As I mentioned previously, I attended part of the Quark Matter 2012 (QM2012) conference last week. (My full write-up can be found at Ars Technica.) The conference ran all week, so there was no possible way I could attend the whole thing, so my sample of talks was smaller than at Pheno 2012. However, just as I found myself frustrated with the quality of the presentations in Pittsburgh, I think I heard the worst scientific talk of my career last Monday.
I admit, I was listening with a significant disadvantage. Physics is a huge discipline, and while I like to think I know a fair amount of it, heavy-ion physics is not very familiar to me—and QM2012 is a conference entirely devoted to that subfield. The presentations are from specialists to other specialists. In other words, I wasn’t the true audience for any of the QM2012 scientific talks. Nevertheless, I still think scientific talks should be held to certain standards, even if they are technical and aimed at a specialist group.
I wrote extensively about this topic in May, so let me paraphrase some key points from my earlier post:
- The total amount of information on each slide should be small if you have any desire for your audience to absorb it. That means a minimal amount of text printed in large, sans serif font. While the talks I saw at QM2012 didn’t commit the Deadly Sin of putting every word the speakers said on the slides, every talk did have too much text, and with one exception the fonts were all too small to be legible beyond the first few rows of the very large conference room. That’s even with two huge projection screens! Lest you think this problem was mine alone, when I discussed matters with my fellow conference attendees, they all told me they couldn’t read the slides from the back half of the room. At all.
- I think we all agree that plots are essential for conveying quantitative scientific information. However, the plots in the talks I saw were generally unreadable: the lines were too thinly drawn, the labels on the axes and legends were too small, and data points were represented with symbols too small to distinguish from each other. Since the plenary sessions were designed for the express purpose of announcing the major new results from RHIC and LHC, having unreadable plots made the talks of little value to the audience.
So what about the “Worst Conference Talk Ever™”? Nearly every sentence was unintelligible. Not because of the person’s accent—I will not fault non-native speakers for difficulty in speaking English, since they all speak English better than I speak their languages, and many scientists from other countries are multilingual. No, the speaker compounded the problems listed above by mumbling everything, and not speaking into the microphone so what was mumbled often didn’t even come through the public address system. Since the person’s slides were badly designed (thin gray lines on the plots, tiny fonts), the audience couldn’t get the information from either the audio or the visual. In fact, the crowd of physicists, specialists in the speaker’s own field, were not paying attention to the talk.
A point I made earlier is worth repeating: the few talks I heard at the plenary were supposedly the big announcements of the conference, the summaries of the latest results from the world’s biggest heavy-ion colliders. The speakers were representing large collaborations involving hundreds of researchers, presenting the fruits of years of hard effort—and the audience couldn’t read their slides. While the immediate comeback is that all the information is available online or in the lab databases, my response is, “What’s the point of a conference talk then?” I don’t care what you’re presenting: if the audience can’t read your slides or understand your delivery, you’ve failed—and you’ve let down the people and projects you’re representing.
Why it matters
The keynote address for the whole conference was former U.S. Representiative Bart Gordon, who served 26 years in the House of Representatives. During his last two terms, he was also the chair of the House Committee on Science and Technology, which obviously placed him in an important position with regard to science policy and funding. His speech was definitely catering to his audience, aimed at showing his continuing support (even while out of office) for big scientific projects, and highly critical of reckless budget cuts to research. (Hey, I didn’t say I disagree with him on that stuff!) He stressed the importance of outreach by scientists to politicians and the general public, saying “It’s often challenging to explain the implications of basic science and research.” He emphasized that politicians are locally focused, mostly interested in what benefit ideas have for their own districts, while good science relies on international cooperation.
The sale of science to non-scientists isn’t always easy, but let’s look at it another way: these colliders are creating quark-gluon plasma. That’s a substance that existed in the Universe’s first moments—and possibly only in very special environments since that time, if then. Our experiments have dissolved atomic nuclei into their constituent parts. While the conference attendees didn’t necessarily need to be reminded of how sexy that is, the public does need that knowledge.
If scientists can’t communicate the basic information of our own research to each other in a supportive environment, we as a group will fail to convey why it’s important to the wider world. If we can’t present the results of major experiments in such a way our own colleagues can comprehend it, we’re failing at a basic aspect of our work. While not every scientist will be a great communicator, it’s well past time for thinking that it’s OK for scientists to suck at presenting. Our work is too important to leave to the “experts”.
Galileo's Pendulum 
12 responses to “The Terrible, Horrible, No Good, Very Bad Scientific Talk”
I prepare material for a TV program regularly. Images have to be 704×480 pixels. It turns out that the analog video reduces the viewable area even more, to maybe 680×440. Fonts smaller than 18 point aren’t readable on the final TV. So that’s the minimum. It’s not a bad way to produce slides for a talk. My goal is to have no text on images. Can’t always do that.
I would say 18 point is probably too small for talks. My general rule is if you can’t read it on your laptop monitor from 10 feet away, increase the font size!
When giving talks to non-professional audiences, I fully agree with the minimal-text rule. Some text is *necessary*. For example, the name ‘RHIC’ should certainly appear on a slide about the research projects being discussed here, and indeed ‘[Relativistic Heavy Ion Collider]’ probably should also appear, so the audience knows what we are talking about.
Different rules apply in the professional context. In particular, it is common to hand out paper copies of slide presentations. In such cases the talking points should be in a big font, but selected details can be provided in a small font, and noted while talking, on the assumption that anybody who cares will have the paper copy. For example, URLs for the details and literature citations can be in those small font details.
A scientist who reads a talk in a dull monotone voice, mechanically, is likely to lose my respect. My attitude is that if their project is worth doing at all, they ought to be able to show some enthusiasm about it! If they expect the public to pay for their project (and their salary), they should accept the responsibility to do a sales job. Actually, even in a professional context, if they expect their peers to rate them and their project well for the process of deciding which projects will get funding (or publication), they should do a sales job.
Why not name the speaker?
Partly because I don’t want to make this about shaming an individual. After all, the speaker’s sins are just more extreme versions of the (nearly) universal sins of scientific presenters. If I named the speaker, it would seem like a personal rather than a general critique. If you *really* want to know, you can go back through my tweets from last Monday (August 13) and find the references to a mumbling speaker.
I generally agree that presentation skills are more critical for scientists than the popular myth would have you believe. I’m a little surprised by the description here, though, because my experience in my own field of physics (Atomic, Molecular, and Optical) is that there’s a reasonably good correlation between the quality of the science and the quality of the presentations. That is, if I were to sit down and make a list of the top twenty or so scientists in DAMOP (in terms of producing impressive and influential scientific results) and the top twenty or so speakers (in terms of giving reliably interesting and informative talks– the sort of speakers I would advise an undergrad student to seek out and listen to, or would invite to give colloquium talks at work), there would be probably fifteen names appearing on both lists.
Maybe this is a difference of large collaborations vs. smaller individual labs. That is, in a field with hundreds of scientists on a single project, maybe the skills that put a person into the position to be the one reporting significant new results don’t necessarily include the ability to give a good talk about the results. Where in a smaller lab situation, the need to compete with everybody else for the grant funding needed to do good science forces individuals to be more skilled at presenting their results (where in a large collaboration, it may be possible to delegate that to someone else).
Or, possibly, the need to address the pet issues of hundreds of scientists from a large collaboration has a detrimental effect on the quality of the talk. That whole “designed by committee” thing.
“Not because of the person’s accent—I will not fault non-native speakers for difficulty in speaking English”
Having an accent is not the same as difficulty speaking English. I have no difficulty speaking English. Your accent is just different from mine.
[…] recently read a post over at Galileo’s Pendulum that discussed the most terrible scientific talk ever. There are many general lessons that one can […]
I wonder how hard it would be to go past our standard “A4” format for presentations. These days, high school students make 3d displays and animations. University research groups often have a technical person who builds interactive web apps.
I went to a talk recently that did not use slides or powerpoint. Instead, the presenter connected his monitor to the big display. He showed people the data then displayed different visualizations, based on audience input. It really was much better than sitting through a series of bullet points.
Completely agree with the author.
I personally prepare my talk in LaTeX using the prosper-based style:
it generates high-quality PDF slides that can be shown in a full screen mode. All coloured text I put in a bold-face sothat it can be seen quite good.
My graphics I prepare using Wolfram Mathematica program and usually I put specification fot the line thickness (PlotStyle->Thickness[0.007]) — this produces output graphics in eps-files with resolved lines.
[…] The Terrible, Horrible, No Good, Very Bad Scientific Talk […]
[…] The Terrible, Horrible, No Good, Very Bad Scientific Talk (0) More » […]