Weekend Reading & Selected Links
Happy weekend! Here are some links to things I've been reading or watching that you might also enjoy:
- My 2025 retrospective, in which listener of the show Zac Gross interviews me. At the bottom of this email, I've included three excerpts.
- Dan Wang's annual letter.
- The Thinking Game, documentary about DeepMind.
- Financial Times interview with Yann LeCun.
- How does Gavin Newsom memorise policy details while having dyslexia?
- Josh Gans on using AI for research.
- Brian Potter on modular housing.
- Joe Carlsmith on fake thinking and real thinking.
Thanks, and best wishes for the year ahead,
Joe
Three excerpts from my 2025 retrospective
1. How will podcasts change?
ZAC GROSS: Derek Thompson, formerly of The Atlantic, now at Substack, I believe, outlined an interesting theory that basically all media is becoming television.
For a bit of behind the scenes — we actually spent much more time on the cameras and the lights here than on the microphones, so that you can all see us. And that’s a reflection of the fact that all podcasts are moving to video. The pivot to video is happening.
How has that affected the business of making a podcast?
JOSEPH WALKER: For me, in two ways.
The first — and most obvious — is just doing video, and needing to do video to grow the show. So I started doing video for every episode last year, and I publish video on YouTube, Substack, and X (formerly known as Twitter).
That has turned out to be kind of mandatory if you’re trying to grow a podcast, because you have access to the upside of those algorithms on those platforms. The discoverability on a platform like YouTube or X is just qualitatively different to the discoverability through a podcast app like Apple Podcasts or Spotify.
I don’t remember the last time I ever used Apple Podcasts or Spotify to find a new podcast. Usually you just learn about stuff through word of mouth, and then subscribe through those apps and keep listening to your same stable of shows. But you don’t use those apps to find new things.
I think the second way in which I’ve noticed the “everything is television” thesis apply to my show is doing ads in more creative ways. And I think this will be a theme for podcasts that we’ll see intensify over the next few years.
It’s no longer enough to just do a read where you’re reading the script the sponsor’s given you and you’re just plugging a product. I think ads will get increasingly more creative, much more personalised to the host and the show, and much more seamlessly integrated into the show.
And that, I think, has been facilitated and driven by the video medium, because there’s so much more you can do with the ads when you’ve got that video dimension.
One example from my show this year was an employer-branding ad for Eucalyptus, which appeared in the Hugh White episode. So I interviewed someone who’s a long-time listener of the show who now works at Eucalyptus.
Instead of me just singing the praises of working at Eucalyptus, the ad was a mini-interview: a two-minute interview where I interviewed that person, Jae, about what it was like to work at Eucalyptus and why they chose to work there.
So that’s an example. We’ll see more and more of that. And I think that’s a special case of the “everything is television” phenomenon applied to podcasts.
GROSS: And it even extended to having a live studio audience this year. Do you see more of that as well in the podcast future?
WALKER: Yeah, I think so.
And I think we’ll start to see a lot of innovation around what a podcast is. So it won’t just be a sit-down interview. There’ll be other elements blended in. It will become more sort of documentary-esque.
GROSS: So like maybe charts in the background, or—
WALKER: Exactly.
GROSS: Flashbacks to a previous episode.
WALKER: Exactly.
One thing I’ve been wanting to do for a while is: podcasts aren’t always the best medium to explore very difficult ideas. And having someone speak extemporaneously about a topic is a very challenging exercise for them. And there’s only so much you can do.
Wouldn’t it be cool if you could cut in other explainers? Or if you were talking to a guest about a very technical topic, maybe you could step aside and they could explain it on a whiteboard or something — and then you could cut that in. That kind of stuff.
I think we’ll start to see the definition of what a podcast interview is start to blur and change and morph.
2. The "dark matter" of Australian state capacity
GROSS: There’s a famous study in economics that looks at the correlation between height and performance amongst NBA players. And perhaps surprisingly — at least surprisingly for a non-expert like myself — they actually find a negative correlation: that within the NBA, shorter players tend to have slightly higher performance than taller players.
And this is not to say that height isn’t important in basketball — obviously that’s not true — but it’s an example of what’s known as selecting on the regressor.
If you only look at a portion of the sample that’s at the top of the field, you might be missing dynamics that filter through the rest of the population.
And so one question I had was: has the series of investigations into state capacity suffered from a similar issue? Because we’re looking all about Australia. We’re looking at the NBA of high state capacity.
Is it possible that instead of comparing, say, different prime ministers, different states, different departments, we might have come up with different answers if we’d looked at, say, the whole OECD and said: what makes Australia so much better than the UK, for example, the US, or France?
WALKER: Yeah, I think that’s a really fair criticism.
Peter Bowers and I, in our literature review on Australian state capacity, did a bit of comparative analysis where we looked at Australia against other OECD countries. So I feel like, outside of the podcast, I’ve thought about that.
But yeah — it’s dangerous to select on the regressor.
GROSS: Always risky.
And this is perhaps a bit more speculative, but: if you were to think about what is the “height” of state capacity — what is the one thing about Australia that it’s so ubiquitous, we have it so deeply, which makes us a fairly well-run country — but it’s so ubiquitous that we don’t even notice it’s there?
Do you have any thoughts about what such a thing could be?
WALKER: So what is “height” for Australian state capacity?
For me, the “dark matter” (to use a physics metaphor) of Australian state capacity is the political culture of faith in the administrative state and obedience towards impersonal authority. Maybe you could summarise all that in one word as statism…
When political scientists look at state capacity, they usually differentiate three different dimensions.
First, extractive capacity: how well can the state extract resources to do what it wants to do — through things like taxation.
Second, coordination capacity, which refers to the kind of Weberian bureaucracy.
And then finally, compliance capacity.
We’ve got a lot of all of those, but the compliance capacity really stands out. There’s a population that trusts its government, that complies with its government.
You know better than me from the Victorian lockdown experience. But when the government asks things of the population in Australia, people tend to go along and make things easy.
That would be the first thing I would think of. How does that sit with you?
GROSS: I think that makes a lot of sense to me.
In fact, I was thinking about how would we multidimensionalise state capacity as a concept. And I think when you look at the body of work you’ve done, there are some aspects that Australia is clearly a world leader at.
So one aspect that I think we do well at is big, technologically complicated projects — such as Single Touch Payroll, or the whole myGov system — which I think actually work really well.
I’ve lived in the UK, and we do much better than them at all these sorts of big technological projects. The Reserve Bank’s payment system, I would add to that.
One area where I think we probably fall down — and this is probably the flip side of that high compliance — is that sometimes we set up a bunch of experts to run public policy and we don’t question them if they start to go off and start making policy errors or start making mistakes.
The examples I’d cite for this are like the planning system. We’ve devolved planning to a whole bunch of architects and urban planners, and I think it took decades for us to wake up to the fact that they were probably stoking a massive housing shortage.
I could tell similar stories — and some of your guests have — about the Reserve Bank making mistakes and taking a long time to be corrected.
So I guess: it’s great to have a compliant population when you want to put them in lockdown, but perhaps the severe cost of that is: when you say, “Look, the experts have it right — don’t bother questioning them,” then even when they make mistakes, it takes a long time to pick up on them.
WALKER: Yeah, yeah. I think that’s a good point.
GROSS: So maybe this podcast is part of the antibody to that — because what you want to do is have experts like Richard Holden and Steven Hamilton on. They’re very good at shouting out when experts are making the wrong decision about, I guess, vaccine procurement.
And giving them a bigger microphone is part of trying to right the balance of that ship, I guess.
WALKER: Yeah, exactly.
3. My biggest intellectual update via the podcast in 2025
WALKER: The Peter Tulip episode was the site of my biggest intellectual update this year.
GROSS: All right. Tell me more.
WALKER: So for me, this was like a huge red-pill moment. No-one else seemed to pick up on it.
This was the highlight I pulled out in my “Eight Things I Learned” roundup of the policy salon series.
But the question I put to Peter was: if we could implement a maximally ambitious deregulatory agenda — so get rid of all of the kinds of planning restrictions he’s worried about — how long would it take for supply to accumulate sufficient to bring prices in Sydney and Melbourne down by, say, 30 to 40% relative to [today’s prices]?
And you can argue whether 30 to 40% should be the goal, but I guess on common measures of affordability, that is what you would want to aim for to make housing affordable in Australia.
And Peter’s answer was: 10 to 20 years. And he kind of extrapolated that using his and Trent Saunders’ rule of thumb around: a 1% increase in supply reduces prices by 2.5%, or something like that.
That was a huge red-pill moment for me because I think it showed me that this problem is not going to be solved any time soon — maybe even within our lifetimes, realistically.
And that was on a maximally ambitious deregulatory agenda, which is much more than the National Cabinet’s target of 1.2 million homes over five years — and even that we’re not going to achieve that. That is a very ambitious goal.
So that just showed me, I think YIMBY is good at the margin, we need more homes, we’ve got to do this. But a lot of the rhetoric around the YIMBY agenda as this being the solution or silver bullet to the housing crisis, I think is wrong.
And if we actually view this as something we need to solve urgently… You don’t want to solve it too quickly, right? Because then you can have—
GROSS: House prices falling by 20%.
WALKER: Exactly — you don’t want too much negative equity.
GROSS: Well, not falling by 20% again.
WALKER: Exactly, so you don’t want to solve it too quickly because you can cause those macroeconomic problems and financial instability.
But say you wanted to solve it in, I don’t know, three political terms or something — I don’t think you can do that under a YIMBY program.
I’m trying to write something on this.
But I think: all of these questions, like all of the things holding us back economically, things holding back our productivity growth — housing would probably be the first cab off the rank, right? — have much more urgency in a world in which Australia doesn’t have a great-power protector.
And that sounds like such a crazy kind of longbow, but I think all these debates — the stakes are going to become much higher. And you want to act now, when you still have the option, because [strategically] we’ll be isolated. We’ll be in a region where the [dominant] power is not an ally of ours. And you want to have a really strong, growing economy.
We should be aiming to solve things quickly.
And we’re not going to solve the housing crisis quickly.
GROSS: No.
If I could channel Peter Tulip for a second—
WALKER: Yeah, please.
GROSS: Bear with me while I get out my divining forks.
I think he would say: look, if a policy lever is relatively weak, that just means you have to pull it all the harder.
And so to the extent that full YIMBYism only gets you maybe half the problem solved, well then you better not dilly-dally around half YIMBYism. You better go full YIMBYism, and then investigate what other policies and fixes might be required on top of that.
And the second point I’d add, aside from my channeling of Peter: the Grattan Institute’s actually got a new report out that looks at the costs of housing, and especially in Melbourne, where the prices are substantially lower, they actually find that a lot of projects don’t pencil out. The economics, even with planning approval, often doesn’t stack up in a lot of different parts of Melbourne.
And so perhaps the companion piece of that full YIMBYism is a real examination of what are the costs going into our construction sector, and how we can bring those down.
But yeah — I guess that’s a podcast for another day.
WALKER: Yeah.
And one of the first questions I asked Greg and Michael in the podcast I did with them was: how big of a lever for supply is productivity?
Because if removing planning regulation is only going to get us so far, what other levers do we have?
GROSS: Yeah.