Sam is involved in innovation policy at Progress Ireland, an independent think tank based in Dublin, and he oversees a publication called The Fitzwilliam. Of particular interest, on his personal blog, he curates a popular link roundup; below is a condensed version of his Links for November.
Blogs and Short Links
1. Is Google search deteriorating? The quality of evidence surrounding this claim is surprisingly inadequate. A closer look reveals some rather poor data.
2. Exploring the game theory behind $1 margarita nights. The concept of a weekly social gathering acting as “office hours” for friends is appealing, especially as social coordination grows increasingly chaotic.
3. Mourning the loss of Peter Temin. His insights into the economy of the early Roman Empire were particularly enlightening. It might be time to delve into the economic frameworks of pre-industrial Europe in my reading group.
4. Recently, for Progress Ireland, I discussed the hesitance of academics to launch startups. Additionally, I provided an update on our initiatives to promote Olympiad-level mathematics in Ireland and announced the introduction of paid subscriptions for my blog.
5. Climate forecasts have proven to be fairly accurate.
6. Will AI intensify NIMBYism? Initial expectations suggested the opposite; the ultimate outcomes remain uncertain.
7. VAT reductions aimed at boosting readership. Could there be an arbitrage opportunity lurking in the paper market?1
8. Urban economics typically suggests that imposing minimum standards for apartments is misguided. However, the Irish government’s plan to reduce these standards is now under threat of legal challenges in the High Court, which may involve the European Union. It’s puzzling how elected officials can create regulations but find it difficult to amend them. Is this a form of selective anarchism regarding political authority, only when it favors less oversight? Or is it merely a manifestation of cynicism?
9. A curious thought: Have I been misinterpreting odd flag cones all this time?
10. Stephen Webb discusses Britain’s surplus of lifeguards. This brings to mind a humorous sketch by Mitchell and Webb about the outrage over zero drownings in a year in Britain:
I aim to highlight the alarming waste of public funds resulting in a year where absolutely no accidental drownings occurred. This raises questions about the implications for fencing, warning signs, swimming lessons, and the maintenance of waterside paths. Such an occurrence indicates significant overspending, as in any urban area of around half a million, a few annual drownings would be expected.
I am aware of various methodologies to calculate the value of a statistical life, but can anyone clarify (a) why different countries, even those with similar developmental levels, adopt varying methods, and (b) whether the significant discrepancies in these values have practical implications? (Once you’re done, I have similar queries regarding the different methods used for calculating social discount rates.)
Every government has its own valuation of human life that dictates investments in areas like road safety. The UK typically values life at around £2.5m (approximately $3m), whereas the US exceeds $12m, and the EU is around $4.5m.
Lastly, a peculiar observation from my inbox regarding Scottish lifeguards:
I’ve noticed that ‘lifeguard’ seems to be an exceptionally common summer job for teenagers in Scotland, unlike other regions of the UK. Why is that?
When discussing teenage jobs with Scottish individuals, I find that over 65% report having worked as lifeguards, compared to a mere <5% from England, Wales, or Northern Ireland. This anomaly is intriguing.
The BBC even aired a drama involving a Scottish lesbian lifeguard in Helensburgh, suggesting that lifeguarding is the go-to job for those seeking summer employment while living with parents.
I still can’t quite grasp it.
11. The New Yorker has taken a stance against non-commutative algebraic structures.
Music and Podcasts
1. Stevie Wonder’s Innervisions is a timeless classic. My top picks include Don’t You Worry ‘Bout a Thing and Too High. Remarkably, Stevie plays every instrument on this album—an astonishing feat. Don’t miss the accompanying You’ll Hear It episode.
2. Podcast of the month: Dean Ball discusses the creation of the AI Action Plan. Will the question of how to incorporate rational technocrats into administrations run by non-rational actors define the political landscape of the late 2020s?
3. Lauryn Hill’s The Miseducation of Lauryn Hill has been hailed as the greatest album of all time by Apple Music. This prompted me to revisit it. My favorite track is To Zion, but the entire album deserves to be experienced in sequence.
4. Hugh Mellor on Ramsey’s take on truth intrigues me for two reasons. First, there’s debate around the biography by Cheryl Misak and its alignment with pragmatism, which I’ve yet to fully explore. Additionally, hosting a conference on Frank Ramsey would be a dream come true.
5. The Marginal Revolutionaries discuss the Baumol effect. Interested readers can refer to the original work where Baumol introduced this concept.
6. Ahmad Jamal Trio’s The Awakening showcases energy that is distinctively different in tracks like Dolphin Dance compared to the original by Herbie Hancock. The latter has been a staple in my playlists.
7. Dan Wang and Stephen Kotkin explore the methodologies of historians. This discussion is best enjoyed in a video format. Kotkin’s experiences in Magnitogorsk are detailed in the Conversations with Tyler podcast, which ranks among my all-time favorites.
Papers
1. Matt Clancy presents How to Accelerate Technological Progress. I admire that Matt allows for nicely formatted PDFs of his work. This overview delves into materials available on New Things Under the Sun; noteworthy findings include how long it typically takes for discoveries to transition to technology (about 20 years) and the prevalence of independent discoveries. I’ve included this paper in the core readings for my economics course at the University of Edinburgh (a longer story for another time).
2. Various authors contribute to The Impact of Advanced AI Systems on Democracy. Surprisingly, in two reading groups, I’m the sole non-political scientist. Among the many contributors, I recognize Bruce Schneier and Saffron Huang.
The second page evaluates the literature on political biases in AI, indicating that LLMs tend to lean towards “progressive/libertarian” views when responding to multiple-choice questions, yet appear politically neutral in open-ended responses. It seems challenging to accurately assess political bias under realistic conditions.
My impression of Nature Perspectives is rather low; their approach to ‘fake rigor’ is exemplified by this paper’s 10 pages of vague opinions, all while citing a staggering 141 sources (without a sense that all authors have thoroughly engaged with them). Can we genuinely assert this contributes more to human knowledge than a thoughtful Substack post by one of the authors would?
Diving into the references, footnote 9 serves as a reminder for me to eventually blog about James Fishkin’s deliberative polling methodology, particularly regarding the potential for consensus among Northern Irish people. Recently, Nan Ransohoff had the misfortune of hearing my extended explanation of this literature when she asked what could be learned from Ireland’s successful policies.
In footnote 60, Hugo Mercier’s “argumentative theory of reason” is cited to support the claim that organizing large-scale in-person events is both costly and time-consuming, and that face-to-face debates can be influenced by social desirability biases where participants prioritize winning an argument over achieving consensus.
Is it necessary to cite the observation that individuals may prioritize winning arguments over mutual agreement? If it is, why reference such a specific and contentious hypothesis in cognitive science that seems irrelevant to the main claim?
This reminds me of a lengthy draft I have that critiques Mercier and Sperber’s argumentative theory, which I initially intended to submit for a contest that I missed. Whether it will ever be published remains uncertain; if anyone is interested in reading it and providing feedback, you can find it here.
This marks the second instance in political science discussions where I’ve found myself sympathetic to a paper’s premise but frustrated by its execution. One wonders if my pedantic tendencies might lead to my eventual dismissal from the group.
Films and Video
1. Joshua Oppenheimer’s The Act of Killing is a groundbreaking documentary that examines the genocide of suspected communists and ethnic Chinese after the 1965 coup that brought Suharto to power in Indonesia. With minimal archival material available and an uncertain death toll, Oppenheimer lived in Indonesia for years, learning the language to engage with the perpetrators who eagerly reenacted their crimes.
I consider The Act of Killing to be a monumental achievement, justifiably regarded as one of the greatest documentaries ever made. In comparison to the works of Ken Burns, it stands in a league of its own.
2. Lastly, from YouTube, I’ve enjoyed performances by Bob James and Tame Impala on NPR’s Tiny Desk. Additionally, I’ve been watching footage from the Chinese Cultural Revolution. Lastly, a thought-provoking talk on cosmological natural selection as conceptual infrastructure for AI alignment was featured at the ODYSSEY conference, part of the ILIAD series on alignment. The all-caps titles are perplexing and don’t seem to form an acronym, though there have been amusing attempts at backronyms like “International League of Intelligent Agent Deconfusion.” I’ll treat the first person to devise a fitting AI safety backronym for EPIC OF GILGAMESH to dinner.
You can find the complete version of Sam’s November links here.
[1] A timely reminder that value-added tax tends to generate less unnecessary pain compared to sales tax.
