2026-05-15 19:11:06

Obviously it depends on the type of presentation, but here's what I currently do for my talks:

Slides should only have images and headlines, not reading-text. People will naturally read whatever text is on a big screen, and they will not listen to you while reading the text, so anything more than ~5 words on a slide is going to kill the mood in the room while people read your slide instead of paying attention to you.

Put timestamps in your presentation notes. Decide in advance how long your presentation should be, time yourself giving the talk beforehand (multiple times), and put timestamps in your notes e.g. [2:00], [3:00], [4:00] at the relevant spots in the slides.

The point is not too hit those marks on the day– you will inevitably talk slower or faster at times, start late, get on a tangent, skip a slide, whatever. The point is to know how you're running relative to standard, like the pacers in a marathon, and then know when you need to slow down / speed up accordingly.

Don't panic. Like parties (and babies), audiences react to your stress. Put serious attention into feeling relaxed so that the audience feels relaxed with you. Also, the audience has no idea what you were planning to say, so it's far better to skip a section or fudge something but stay calm than to stress the audience out by spending a bunch of time backtracking.

This is also important in case of tech issues, which are legion. Whatever happens, just smile and roll with it. On that note:

Set a person in the audience who will tell you if something bad is happening technologically, and assume otherwise everything's fine. It's bad if the speaker is inaudible, but it's also bad if the speaker keeps stopping to ask "can everyone hear me?," or gets flustered about the fact that they think people can't hear them.

Microphone and speaker technology are weird, as professional musicians know. I once gave a talk where, due to the audio setup, I was hearing myself echo on every word; I briefly thought about stopping the talk to check about that, but that's exactly the kind of stressed-out behaviour that stresses audiences. And when I asked my friends afterwards they had no idea what I was talking about it, it was just an issue for the speakers on stage.

It turns out the solution to all of this is to have ONE person (in the audience, or running the talk) who has solemnly sworn they will give you a signal if there are any issues, then you as speaker assume there are no issues until told otherwise. Thanks to the excellent NerdNite for showing me how this works.

Practice a ton, to the point where you've given the entire talk multiple times in full before you're doing it live for an audience. I'm repeating this one because it's probably the most important.

In some sense I feel like "practice a lot" is obvious but I also see a lot of people not-doing it, and I'm not sure I fully internalised how much you need to do it until Friend-of-the-blog N. gave me advice before a talk once.

For my most recent talk, I downloaded the screen recording software OBS and did four full run-throughs of the talk in the week leading up to it. I didn't watch all those recordings in full, but it was super helpful both in terms of helping me feel calm with the material and in terms of figuring out the structure and timing of the talk.

I think there's all kinds of cope that people tell themselves about being "better when improvising" or whatever, but I mostly think we wish that were true because practicing is annoying and hard. I do think that talks benefit from some live flexibility – I wouldn't want to watch someone recite a talk word-for-word from memory – but realistically you're going to be a better Flexible Improviser if you've got a really solid foundation first. So have a clear plan for your talk and get to the point where you can do that version unhesitatingly, even if you later deviate from your plan on the day.

2026-05-14 19:11:45

In my life I have struggled mightily to find situations that allows me to spend my time working at 95-105% of my sustainable long-term capacity.

I have had situations where I worked at 120% of my sustainable long-term capacity, and burnt out.

I have had situations where I worked at 20-50% of my sustainable long-term capacity, and was bored.

Where do I find the situations that lets me work hard but not unreasonably hard? Why is this so hard to find?

2026-05-13 19:11:33

One of my long-time hobby-horses is that people are way weirder than they realize, and that our politics would look different if more people understood how massively unrepresentative they are of the rest of the population. So here's a little quiz designed to demonstrate how unrepresentative you are (of the US population – sorry non-Americans).

I did shockingly little of the work on this, so compliments should largely go to Claude, but if you have complaints or think this is wrong please do let me know. If you're interested in the process you can see the conversation here.

2026-05-12 19:11:33

1) Evolution of species by natural selection is the classic of the genre, of course. If you have a bunch of finches trying to drink nectar out of flowers, the finches with the long pointy beaks can more easily get deep in those nectaries, and over time the finch population becomes dominated by the long of beak.

2) I think some of my more anti-capitalist friends are flabbergasted at the idea that the invisible hand could possibly guide businesses to figure out what customers want. If you've ever been inside a business, that doesn't seem to be how it works! But that's not the theory of capitalism, really: it's that if a bunch of different businesses try to please customers, even if they're all wildly flailing around at random, the ones that happen to figure out What People Want will stay in business, and the ones that don't will die out.

3) What we post online in any algorithmic environment is subject to brutal evolutionary pressures. Even here on my relatively insulated blog, it's hard to ignore the incentive to write what people seem to enjoy. E.g. you guys love lists, empirically, and all my most popular posts in the last few months started with a number, so here I am writing another list. But this effect is 1000000x fold on a platform like Instagram or Twitter or YouTube, where your most popular offerings can get 1000000x the attention. Soon you find yourself desperately iterating towards whatever the algorithm most rewards from you, until one day you wake up and you're posting bulls**t rage-bait because your audience applauds when you do.

4) What I don't think is nearly as heavily acknowledged is that the same pressure exists in our non-algorithmic environments, like our everyday lives. Every time you say an opinion at a party you get feedback based on other people's responses; without even necessarily knowing it consciously, you get trained that some opinions are far more popular than others; unless you really make huge and deliberate efforts, the natural glide-path is towards only saying the views that get you good responses. It's like social media just without the quantification, and with real people whose feelings about you matter immensely, and who are more likely to be selection-biased to have similar views to each other versus the diversity of the internet as a whole. I sometimes wonder if the only progress in society is made by disagreeable people who are able to withstand the crushing social pressure to only say (and think) already locally-popular things.

5) At least in Europe, professional sports leagues are evolutionary: the teams at the bottom of the league get demoted, the teams at the top (of the league below) get promoted in their place. American sports somehow don't function this way, they're like a managed zoo where the animals are insulated from evolutionary pressures, I am pretty confused about it.

6) Corporate politics is all about evolution. The most successful managers sneak more and more of their proteges into the firm, the least successful managers get booted from the firm and don't get to hire any more people in, and can't protect the people they hired in previously. Fitness is determined by your fitness at playing the game, not at how good you are at [thing the organization was supposedly set up to do].

7) Political parties get formed by evolution: the ones whose platforms are unpopular wither away, the ones whose platforms are popular live to fight another election. And this is true fractally for coalitions within each party, and politicians within each coalition. The upshot is that, over time, parties can evolve beyond recognition as they seek to respond to the changing environment and survive within it. And as always with evolution, this doesn't necessarily mean the actors are manipulative or even conscious of what they're doing: it could be that 100% of politicians are 100% sincere in their beliefs, but over time the politicians with unpopular beliefs will get selected out of the party, until the only people left are the ones who hold the views that the majority of the selectorate endorses. Survival of the fittest, where "fitness" doesn't mean "good" but just "good at surviving."

2026-05-11 19:11:53

Look, I realise this is an oddly specific gripe but I do feel it strongly whenever I see a highway or a public building and the name of a politician (often a Mayor, sometimes a Governor or President) on it: this just shouldn't be allowed. The politician did not pay for it, the public did. And it's naked electioneering on behalf of the incumbent, trying to get voters to associate the incumbent's name with public spending, which is unfair on the non-incumbents who will sooner or later be trying to unseat her. It's nakedly against the spirit of liberal democracy, I don't know where it started and I don't know why we put up with it.

2026-05-08 19:11:49

For most newsletters, I would like to receive neither All Newsletter or No Newsletter, but Some Newsletter – depending on the author, I'd like to get 10% or 20% or 50% of their best posts.

I don't think as a writer I have any way of knowing which posts are great. (There's definitely posts where I feel confident they are not-amazing, I just want to get that idea out my head, but I don't think I've ever correctly predicted a break-out post would break out).

Here's how I wish the newsletter-sending business worked: I would just mail out whatever's in my head, at whatever volume, and you all would opt in to a certain threshold of email receipt. You could choose whether you want to get 100% of my posts, or only the best 50%, or only the best 10%.

Each day the newsletter would first go out to the people who agreed to receive 100% of posts, and they would vote on how good it was, and only if it crossed a certain threshold of positive votes would it get sent to the 90%ers, who would then vote on how good it was, and so on till it had either hit its quality-bucket or (in the rare 10% of cases) been sent to the whole list.

If you're worried that superfans have different tastes than the average, you could instead set it up so that each post gets sent to a random 10% of the audience, that 10% votes, and then the post gets sent out to a suitable share of the remaining audience. So it's no longer the case that you can get only the best N%, but you can get the best N% plus some ungraded posts, where your contribution to the project is to grade them.

Another theoretical way to do this, especially for a paid newsletter, would be to have a small pool of people who volunteer as the Test Pool (in lieu of paying for the newsletter), and they vote on the quality, and then the curated selections get sent to the paid reader pool. I'm not sure how you could possibly incentivize the voters to vote accurately, though – famously the Oscar voting was historically full of people who had not-seen the movies voting on which was the best one, and that's in a setting where the result matter a lot.

One problem with this system is if/when people have idiosyncratic views about which posts are best. This seems obviously true, some posts will resonate with some people and not others, but I'm not sure how to avoid that. At the extreme you could imagine a recommender algorithm, "people who enjoyed Post X also enjoyed Post Y", but 1) that totally seems like too much work for a simple newsletter, and 2) maybe it's just reinventing social media from first principles?

I'm pretty sure I wrote this idea before, maybe in an Is This Anything? post, but since I'm not sure maybe you won't be either.