Lightning talks, and not…

Two years ago at SCaLE I attended a workshop on building ignite-style lightning talks. I failed abysmally.

This week at PyData Socal I have been referred to by several people in attendance as somewhat of an expert. There are two lessons here:

  1. I’ve put in a lot of hard work and practice.
  2. Most people suck at giving these, or have no clue at all how to start. When everybody else is blind, the one-eyed person is king.

A number of people at the conference I attended the past two days expressed abject fear of doing a lightning talk, even those who thought they were fine doing longer talks. But in my experience, the constraints of lightning talks make them one of the easiest types to pull together, and doing them builds useful skills for longer presentations.

I you can’t do a good lightning talk, you probably aren’t organized and focused enough to give a longer talk. It will end up a disorganized rambling mess. Even some lightning talks manage to be disorganized and unfocused.

What’s a lightning talk?

Wikipedia’s entry about them is a bit too vague. Every conference I’ve attended has meant a five-minute, time-limited presentation. Most conform to the “Ignite” style of talks, in which you have 10 or 20 slides, respectively either 30 or 15 seconds each, and no control of the presentation or projector. Some are a bit more relaxed but all are strict about the time.

A lightning talk is an opportunity to quickly present an idea, observation, or lesson in a manner that will get the key points across and (possibly) generate interest for people to do more learning on their own. It is not a forum for exhaustive exploration of a topic, and it is only suited to tightly-defined ones. Often lightning talks are an opportunity for conference participants to quickly bring up something relevant that isn’t addressed anywhere else.

What it’s not

It is not a second chance to give the talk that was rejected by the conference, by cutting it down to 5-10 minutes and attempting to rush through slides that were designed for in-depth treatment. I’ve seen people try and it never works.

It is also not an opportunity for lots of interactivity (live demos) or question-answer sessions. There just isn’t enough time. Again, I’ve seen people try and it always fails. For a couple of my talks I feign interaction by asking the audience a question and asking for a show of hands. But that’s just an effort to get people engaged. I know most of the hands in the room are going to go up.

What’s a good topic?

Good topics address very specific events, ideas or activities. It’s helpful to start with a question. For example “An intro to the open source project I’m working on” is probably not a good topic. Software packages are big and have lots of factors to them. Even a smaller one can be done no justice in five minutes, though I’ve seen people try. “How I solved specific problem X with open source project Y” is far more likely to yield a good result.

Good topics I’ve given or heard talks about include “You’re a Failure! Now what?” “How I got involved with the stupidest man in the world” (I had the misfortune of being tangentially connected to this debacle), “Operations lessons from a 17th century Samauri,” “4 steps for making great technical documentation,” “CS skills they don’t teach in school,” and “Sex, Secret and God: A brief history of bad passwords.”

You’ll note that most of the topics are tightly constrained. The samauri probably only had a few lessons relevant to computer ops, and you only need to pick the top three or four to fill five minutes; “4 steps” is constraining from the start; same for “skills they don’t teach in school,” as you don’t need to present all of them, just pick 3. The topics are often humorous. Lightning talks are not lectures and work best if they’re a light-hearted in a way that drives the key points home.

Some day I’ll give a talk about “how my vacation to New Zealand changed my work in tech.” The key points can be covered in five minutes.

How to approach it?

It shouldn’t be intimidating. If you do it well, it can become a fun hobby and it’s a good way to organize your thoughts about a topic even if you never give the talk.

Key points

Once you have a sufficiently narrow topic, identify the key bullet points for the story. In a 10-slide presentation you’ll have an intro, a summary, and 8 slides explaining how you got there. In a 20-slide presentation you’ll have 18 slides. You will need to think more about transitions, and about how to talk about a single point over several slides, which makes them a bit tougher. I think ten is the optimal place for most people to start if you have a choice.

Each of the major points will be covered in a slide or two in a 10 slide presentation.

Images or illustrations

Lightning talks go by fast. Don’t put lots of word salad on the screen. Many lightning talks put no words on the screen at all. Pick an image for each point and make that image your slide. Maybe add a caption or title. Don’t overthink it. A Google image search is usually enough. In some cases a flow chart, graph or other illustration might work well. Keep it simple, because it will be up for 30 seconds or less and people need to get the idea immediately. More than anything, you want to ensure that people remain focused on what you are saying, not on what the slide contains. If the goal is to share lots of detailed information, slides and presentations are the wrong tool.

Talk through it

Don’t worry about timing initially, just talk through the slides. This will help you notice things that don’t fit, aren’t in the right order, or are obviously missing. I’ve found it interesting how much I need to reorganize things when I’m talking through them rather than writing them. Our brains work differently when reading than when talking/listening. Rework the presentation to be a spoken address, not a written one.

At this point, you have a very workable presentation. It can take as little as 2-3 hours to pull it together once you’ve done it once or twice. I’ve seen people come up with a great presentation, having never done it before, in a 3-hour workshop.


You don’t want to spend your entire presentation looking at the screen. Practice what you’re going to say and do it with the slides advancing automatically at 15 or 30 second intervals. Adjust what you’re going to say about each. Trim it mercilessly if you need to. Not every minor point or anecdote needs to be squeezed in. It is far better to have a quick breather if you don’t have enough to say, than to find yourself still talking about the previous slide.

Why you should do this

All the rules for lightning talks apply to presentations in general. Avoid word-salad on the screen? Always a good practice. Well-defined topic? Essential. Lots of images and graphic aids? It’s a best practice. Rehearse in advance and get your timing right? All good speakers do this.

Lightning talks seem scary, but they are one of the best ways to introduce yourself to public speaking in general. Once you can do one of these, you can easily relax some of the rules for longer presentations, add in demos, Q&A, etc.

There are lots of forums for giving these. Just do it.

Here are some examples, some better than others. I attended all these in person. Not all are great presentations, but all adhere to the 5 minute rule and all the ones at SCaLE are “ignite” style talks with timed slides. What do you think will work for you? What never works?

You’re a Failure! Now what? (Yes, this one is me. It was given at a post-drinks, pre-afterparty event, so it’s mildly NSFW in a very juvenile way. I have a SFW version if you want me to give it.)

UpScale talks at SCaLE15x

UpScale talks at SCaLE16x

Lightning talks at PyCon 2017

More lightning talks at PyCon 2017