We’ve all been there. You’re sitting in a conference room at an AMS Sectional meeting, coffee in hand, ready to hear about some exciting new results in harmonic analysis. Twenty minutes later, you emerge confused, overwhelmed, and unsure what the speaker actually proved. The talk has ended, but you couldn’t tell anyone what it was about.
What went wrong? More importantly, how can you avoid inflicting this experience on your own audience?
Let’s start by examining the most common ways mathematicians sabotage their own talks. Then we’ll discuss what actually works.
The Gallery of Common Mistakes
The Information Overload Trap
You spent two years on this project. Your paper is 47 pages long with five major theorems, three key lemmas, and two important corollaries. Surely you can fit all of this into 20 minutes if you just talk fast enough, right?
Wrong. This is perhaps the most common mistake in mathematical talks. Speakers try to cram an entire research program into a short presentation, racing through results at breakneck speed. By slide 23 (of 50), the audience has mentally checked out. They’re not following the logical thread anymore because there are too many threads to follow.
The painful irony: when you try to tell your audience everything, they remember nothing.
The “Everyone Obviously Knows This” Fallacy
This mistake comes in two flavors, both bad.
Flavor 1: Assuming too much. You’re at a sectional meeting in analysis, so everyone must know the precise statement of the Calderón-Zygmund decomposition you’re using, right? You skip the setup and dive straight into “applying the standard CZ argument,” leaving half your audience lost because they work in a different corner of analysis where this isn’t “standard.”
Flavor 2: Assuming too little. You spend five minutes carefully explaining what a Hilbert space is to a room full of operator theorists. Your audience feels insulted and you’ve wasted precious time.
Even at specialized conferences, people come from adjacent areas. They’re experts, but not necessarily in your exact subfield.
Death by Dense Slides
Picture this: a slide appears containing 15 lines of equations in 10-point font, complete with multi-line display formulas, subscripts with superscripts, and no white space. The speaker says “as you can see here…” but you can’t see anything because you’re still trying to parse line 2.
Or worse, the speaker has essentially LaTeX-ed their paper directly onto slides and is now reading their paper to you, which you could have done more comfortably in your office.
Dense slides with tiny text don’t make you look sophisticated. They make your talk unwatchable.
The Proof Recitation
“So from equation 3, we apply the triangle inequality to get equation 4. Then we use Hölder’s inequality with p equals 3 halves to obtain equation 5. Substituting equation 5 into equation 6 and using the bound from Lemma 2.3…”
Stop. No one is following this, and even if they are, they’re not learning anything except that you can indeed execute a standard argument.
The audience doesn’t need to verify that you did the proof correctly. They’ll trust you on that (and if they don’t, they’ll read your paper). What they need is to understand the key idea that makes the proof work.
Vague Main Results
Ten minutes into your talk, someone in the audience is wondering: “Wait, what exactly is this person trying to prove?”
You’ve been talking about Sobolev spaces, maximal functions, and oscillatory integrals, but you haven’t clearly stated what your main theorem actually says. Or you stated it in such generality (“We prove optimal estimates in certain function spaces under appropriate conditions…”) that it’s meaningless.
If your audience doesn’t know what you proved, your talk has failed at its most basic job.
No Motivation or Context
“Let H be a separable Hilbert space and let T be a bounded linear operator…”
Wait, why? Why are we studying this particular operator? What problem does this solve? How does this connect to anything else people care about?
Starting with cold definitions and no motivation is like inviting someone to dinner and serving them the food with no explanation of what it is or why they should eat it. Technical sophistication doesn’t excuse a lack of context.
The Time Management Disaster
This comes in several depressing varieties:
- Spending 15 minutes on “background” (most of which the audience knows), then cramming your actual new results into the final 3 minutes
- Running over time and getting cut off mid-sentence by the session chair
- Finishing in 12 minutes because you prepared for a 30-minute talk and just… talked faster
All of these signal that you didn’t prepare properly or don’t respect your audience’s time.
Missing the Key Insight
You’ve walked through twenty steps of the proof. You’ve shown all the estimates. But you never explained the one clever trick that makes it all work; the geometric insight, the unexpected cancellation, the reason this approach succeeds where others failed.
Your audience leaves knowing you proved something but having no idea how you proved it or why your method matters.
So What Should You Do Instead?
Now that we’ve catalogued the mistakes, let’s talk about strategies that actually work.
Know Your Audience and Calibrate Accordingly
At an AMS Sectional meeting in analysis, you’re speaking to experts, but experts in a broad area. Someone working in harmonic analysis will know different things than someone in PDE or operator theory.
You can use technical terminology freely, but don’t assume everyone has memorized every theorem in your specific subfield. A brief reminder (“Recall that the Hardy-Littlewood maximal function satisfies…”) takes ten seconds and keeps everyone on board.
Choose ONE Main Idea
This is the most important advice in this entire post: pick one main result or idea and communicate it clearly. Not five results. Not your whole research program. One thing.
If your audience leaves understanding one clear mathematical idea and feeling motivated to learn more, your talk was a success. If they leave with a vague sense that you proved some stuff about some spaces, you’ve failed.
Twenty minutes is enough time to explain one thing well. It’s not enough time to explain five things even poorly.
State Your Main Result Early and Precisely
By minute 3 or 4, your audience should know exactly what theorem you’re proving. Put it on a slide. Make it visible. Say it clearly.
“The main result I’ll discuss today is the following estimate: the L^p norm of Tu is bounded by the H^s norm of u, for p in this range, with this regularity gain.”
Now everyone knows where you’re going. They can follow your setup and proof strategy because they know the destination.
Focus on the Key Insight, Not Every Detail
Your audience consists of professional mathematicians. They trust that you can execute a standard epsilon-delta argument or apply Hölder’s inequality correctly. They don’t need to watch you do it.
What they can’t figure out from hearing your theorem statement is: What’s the clever idea? What’s the geometric intuition? Where did previous approaches fail and how did you overcome that obstacle?
This is what you should spend your time explaining. “The naive approach breaks down because of these boundary terms. The key observation is that if we decompose the domain this way, we get unexpected cancellation…”
Use Visual and Geometric Intuition
Even in technical areas like harmonic analysis or operator theory, visual intuition helps. If you’re proving an estimate, can you sketch why it’s sharp? Can you show a picture of the domain where the interesting behavior happens? Can you illustrate why certain regions contribute differently?
Complex analysts especially appreciate seeing the geometry; where are the singularities, what paths are you integrating over, why does this region matter?
Don’t hide behind formulas when a picture would clarify.
Highlight Key Estimates and Where Techniques Matter
In analysis, the heart of most results is “this norm controls that norm under these conditions.” Make these estimates visible. Don’t bury them in prose.
Similarly, if you’re using a sophisticated technique, a clever commutator estimate, a maximal function argument, an adaptation of a tool from another context, emphasize this. Operator theorists and harmonic analysts care about methods. They want to know if your technique might apply to their own problems.
“The key step is applying Cotlar’s lemma in this non-standard setting” or “We use a Christ-Kiselev-type argument but with this modification” tells your audience something valuable about your approach.
Address Sharpness and Limitations
Analysis audiences want to know: Is your result sharp? What happens at the endpoint cases? Are the exponents optimal?
If you have a counterexample showing your result can’t be improved, mention it briefly. If you believe the result is sharp but can’t prove it yet, say so. This kind of honesty about the boundaries of your work strengthens rather than weakens your talk.
“We conjecture this extends to p less than 2, but our method breaks down because of this issue” is useful information.
Manage Your Time Ruthlessly
Practice with a timer. Plan for 18 minutes of content, leaving buffer time for transitions, pauses, and the unexpected.
A reasonable time budget for a 20-minute talk:
- 2-3 minutes: motivation and context
- 2-3 minutes: setup and notation
- 10-12 minutes: main results and key ideas
- 2-3 minutes: implications, related questions, or conclusions
If you’re running long in practice, cut content. Don’t talk faster. Speed-talking doesn’t save your presentation; it ruins it.
Keep Slides Clean and Readable
Use large fonts. Leave white space. Put one main idea per slide, not five.
If you have a complicated inequality, build it up step by step across multiple slides rather than dropping the whole thing at once. Guide your audience’s eyes to what matters.
And please, never put more than about 6-7 lines of mathematical content on a single slide. If you need more than that, it should be multiple slides.
Connect to Known Work (Briefly)
Your audience likely knows recent important results in the area. Brief connections help them place your work: “This generalizes Smith’s 2023 result by relaxing the compactness assumption” or “This provides an alternative to the approach Jones considered.”
Keep these connections brief, a sentence or two. The goal is orientation, not a full literature review.
Invite Questions and Engagement
If the format allows, leave a minute or two for questions. At sectional meetings, questions during or after your talk are a sign of engagement, not hostility. Welcome them.
If someone asks for clarification, answer clearly and briefly. If someone points out a connection to their work, acknowledge it graciously. This is how mathematical conversations happen.
The Bottom Line
A 20-minute math talk is not a compressed version of your paper. It’s not a complete proof verification. It’s an invitation.
You’re inviting your audience to engage with an interesting mathematical idea. You’re showing them something beautiful or surprising or useful. You’re giving them enough understanding that they could explain your main result to a colleague over coffee, and enough motivation that they might want to read your paper or think about related questions.
If your audience leaves your talk understanding one clear idea and feeling curious about your work, you’ve succeeded. That’s the goal. Everything else is in service of that.
So next time you’re preparing a 20-minute talk, remember: less is more, clarity beats comprehensiveness, and the clever insight matters more than the technical details. Your audience will thank you for it.
What strategies have worked for you when giving short conference talks? What mistakes have you learned from? I’d love to hear your experiences.