SRECon Fieldnotes: Storytelling and Rhetoric in Conference Talks

Published: 2024-04-11

Tagged: essay rhetoric thoughts

I attended 18 talks at SRECon this year. In a normal year, I attend 0, so I sat and listened with greedy attention. Each one was fascinating, but some worked better than others--a few even attained "fond memory" status! Thus, on the flight back home, I pondered: What makes a conference talk stand out?

Two ways of getting the answer occurred to me, both equally good I think: storytelling and classical rhetoric.

But first, a quick disclaimer: we're all Internet strangers, so you probably don't know that my body mass is mostly pure, USDA-approved kindness. So know that my aim here is light, not heat. It takes guts and effort to give a talk, and each speaker referenced here has my respect and thanks.

Talks as Stories

Broadly, talks that sparked less interest than others didn't work well as stories.

Take, for example, the Shopify talk wherein the speaker described how large search datasets were migrated into new regions. It went something like this: Shopify wanted to spread out its data to reduce latency, so we looked at options A, B, and C, and we thought hard and decided on C, and then we built it, and after fixing some bugs it worked, and our customers were happy.

Now, I'm sure this undertaking featured many an exciting twist, but the speaker made it seem sterile, like an example from a textbook on system design interviews.

The Slack talk serves as great contrast to this. Here, an engineer shared an exciting tale: We had a series of incidents with chef, so we set up a manual approval process, but it was slow and frustrating, and for me it even made the vibes worse, so I vented to my manager, but she turned around and asked, "Why don't you do something about it?", and I wanted to, but I felt unsure, so I... (watch the talk!)

Where the Shopify story was more like a timeline ("A, and then B, and then C..."), the Slack one zigged and zagged ("A, but then J, so K, but suddenly C..."), each turn upping the stakes and drawing the audience in deeper, fanning our curiosity. I couldn't help think how truly it reflected real life, what with all the uncertainties present in larger software projects.

What also made the Slack talk effective was how much detail it included. The speaker shared dates, hard numbers, and screenshots of Slack messages. All of that gave the narrative some hard edges--characters weren't abstract personas but actual people; events occurred in a messy, confusing reality and not some generic scenario.

However, it's possible to overdo details. An interesting example of what not to do appeared in the Atlassian talk about synthetic testing where the speaker spent a few minutes describing the concept of the testing pyramid. Given how popular this idea is, I'd expect most of the audience knew it. And given how simple it is, I'd expect everyone else to quickly figure it out. So it's critical to include only such details that keep the narrative advancing smoothly.

Drowning the reader in too much detail is actually a common mistake among newbie writers--especially those that dabble in fantasy settings. After building out a whole world, it's tempting to open a story with a massive descriptive dump about it: the kingdoms, types of dragons, rules of magic, royal lineages, etc. However, doing so repels readers because it denies them a character they can care about.

Anyway, the point is that a good story makes for a good conference talk. There are plenty of books and courses for learning about that. But I find it more interesting that we, being creatures that think in narratives, already possess powerful tools for judging and producing stories.

So, next time I'm putting together a presentation, or even demoing something at work, I'll spend some minutes trying to imagine how my colleagues would react to it. Would they laugh? Would they listen intently? Would they scroll reddit on their phones?

Talks as Speeches

Another way to analyze talks is through classical rhetoric.

Aristotle described three means of persuasion: ethos, pathos, and logos. (There is also kairos, stasis, and a whole lot more, but the three are sufficient for our purpose.) Think of them as dimensions, like height or width, that outline the overall shape of a presentation. And to be clear, persuasion here is less about selling used cars and more about conveying ideas in a way that makes them amenable to critical thinking.

Ethos

The first dimension, ethos, concerns trust. After all, why should anyone listen to what a stranger has to say? We're all constant targets of manipulation, and we've all developed more or less effective defenses against it. So an important objective for the speaker is to prove to the audience that their minds are in good hands.

Everyone giving a talk at SRECon gets a little help just because they've passed the selection process. The attendees trust the organizers to filter out irrelevant or even malicious speakers. But that ensures just a few minutes of good will at the beginning--if these aren't used well, people will (and have) switch to their phones and laptops or go chat in the hallway.

That's why most speakers start off with introducing themselves: They're showing the audience their qualifications to give this talk. That relationship should be proportional, meaning, the bigger the topic, the stronger the qualifications. Consider what happens when this rule is violated. High qualifications, but trivial topic is a recipe for boredom. But flip that around and you get "crazy uncle talking about strange economic theories at the holiday table."

Pathos

The second dimension, pathos, is about emotions. It forms the core of modern advertising. Ads no longer sell us on a product's benefits, but on the feelings associated with the brand. Something as simple as water can't be just pure and fresh; it must make its consumer a more productive or virtuous person as well.

Thankfully, such ugly manipulation doesn't take place at a tech conferences. But what does take place is a ritual of emotional attunement between the speaker and the audience.

Honestly, I don't feel qualified to talk about this aspect, except for one aspect of good talks: humor. Humor is amazing at creating a connection between the speaker and the audience.

Some speakers tried to achieve connection in other ways. For example, by asking for a show of hands, eg. "raise your hand if you've been paged at 2am. Ok, now keep your hand raised if the alert resolved before you booted up your machine." Doing this helps listeners focus and creates a good vibe; it's nice to be part of a bigger group doing the same thing (like: dancing, cheering, etc.)

But humor was an order of magnitude more powerful. When a speaker executed a good joke, you could feel the air vibrate. I remember thinking, "wow, I'd love get a beer with him/her." But above that, hearing so many other people laughing around me made me feel part of something greater. It turned a presentation about cloud costs into a wonderful shared memory. I suspect it also helped the audience stay in lockstep with the speaker--people have to be listening and following along for a joke to work.

Logos

Finally, we have logos, or persuasion through argument. Here, the objective is to arrange ideas to make cause & effect clear, and to furnish the audience with sufficient data to judge the topic for themselves. It's something we've all been taught to do since our earliest days in school. Partially because it's the foundation of maturing as a human being, as in moving from believing in Santa Clause to believing in thermodynamics. And partially because it's so easy to discern and grade.

In the context of a talk, the greatest risk is the curse of knowledge, or the assumption that the audience knows everything you know. If it's an acronym not known outside your company, that's just annoying. But if it's a larger piece, and one that's crucial to the point of the presentation, like why someone decided to move from cloud to on-prem in a talk about hosting costs, that can be downright confusing. That can instantly disengage any listener.

Examples

Going back to the Atlassian talk about synthetic testing, there's a tension present. Explain too much, and you're boring the audience. Too little and you're confusing them. I don't think there's a universal principle to get this just right, except doing a bunch of test runs with one or more groups that are similar to your expected audience. Of course, if you're an extrovert that haunts every conference, you might already have developed a feel for this.

Because examples are so helpful in getting ideas across, let me backtrack a little to ethos and pathos.

Compare the ethos of the Slack and Shopify talks. In the first one, the presenter was also the protagonist of the story; everything she described, she interacted with directly. That lends her credibility because it's hard to "touch" something and get it totally wrong. We may wonder if she might have missed some detail, but we can always ask about it at the Q&A.

In contrast, the presenter of the Shopify talk covered a much bigger project. She probably did not inspect every commit or attend every meeting, so to trust her, we need to also trust all the people she interacted with it. That's hard to do when they're not on stage right there. However, the speaker managed this by strengthening logos--specifically, going into deeper detail on how they decided on the right replication scheme.

An interesting example of when ethos doesn't work was in the "Resilience in Action" talk. The speaker chose to forego an introduction and proceeded to talk, authoritatively, about some pretty abstract concepts. Despite the fascinating topic and some juicy insights, I found myself growing increasingly resistant. "Who is this guy to so definitely tell me what is or isn't true?", I remember thinking to myself.

The best example of deploying humor is "Scam or Savings? A Cloud vs. On-Prem Economic Slapfight" by one Corey Quinn. Going in, I wasn't expecting much. Pricing and cost decisions? Thank the gods I had some coffee beforehand. But then Corey made everyone laugh. And then again. And again. And oh boy, I learned more about the topic in just 25 minutes than in a year or two of lazy observation.