Sam works on innovation policy at Progress Ireland, an independent policy think tank in Dublin, and runs a publication called The Fitzwilliam. Most relevant to us, on his personal blog, he writes a popular link roundup; what follows is an abridged version of his Links for November.
Blogs and short links
1. Is Google search getting worse? The quality of evidence around this is remarkably poor.
2. On the game theory of $1 margarita night. As this post describes, it would be great to have a standing social occasion every week at the same time and place that could serve as my friend group’s “office hours”. The changeability of plans makes me feel that more and more of my time is being spent on social coordination.
3. Rest in peace to Peter Temin. I enjoyed his paper reviewing the economy of the early Roman Empire. At some point, we should be covering the economies of pre-industrial Europe in my reading group.
4. For Progress Ireland, I recently wrote about why more academics don’t start companies. I also gave an update on our efforts to support Olympiad-level mathematics in Ireland, and turned on paid subscriptions for my blog.
5. Climate predictions have been relatively accurate.
6. Is AI going to supercharge NIMBYism? The initial hope with efforts like Tract was the opposite, and I have no prediction on how these things will shake out.
7. VAT cuts to create more readers. Do I smell an arbitrage opportunity in the paper market?1
8. The conventional wisdom in urban economics is that minimum apartment standards for floor space and other matters of personal preference are a terrible idea. The Irish government’s stated reduction in minimum apartment standards is now being delayed, or possibly cancelled entirely, because of a challenge in the High Court. The court in question is intimating that the European Union may need to get involved. Even if you think that modestly reducing minimum apartment standards is a bad idea, I’m baffled by the worldview in which democratically elected officials had the authority to introduce these regulations in the first place, but not to modestly reduce them. Is this anarchism with respect to the problem of political authority, but only when changes are in the direction of less regulation, rather than more? Is there even a worldview here, or just bitterness and cynicism?
9. Another thought: Have I been looking at weird flag cones this entire time?
10. Stephen Webb on why Britain has too many lifeguards. I’m reminded of the Mitchell and Webb sketch about how much of an outrage it is when a year goes by and zero people drown in Britain:
I’m trying to draw attention to the massive waste of public money that’s led to a situation in where absolutely nobody in a whole year drowns by accident. What that must mean in terms of fencing, warning signs, swimming lessons, people coming into school to tell children to be careful, life belts and the maintenance of waterside paths is just staggering. There has clearly been a massive overspend, because in any conurbation of up to half a million people such as Westchester that’s run with the proper priorities, at least two or three people should drown every year.
I am familiar with the different methodologies for how to calculate the value of a statistical life, but can anyone explain to me (a) why different countries, even at similar levels of development, chose different methods, and (b) whether the large divergence in these numbers has any practical significance? (After you’re done with that, I have the same confusion about the different methodologies for calculating social discount rates.)
[E]very government has a value for a human life which determines the appropriate level of investment in, say, road safety measures. The UK is typically at the lower end here—at around £2.5m (say $3m) compared to over $12m in the US and about $4.5m in the EU.
Finally, from my email inbox, I have been informed of the mystery of the Scottish lifeguards:
One of my minor obsessions is that it seems that ‘lifeguard’ is an extremely common teenage summer job in Scotland but not anywhere else in the UK. I have no idea why.
Every time I speak to a Scottish person now I have to stop myself from asking “did you work as a lifeguard as a teenager?”. But every time teenage jobs come up in conversation, >>65% of Scottish people I’ve met say they were lifeguards, compared to <<5% of English or Welsh or (Northern) Irish people. I’m not randomly sampling, obviously, but that’s crazy.
The BBC even had a lesbian romance drama set in Helensburgh where the whole premise is that ‘lifeguard’ is the default job you take when you’re just looking for something to make money while living with your parents.
I do not get it.
11. The New Yorker has come out against non-commutative algebraic structures.
Music and podcasts
1. Stevie Wonder, Innervisions. I had forgotten how superb this album is. My favourite tracks are Don’t You Worry ‘Bout a Thing and Too High. Except for some backup vocals and minor synthesiser, Stevie Wonder plays every instrument on this album—absolutely insane. Here is the associated You’ll Hear It episode.
2. Podcast of the month: Dean Ball on how he wrote the AI Action Plan. Is “How do you insert sane and intelligent technocrats to be largely left alone in administrations run by lunatics?” going to be the defining political question of the late 2020s?
3. Lauryn Hill, The Miseducation of Lauryn Hill. Realising I should blog more about genres other than jazz and classical. I was prompted to re-listen to this after Apple Music declared it the greatest album of all time. Perhaps my second favourite rapper after André 3000, although it has been a while since I have thought about such things. My favourite track is To Zion, although you should certainly be listening to the whole album in order.
4. Hugh Mellor on Ramsey on truth. I never did get around to reading Truth and Probability, but this is of interest for two reasons. First, I’ve heard philosophers claim that the Cheryl Misak biography completely shoehorns Ramsey into the pragmatist tradition, but I have not waded into this debate. Second is that, because of the breadth and depth of his thought, running the teeny tiny conference on Frank Ramsey would be a dream.
5. The Marginal Revolutionaries on the Baumol effect. Here is the original book where Baumol proposed the idea with William Bowen.
6. Ahmad Jamal Trio, The Awakening. Pay attention to how the energy is different on Dolphin Dance compared to the Herbie Hancock original (I’ve also been re-listening to that album, one of my favourites). You probably recognise the lick around 0:40 from often being sampled in hip hop.
7. Dan Wang and Stephen Kotkin talk about how historians work. This one works better as a video podcast. You can hear more about Kotkin’s time living in Magnitogorsk on the Conversations with Tyler podcast, one of my favourite podcast episodes of all time.
Papers
1. Matt Clancy, How to Accelerate Technological Progress. One of the many (many!) things I appreciate about Matt is that he allows his webpages to be downloaded as nicely-formatted PDFs. This is an overview of some of the materials on New Things Under the Sun; the most interesting new ones I found out about are reviews on how long it takes to go from science to technology (answer: about 20 years), and how common independent discovery is. I assigned this paper as one of the core readings for an economics class I teach on at the University of Edinburgh (long story, I’ll post about it soon).
2. Various, The Impact of Advanced AI Systems on Democracy. I have now joined not one but two reading groups in which I am the only person who is not a professional political scientist. From the long author list, I only recognise Bruce Schneier and fellow Interact-er Saffron Huang.
Page two reviews some of the literature on political biases in AI: they say LLMs are biased toward “progressive/libertarian” views (aren’t these often opposites? What does this mean?) when presented with multiple choice questions, but are politically neutral when they are able to respond freely. My sense is that it’s been difficult to measure political bias under actual realistic usage conditions.
I generally came away with a pretty low opinion of Nature Perspectives, or editorials associated with Nature in general. One aspect of ‘fake rigour’ this paper excels in is that these 10 pages of relatively vague and milquetoast opinions cite a whopping 141 sources (and I do not get the impression the authors have actually read them all). Can we honestly say this was a greater contribution to human knowledge than a thoughtful Substack post by one of them would have been?
Diving into those sources so you don’t have to, footnote 9 is a nice reminder that I should blog at some point about James Fishkin’s deliberate polling methodology in the context of whether we can ever get Northern Irish people to agree on anything. Nan Ransohoff recently had the misfortune of listening to me talk about this literature in more detail than she could have possibly wanted to explain why my answer to her question of “What can other countries learn from what Ireland has done well in policy?” was not “citizens’ assemblies”.
In footnote 60, Hugo Mercier’s heterodox “argumentative theory of reason” is cited as “evidence” of the following claim:
Does the claim that people sometimes want to win arguments, rather than come to a mutually agreeable outcome, really need a citation? Even if somehow the answer is yes, why would that citation be one extremely specific and highly disputed hypothesis in cognitive science, that is largely irrelevant to the claim as stated in the main text? (Sorry for being mean.)
This also reminded me that I have a 10,000-word absurdly bloated draft of a blog responding to Mercier and Sperber’s argumentative theory of reason, which I never finished on time to submit to the Astral Codex Ten book review contest. It may or may not ever see the light of day; if you want to read a copy and let me know whether it’s salvageable, you can do so here.
This was the second time in the political science group that I was instinctively sympathetic with the idea of a paper, but frustrated with the sloppiness of the execution. One suspects that if I keep coming across as so pedantic, they are going to kick me out.
Films and video
1. Joshua Oppenheimer, The Act of Killing. This is a documentary about the genocide of alleged communist sympathisers and ethnic Chinese following the 1965 coup by which Suharto came to power in Indonesia. Given that the governments since the transition to democracy have still been sympathetic to Suharto, very minimal archives have been released, and it’s unknown how many were killed, even to the nearest million. To make this film, Joshua Oppenheimer spent several years living in Indonesia and learning to speak fluent Indonesian, for the purposes of filming and partially befriending the perpetrators, who happily volunteered to reenact the executions.
I think The Act of Killing is a great achievement, and deserves its reputation as possibly the greatest documentary of all time. As someone who has watched a fair amount of Ken Burns, this is in a completely different league.
2. Finally, from YouTube, we have Bob James and Tame Impala on NPR Tiny Desk. I have also been watching footage from the Chinese Cultural Revolution. Finally, a speculative talk: cosmological natural selection as conceptual infrastructure for AI alignment. That is from the ODYSSEY conference, the second in the ILIAD series of alignment events. Confusingly, the all-caps titles don’t seem to indicate an acronym, although some backronyms have been suggested (“International League of Intelligent Agent Deconfusion”). I will buy dinner for the first person who can come up with a good AI safety backronym for EPIC OF GILGAMESH.
You can read the full version of Sam’s November links here.
[1] Also a good time to be reminded that value-added tax creates less unnecessary pain than sales tax.
(0 COMMENTS)
Source link
