuable time listening to you.

Here are some questions to guide you in preparing a good talk.

Let's take these questions one by one.

Whom are you speaking to?

Most scientists at conferences appear to be speaking to themselves, or, perhaps, to the people who will eventually be reviewing their paper. Perhaps that's all you care about, in which case you can stop reading this post right now. But if you actually want people in the room to learn anything from you, you need to think about who they are and what they will come in to the room knowing and not knowing. The wider an audience you are addressing, the more contextual information you will need to provide to them. Deleting necessary context from your talk in order to present more of what you did cuts out large swaths of audience. It is an act of disrespect to your audience. If you do not provide the people in your audience with information that they require in order to understand you, it is the same as telling them that you do not care if they understand you or not.

Let's take a specific example: let's say your work concerns using the CRISM imaging spectrometer on Mars Reconnaissance Orbiter to map clay and sulfate minerals on Mars and that you are using your maps to infer things about Mars' geologic history.

I usually think of this in terms of widening circles within your audience. You and your coauthors and/or advisors are a very tiny circle. Widen that circle a little bit to the people doing the same kind of work in your field, and you don't have to change your language very much. Widen the circle a bit more to take in other kinds of geologists, though, and you begin to run into trouble. Mineralogists who do not study Mars will understand the significance of mineral names, but they may not know what a CRISM is or what it does or what its resolution is or what its other limitations are or where clay minerals have and haven't been found on Mars. On the other hand, people who study Martian geomorphology from orbit know what CRISM is but may lack your ability to instantly mentally translate a mineral name to its chemical formula, and they might lack the background knowledge of the kind of sedimentary environments that typically produce the mineral.

And we're still talking about pretty specialized Mars scientists here. What about Mars atmospheric scientists attending your talk? What do they know? What about graduate students? What if an asteroid astronomer wanted to try to find out what's going on at Mars? What about impact crater experimentalists? What about a space journalist? A NASA administrator? A high school earth science teacher? Their students? The conference venue's A/V technician? You can widen your circle ad infinitum.

At some point, you do have to stop. You do have to identify the circle containing the people who already possess the necessary contextual knowledge that will permit them to understand everything you are telling them, and recognize the fact that some people who are there will not understand everything. Please note that I am not speaking of intelligence here. I am speaking of background knowledge. Still, context trumps data, every time. You can spend a whole conference presentation talking about TLAs to an audience of incredibly intelligent people but if they don't know what a TLA is, it's likely you won't have communicated a danged thing.

("TLA" = Three Letter Acronym.)

Really good speakers are ones who manage to communicate something to everybody in the room, no matter who they are or how much they already know. To the relatively ignorant, you should at least convey the driving questions behind your work: why should anyone care what different kinds of minerals appear in different places on the surface of another planet? What did you learn, and why does it matter? At the same time, to the well-informed, you should convey how your work has added to or broadened or contradicted what has come before it. Even the specialists spend so much time lost in the weeds in their fields that it benefits them to be reminded of the big picture.

If this is overwhelming, pick somebody in the middle of the ignorance/specialty spectrum to pitch your talk to. When I write, I am usually imagining that I am writing for a college-educated person whose degree is not in the sciences. At a conference, I'd imagine a graduate-educated person whose specialty is not the same as mine. And then I ask the next question.

What do you want your audience to learn?

It amazes me that people prepare talks without ever asking this question, but they appear to. A lot of people really spend a lot of time describing their research methods -- what they did, and what their data look like. It's easy to understand why people make that mistake: what you did is, after all, what you spent most of your time doing. But the whole point of your work was to learn something that you could then communicate to others. Don't force the audience to go through the same process you had to, in order to get to the result. You can save your audience all that work by telling them what it was you learned.

If, through your presentation, you interest an audience member in your work, they can read all about your methods and data in your paper. (That's what a paper is for.)

Here's an exercise that I highly recommend: Compose a Tweet summarizing your talk. You get 140 characters. Maybe even just 120 (there's always overhead, for a conference hashtag and your surname, or a link). You don't get to use text-speak or Esperanto. It needn't have perfect Strunk & White grammar, but it needs to be sensible, comprehensible English. In that limited space, you are not likely to say a whole lot about your methods! If you do, you are boring. "I mapped clay minerals on Mars." Who cares? "Large areas of Mars experienced rainfall over tens of thousands of years." Cool.

Make that Tweet your conclusion slide. Make sure that your talk delivers that conclusion. How are you going to do that? Well....

What is your story?

It is impossible to overemphasize the importance of narrative in a talk. You, standing up in front of an audience, are telling a story in which you are the principal character. What's your motivation? What are the big questions that drive your professional curiosity? Did you answer your questions, or was your search fruitless?

There are several fairly standard kinds of stories that work great for scientific talks. The easiest ones for space science: you are solving a mystery. Or you are an intrepid explorer who has gone to a place no one has gone before. Maybe you have fought a pitched battle with a legendary monster of a data set. (This is a great framework for a presentation about a null result; you get to be the tragic hero.)

Stories are fun. If you tell a good story, you hook your audience and then they will willingly follow you even into dark corners of your subspecialty.

Stories are also functional, especially for people in the audience who may be struggling to follow you on that journey. If, for example, you have managed to tell your audience that this is a crime story, pretty much everybody in the room should be able to understand what the crime was at the beginning of your talk. Then, if you lose them while you're talking about evidence gathering, you still have a chance of picking them back up again when you tell them: that was the evidence, and this piece of evidence led me to the perp. Even if an audience doesn't get spectroscopy or understand what a general circulation model is, they probably get how crime stories work.

Narrative is not just helpful to your audience; it's helpful to you, too. It provides a structure for your talk, and helps you determine what is crucial to conveying your message, and what is not. Which is very important when you consider the following question:

How long do you have to speak?

You cannot say all the same things in a 15-minute talk slot as in a 1-hour colloquium. You just can't. Don't even try. However, you can tell the same story, which is why I put story before talk length in this blog post. Do you have a favorite novel that's been made into both a miniseries and a movie, and maybe even a 1-hour show? Think about the differences in story among these. As you go from longer to shorter versions, you see reductions in characters; in settings; in subplots; and finally in the main plot line itself. Yet the story (usually) remains recognizable. Exactly the same process is necessary to go from a scientific paper to a colloquium to a long conference talk to a short conference talk.

It is especially important for very short talks (like at the Division for Planetary Science meeting, where the slots are only 10 minutes long, meaning 6 minutes for speech) to practice your talk and then, if it is too long, cut out information that is not needed to tell your story. You cannot solve the problem of a too-long talk by talking faster. You must simplify the story that you are trying to tell.

It is only now, once you have identified your audience, your take-home message, and your story, that you should begin to think about making a PowerPoint presentation.

What visuals will serve to amplify your story?

I've observed that a lot of people use the phrase "prepare a talk" as though it is synonymous with "compose a PowerPoint presentation." Don't do that. Please.

I don't hate PowerPoint, not at all. PowerPoint doesn't kill scientific conferences. People kill scientific conferences with bad PowerPoint presentations. PowerPoint -- or any other means of projecting visual content in front of a large audience -- is a tool, and like any tool, it can be used for good or for evil.

When PowerPoint is used for good, it serves to emphasize or amplify points that you, the speaker, are making with your voice and your body language. No matter what, your slides should serve to enhance your presentation, not to distract from it.

The number one error that almost everyone makes with PowerPoint presentations: There are too many words on your slides.

We use the same parts of our brains to process spoken language and written language. If you show me a slide containing more than a few words, I must choose between reading your slides and listening to you speak. I am physically incapable of doing both at the same time. If I try, I am liable to jump between reading some text and listening to some speech and then I miss things and I get lost. If your entire talk is written out on your slides, why the heck are you even talking to me? I read faster than I hear. You could just stand up there silently and advance your slides periodically. But of course that would be boring and I'd stop paying attention. Which is what I do anyway when you put up slides with too many words on them.

(Some speakers compound this evil of reading their entire talk aloud from their slides by using a laser pointer like the bouncing dot on karaoke lyrics, zapping each word as they read it. There are few quicker ways to make me stop paying attention either to your slides or your speech.)

Try this exercise: Put words on your title slide -- your title, name, coauthors, acknowledgments. Put words on your conclusion slide -- that Tweet I suggested earlier, and your name and contact information. On all the slides in between: no words. Just pictures. And pictures only when necessary. If a picture would not help me understand your point, put in a blank slide. Yes, I'm quite serious: a completely blank slide.

There is a great deal of power in a completely blank slide. Put one up and watch the entire audience suddenly make eye contact with you. I like to put blank slides in places where I am making transitions in talks. It is a reminder to me to remind the audience where we came from, and inform them where we are going. I can look them in the eye and check in with them to see if they are still with me.

Tips and Tricks and Do's and Don'ts

Now that I've gotten my lecture on talk organization out of the way, here's a list of bits of advice gleaned from personal experience, from the advice of friends, and from a wide variety of sources on the Internet. Unfortunately, I do not remember the ultimate source of most of this advice; a lot of it originated in multiparty hallway conversation. If you see something here and say "Hey! That was my idea!" feel free to take credit in the comments.

Speak Differently Than You Write

Do not use initialisms or acronyms in your speech at all, unless you are confident that they are understood by everyone present. A funny example of this one is "LPSC." LPSC stands for the Lunar and Planetary Science Conference. Everybody at LPSC knows what LPSC is, so that one is okay, as long as I'm at LPSC. But the people who are reading my blog? and my husband? They don't know what an LPSC is, so I will have to tell them. Every time. And honestly it's often easier to say "that conference I just went to."

Look suspiciously at any abbreviation that you are tempted to use. Ask about each one: will everyone in my audience know what I mean when I say this? If not, then I'm sorry, but you can't use it. You don't have to say "Mars Reconnaissance Orbiter" instead of "MRO" every time. You can just say "the spacecraft" or "the ship" or "the orbiter." More people will understand what you mean by that than will understand "MRO."

I hate talks where people define a new initialism in the beginning of the talk and then carry on using it throughout the rest of their talk. If you are writing a long paper that is all about something called recurring slope lineae, it is okay, even perhaps useful, to define "RSL" early and then use the initialism to refer to the features in the rest of your paper. It makes the text easier to scan. However, it is BAD BAD BAD to do this in a talk. If someone misses your definition, they will spend your entire talk wondering what the heck an "RSL" is and they will learn nothing from you.

Put two initialisms in your talk, and you become four times harder to understand. It's exponential, I swear.

The same goes for any piece of technical jargon. Jargon exists for a very good reason. Just like an initialism, jargon acts as shorthand, making communication among people who know the jargon much more efficient. But jargon is an impenetrable barrier to people who can't translate it. And if the piece of jargon that you are using is the subject of your talk, and you insist upon using it, you've just walled off a big chunk of audience. Don't disrespect your audience by using jargon they won't know. It's not good enough to define it once in the beginning of your talk.

So, wherever possible, replace jargon with a few words that can stand in for it. Yes, this takes a little more time. Which is more important to you, squeezing in a couple more sentences, or speaking to more of your audience? (There is only one right answer to that question.)

You must use less jargon in a spoken talk than in a written paper. In a paper, if I come across a term whose meaning I don't recall, I can look it up. In a talk, I can't do that. Consider, for example, the terms "stress" and "strain", which have very specific, precise meanings in physics. (Stress is a force acting upon a material, while strain describes how the material deforms, or changes shape, in response to a stress.) Even now, it takes me a split second to parse the terms "stress" and "strain", making sure I've always got their different meanings straight in my head. The professor who taught me my first structural geology course often (not always, but often) substituted words like "push" and "squish" for "stress" and "strain" when lecturing, along with appropriate hand gestures that emphasized which was the push and which was the crumpling. She used the proper terms in written assignments. The words she spoke were not as precise, but if she used only the precisely correct terms I would've understood less, not more.

Related to that, here's a cool trick: Don't use polysyllabic Latinate words when you can use Anglo-Saxon ones. Speak these two sentences: "Curiosity remained immobile." And: "The rover stood still." Both convey the same information; they work pretty similarly, when you're reading silently. But you can be so much more emphatic speaking the second one. Here's another comparison: "The compressive stress resulted in a crustal length reduction of thirty-one percent." And: "The crust got squished to just two-thirds as wide." Use the first one in your paper. Speak -- and gesticulate -- the second one. I wish I could remember who told me this trick, because I love it.

Whatever you do, don't call this "dumbing down" your language. You are not dumbing it down; you are oomphing it up. (Yes, I did just use "oomph" as a verb. Sue me.) Language is a tool for the communication of information. Use the words that will produce the strongest signal in your audience's receiving brains.

Simplify your sentence structure. In a technical paper, a single sentence can span a whole paragraph. Modifying clause piles on top of modifying clause. It's a way to armor sentences against criticism. But excessive dependent clauses are deadly in speech. If I lose track of which noun your lengthy list of clauses is modifying, I lose the whole sentence.

Simplify. Finish making one statement before you move on to another. Don't conjoin sentences when you can split them into separate ones. If your audience contains people who speak English as a second language, simplifying your sentences will greatly increase how much you are communicating to them.

If a point is important, repeat it. You can't bold it or underline it when you're talking. But you can repeat it. Repetition is verbal underlining. It also functions like putting up a blank slide. Repeat something and you'll start making eye contact with your audience again. Say something three times and they'll all be looking at you.

Passive voice stinks. Actually, this is as true in written text as it is in speech. Don't say "it was observed," say "I observed." A lot of dramatic things happen in space, but we suck all the drama out of them by making sentences passive. Rearrange your sentences into subject-verb-object form and see how strong a verb you can stick in there.

Don't go over your time. DON'T GO OVER YOUR TIME. Speaking so long that there is no time for questions informs your audience that you do not care what they think of your work or whether they understood your presentation. Speaking so long that you run in to the next speaker's time informs your audience that you think you are more important than the next speaker and more important than anything else the audience had been planning to do at that time. Both are insulting and disrespectful. You look like a jerk if you go over your time.

If, when you talk, you seem to vary widely in your talk length, bring your smartphone to the podium and run an app that has a countdown clock, and make sure the screen will stay on through the length of your talk. If you seem to be running long, do not talk faster. That will reduce, not increase, what audiences take away from your talk. Skip ahead and finish on time.

Anyway, if your talk length varies widely, odds are good that you're going down side paths that you should not.

Your slides have too many words, and other problems

If you must put words on your slides, use very few. No complete sentences. A title for a slide is okay. If you are talking about something that may be unfamiliar to some (say, recurring slope lineae), please show a photo of that thing and label it as being that thing. That will aid understanding. But don't put a several-sentence definition of the thing on your slide. Then people will read your definition instead of listening to you.

No text below 20 points. If most of the audience can't read it, why bother putting it on the slide?

Graphs can be a really big problem. There is a tremendous amount of information in a graph. Different color symbols! Different axes! Which way do the axes increase? Are they logarithmic or linear? If they are linear, is zero on the graph or are you just showing me a little bit of the range? Are there error bars or not? Are the lines thick enough to see? Do the colors show up on the screen? Can I even read the axis labels? Even if the font is big enough, are the labels so low on the slide that the rows of heads in front of me are blocking them? Which part of all of this is actually important to the point you are trying to make? And how should I figure all of this out while you're talking at me? Think very carefully before you include a graph in your talk, and then be prepared to spend minutes explaining it. And take out every line or label that is not strictly necessary. Graphs that are suitable for papers are generally NOT suitable for talks.

There are some kinds of graphs that have a very common form so are instantly recognizable -- to people who possess the necessary background knowledge. Those of you who live and die by R-plots or infrared spectra probably have to show them in your talks. Sometimes. Maybe not as often as you do now, but sometimes. But make sure that, one way or another, you present them in a way that someone who doesn't know what the heck an R-plot is will still understand the point that you are trying to make by showing it. Unless your talk is to the R-plot Appreciation Society.

Equations are even worse than graphs. Seriously, don't put equations on your slides. On very rare occasions I have seen people show an equation and effectively use it as a means to convey to the audience how much more strongly one term controls the behavior than another term. You wizards who can do that effectively may go on putting equations on your slides. The rest of you: don't.

Data tables: BAD. No. Don't.

Approximately one in ten of the men in your audience is color-blind. What this means: never, ever use a ROYGBV spectrum to represent a continuously varying property. Vischeck is a super website to use to ascertain whether your graphics will be incomprehensible to the color-blind.

One thing per slide. You can only say one thing at a time. What is on the screen should be serving to emphasize what you are saying right now. If it is illustrating something else, it confuses rather than aids people in understanding what you are saying. It works very well to start with a relatively empty slide to which things get added as you speak -- for instance, starting with a photo and then adding circles around interesting features as you point them out. That's a great method when it works, but if you find yourself in a situation where you are not controlling the clicker and you have to verbally cue the poor A/V person to advance to the next time-step to put every circle and line on your slide, then you're in trouble. Still, I have found that A/V techs (or the session chair or whoever is driving the presentation) are getting better at listening and queueing slides that get built incrementally in this way; speakers just point or wave and the tech clicks.

Without words, your slides will almost certainly not be able to serve as a stand-alone record of your presentation. If your slides could stand alone, then your presence wouldn't be necessary. Obviously this means that you can't make a presentation by reading from your slides. People who read their slides to the audience are usually facing away from the audience. It is disrespectful to your audience to turn away from them, and moreover, it looks stupid. You're going to have to have separate notes. I mostly don't use notes when I speak except when I'm very nervous, but I still have notes from when I was planning my talk, and I keep digital copies of those notes in a folder with my PowerPoint presentations. If your PowerPoint presentation is going to be archived somewhere, feel free to add a slide after your final one and just paste in your notes there.

DO put your name on your slides, maybe even your abstract number. I can't tell you how much of my time in the audience at conferences I spend shuffling around the program trying to figure out the name of the person currently giving the talk so that I can find and read the abstract later. (This is usually because I was struggling to compose a tweet-length summary of the preceding talk, and I missed the introduction of the next speaker; or because the previous speaker in a different room ran long, and I showed up late to the current talk.) Put your name and maybe your abstract number in the corner and you make my life so much easier. You also make it more likely that you'll be contacted after your talk is over by someone who was interested by it. If you want people to be interested in your work beyond the few minutes of your talk, especially if you are a relatively obscure person in your field (say, a student), put your name on every slide. If nothing else, make sure to put your name and contact information on your conclusion slide.

Your last slide is the one place where you ought to have a sentence. I like one Tweetable sentence that summarizes your talk. Don't make your final slide say only "Thank You" (though there's no harm in also putting that on the slide) and especially don't make it say only "QUESTIONS?"

Odds and Ends

A word on the number of your slides. It's a commonly cited rule of thumb that you should have about one slide per minute. I think, though, that that assumes that people (or you) will be reading your slides. If you have a slide full of 20-point font, it will probably take about a minute to read that slide aloud. This one-per-minute rule of thumb doesn't work as well if your slides aren't word-heavy. And it makes the PowerPoint presentation drive your talk organization, rather than the other way around. So I don't find that rule of thumb particularly useful. Focus, first, on what you want to say. Have slides at appropriate places to emphasize what you are saying. If you can't say what you need to say in your allotted time, you need to say less. Eliminate slides or slide content that are no longer needed to support what you are no longer saying.

A word on animations. I have to say that this recent LPSC is the first conference I've attended in a while where I did not witness many train wrecks involving animations not playing properly. Still, if your presentation contains an animation (and they can be awesome visuals), make sure you have tested that your animation works. If you do not have an opportunity to test using the exact system that is employed in the conference hall, have a backup plan that does not involve berating the hapless A/V technician. I like animated GIFs in PowerPoint presentations because they always seem to work. If you know you will have control of the clicker, an even easier way to do a not-very-many-frame animation is just to put one frame per slide and advance them manually. That will work even if (horrors!) your PowerPoint is turned into a PDF.

A word on anxiety about forgetting your talk. I think a lot of people write their entire talk on their slides because they're afraid they'll stand up in front of all of those people and forget what they want to say. I have a lousy memory, and find it impossible to memorize the specific words I want to use; getting away from using text on slides as a crutch was a serious challenge for me. PRACTICE helps a great deal.

Here's a method I use, especially for short talks. I try to memorize the first sentence I intend to say about each slide or sequence of slides. Then I speak more off the cuff for a few sentences about that part of the story. While I am doing that, I visually check in with my audience to make sure they are with me. When I advance the slide, I glance at it, and that triggers the sentence I intended to say when I advanced that slide. If I have words on slides, they are usually just titles; those titles also serve as my cues to help me get my intended first sentence out.

A word on what to wear. When you are speaking at a large meeting, you should be slightly better-dressed than your audience; being better-dressed commands attention and that's what you're there to do, to command attention. At geology meetings, any amount of effort spent on your appearance will make you look sharper than your audience, so this is not a hard standard to meet. Particularly if you are young and have not yet earned much in the way of professional respect, dressing nicely is a way of adding to your respect-worthiness. Personally, I view a certain amount of effort spent on personal appearance as a sign of respect to the people who have to look at you. It's a sign that you take yourself seriously, which is a suggestion to me that I should also take you seriously.

There are people who successfully pull off not being dressy. These are usually senior and/or highly confident people, who can succeed without the respect boost that dressing nicely gives them.

For women: the fact that you get to use makeup when you give a talk is a rare situation in which being female is an advantage; use it! Even if you (like me) never wear makeup on a day-to-day basis. When you are on stage and are fighting for people's attention, it is easier to make eye contact with people if your eyes are more noticeable. It's easier for people to read your lips if they are easier to see. If, like me, your skin tends to get even more red than usual when you feel attention on you, wear foundation and powder to minimize how flushed (and hence nervous or embarrassed) you look. You don't need to look like Tammy Faye. Even just a subtle quantity of eyeliner and mascara and lipstick will make it easier for you to connect with your audience. That being said, you are on stage and it's actually hard to wear too much makeup on stage. Though if you don't want to look like a freak in the hallways, you shouldn't go overboard.

(Men: I challenge some of you to use a little makeup before a public talk. Even just a bit of powder to knock down shiny skin and prevent your specular reflections from drawing attention away from your eyes and mouth. The pros do it.)

For either women or men: if you have long hair, make sure it does not fall forward where it might brush a lavalier microphone.

QUESTIONS? COMMENTS?

I welcome both!

Further reading:

Scientists are Humans, Too

Giving an Academic Talk

Spectacular Scientific Talks

End of the Rainbow  

Advice on Giving a Scientific Talk