Sam Enright, a proponent of innovation policy at Progress Ireland, an independent policy think tank in Dublin, also oversees a publication called The Fitzwilliam. For those interested in his insights, he maintains a personal blog on which he often shares a well-received link roundup; below is a distilled version of his Links for June.
Blogs and Short Links
1. Henry Oliver argues that the notion of an English literature degree being ideal preparation for a prime ministerial role is utterly absurd. Read more here.
2. Patrick Collison offers a glimpse into Hibernofuturism, a concept that’s both intriguing and perplexing.
3. José Luis Ricón discusses Claude’s constitution. He also shares additional links, which delve into the bizarre debate over whether worms can gain memories through consuming other worms. Yes, you read that right!
4. Goodreads has its share of documented flaws, yet perhaps its most philosophical failing lies in its blanket recommendations, which naively presume the existence of an “average reader.” Personally, the fact that Why We Sleep by Matthew Walker ranks in the top 5% of ratings does not pique my interest. In response to this issue, Mark Greville has developed BookTribes, which curates book suggestions based on a user’s unique taste rather than mere popularity.
5. A round of applause is warranted for the establishment of the Arq Foundation, a new Brussels-based think tank focused on preparing Europe for the transformative impacts of AI.
6. The latest edition of “unexpected Timothy Williamson” is worth checking out. Perhaps someone should gift Ava a copy of Timothy Williamson’s crochet puppet for her collection.
7. Frequently, I’m asked if there’s a definitive book on the Irish housing crisis that doesn’t suffer from economic ignorance. The unfortunate answer seems to be no. However, Ronan Lyons is set to change that with his upcoming release Brittle: How the Housing Crisis is Breaking Ireland—and How to Fix It, now available for pre-order and set to debut this October.
8. Alex Tabarrok has effectively dismantled the Coase conjecture.
9. I often ponder how much Jorge Luis Borges would have relished experimenting with large language models. Many themes in his works foreshadow contemporary discussions, notably in his fiction. If you’re a Borges aficionado, you’ll appreciate Pierre Menard, Author of the Quixote. I found Linch Zhang’s attempt to fine-tune an LLM to reproduce Borges’ original story, token-for-token, to be quite amusing. Check it out here.
10. Ben Yeoh pens a weekly letter that I find particularly valuable, especially regarding the ongoing debate on whether cancer outcomes are genuinely improving after accounting for earlier testing and the aging population. It appears that cancer mortality has decreased by approximately one-third since the early 1990s.
11. A new magazine titled In Development has emerged, featuring an essay by Paul Niehaus discussing the origins of GiveDirectly. He humorously notes in footnote 13, which resonates with my own skepticism regarding microfinance:
In 2005, my co-founders and I managed to secure invitations to the launch event for the International Year of Microcredit at the United Nations. I recall the cocktails being significantly stronger than the evidence.
In Development’s editor-in-chief, Lauren Gilbert, also curates her own linkpost, which leads me to Chris Blattman’s recommendations for essential readings for development economists.
12. In my essay on Milton Friedman, I noted my surprise at the lack of scrutiny regarding a well-known doctored image of him. A reader pointed me to a Forbes article that confirmed the image’s falsity; Friedman found it so amusing that he shared it with all his friends.
13. One of my all-time favorite linguistics texts is The Language Instinct by Steven Pinker. Here are some reflections on it from Dwarkesh Patel.
Music and Podcasts
1. Dario Amodei makes a return on the Dwarkesh Podcast. One might jest that he has, in a way, wagered humanity’s future on a model based on 19th-century bottled-water companies:
[L]et’s just simplify the entire industry. Imagine we’re in an economics textbook featuring a limited number of firms, each with constrained investment potential. They can allocate a fraction of their resources to R&D, and the marginal costs associated with serving customers are significant. The gross profit margins on those marginal costs are quite high due to efficient inference. While competition exists, the models are also differentiated.
In this scenario, companies will compete to enhance their R&D budgets. Yet, because there are few players, we encounter the… what is it called? The Cournot equilibrium, I believe. The point is, it doesn’t settle into perfect competition with zero margins. If there are three firms in the economy, and all behave rationally, it won’t equilibrate to zero.
2. Tom Holland interviews Paul McCartney. Listening to McCartney always reminds me of Simon Pegg’s spot-on explanation of the embouchure of Beatles impressions.
3. ChinaTalk features discussions with representatives from the Institute for Progress and the Foundation for American Innovation about the evolving role of think tanks in America.
4. My latest discovery in Chinese history podcasts is Barbarians at the Gate, which covers the An Lushan Rebellion, a devastating civil war that marked the decline of the Tang dynasty’s glory years.
Many posts online discuss how the deadliest events in global history are often dominated by obscure Chinese civil wars. I have long struggled to reconcile my confusion regarding historical population estimates in China. The figures for city sizes and war casualties often seem remarkably outside what one might expect based on general technological development and other contextual factors.
Reflecting on my teenage reading of The Better Angels of Our Nature, I was astounded by the death estimates attributed to various Chinese civil wars, many of which were unfamiliar to me. In Chapter 5, Steven Pinker mentions that the An Lushan Rebellion “according to censuses, resulted in the loss of two-thirds of the empire’s population, a sixth of the world’s population at that time” (p.195).
Can that possibly be accurate?!
These estimates stem from comparing the Tang’s census figures from 754/755 AD (around 53 million) and the 764 census (approximately 17 million).
The issue is that few credible sources support the idea of 36 million as a plausible death toll; it’s more of an upper bound. The podcast hosts suggest that this number primarily reflects societal dislocation and the collapse of administrative structures during that period. Historical data from the Chinese census dates back to 2 AD, and equating census differences with deaths tends to misrepresent more functional areas, as they at least maintained sufficient state capacity to conduct a census. This has not generally been true throughout history.
To my knowledge, no one has rigorously reassessed Pinker’s violence history analysis while considering the credibility of census data as upper bounds. Regrettably, most counterarguments to the Better Angels thesis have been rambling and convoluted.
Interestingly, one of the world’s most famous poems centers on An Lushan. Anyone who attended school in China can likely recite the opening couplet of Spring View, where Du Fu reflects on the destruction wrought by the rebellion:
The country is broken, though hills and rivers remain,
In the city in spring, grass and trees are thick.
All this serves as a lengthy preface to what was ultimately my second favorite podcast episode about the An Lushan Rebellion I encountered last month. My top pick was an episode by Melvyn Bragg on In Our Time. In that discussion, when Bragg mentioned Pinker’s analysis, the expert guests erupted in laughter. Guffaws!
5. A fascinating episode on the life and philanthropy of John Arnold. I found this particularly enjoyable: Arnold made his debut in the (highly profitable, not fraudulent!) natural gas trading division of Enron. Following the Enron collapse, he founded one of the most lucrative hedge funds in history. At just 38, he decided to return the fund’s capital to investors and devote his efforts to philanthropy.
6. Paul Gillingham shares insights on life in Mexico. Although I know little about Mexico and have yet to visit, I could easily spend my days wandering with my AirPods, absorbing podcasts that explore diverse places and their histories.
7. 80,000 Hours discusses their rebranding, which now focuses almost exclusively on guiding careers towards the transition to AGI.
8. Gustav Mahler’s Symphony No. 5, accompanied by the related Sticky Notes episodes, was performed live under the direction of Jonas Alber. This event also inspired my first classical music discussion group—a bold experiment to see how far the reading group format could stretch. You can view my discussion prompts here. The third movement (Adagietto) is exceptionally poignant, but its connection to the other movements is rather abstract. Generally speaking, it’s easy to get lost while listening to Mahler. Thus, my conclusion is that this piece may have been too intricate to start with. I’m now three down and seven to go in my quest to experience all Mahler symphonies live before I kick the bucket.
9. Benjamin Britten’s Violin Concerto has emerged as one of my recent favorites; check out this YouTube version by Janine Jansen. Written in 1939, it’s often viewed as an elegy for the Spanish Civil War, though to me, it conveys a sense of dread over an impending war. While Britten isn’t usually among my favorites, this piece contains moments that resonate, particularly the pizzicato in the first movement.
In keeping with my habit, I sought Tyler Cowen’s thoughts on Britten. For what would have been his 100th birthday, Cowen reflected on how Britten serves as a lens for understanding John Maynard Keynes’ economic theories and “the broader history of homosexuality in England.” Because, of course, why would we focus on just one topic?
Books and Papers
1. DeepSeek, DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning. The initial DeepSeek paper from January 2025 triggered a plummet in American tech stocks, as a Chinese language model showcased near-frontier performance while being far more computationally efficient. It’s surreal to think this event occurred merely a year and a half ago; it feels like a decade has passed.
DeepSeek-R1 was the second “reasoning” model after OpenAI’s o1. Prior to their development, language models were indeed improving, yet there wasn’t a mechanism to transform more tokens into higher-quality responses. With reasoning models, reinforcement learning is employed to prioritize tokens in the “chain of thought” that led to correct answers. This shift has led LLMs to begin saying things like “wait, let me reconsider…” and self-correcting.
You can find my discussion prompts for the meetup at the Fitzwilliam AI Circle, concerning both this and the DeepSeek paper here.
2. Amanda Askell, Claude’s Constitution. Admittedly, I’m late to the party, as this topic garnered significant attention back in January. It’s challenging to evaluate this document since the specifics of its implementation in the training process remain largely undisclosed. When Claude makes a moral decision, how much of that stems from pretraining, how much from the system prompt, how much from reinforcement learning via human feedback, and how much is the result of Anthropic’s unique approach to “model character”? Does anyone outside the labs truly know? Does anyone inside the labs even understand?
From my understanding, the constitution aims to establish a concise set of stringent constraints incorporated into the training process, enabling the model to exercise sound judgment on which user requests to decline or challenge while comprehending the rationale behind those decisions. This also reflects Anthropic’s vision (or, depending on your perspective, delusion) of humanity evolving alongside Claude. If we provide a sufficient number of high-quality examples of human reasoning, he might surpass our wisdom.
A truly outrageous essay in the New Yorker offered a perspective on the Constitution from someone who clearly made no effort to grasp the technical intricacies involved, despite the fact that those details are crucial. Consider:
One way to interpret Claude’s Constitution is to view it as a response to a state collapse. It suggests that the failure of the U.S. Constitution has led to the necessity for Claude to have a constitution, which supposedly protects American citizens (and foreign entities) from the overwhelming might of the United States military.
What on earth is this individual talking about? Did the renowned fact-checkers at the New Yorker not inform her that this document was only published in January? Or that other labs release models (sans Grok) that are not evidently less ethical, yet lack a constitution?
On a lighter note, I stumbled upon an amusing detail in the acknowledgments, where a bishop and a priest were thanked. One can only assume the Pope himself was unavailable for comment.
Fun fact: I happen to own one of the few physical copies of Claude’s Constitution. The essence of artificial intelligence is represented by a charming #DE643C in hexadecimal color code.
3. Alan Ahearne, Entrepreneurial Activity and Living Standards in Ireland. This paper, authored by the former economic advisor to Micheál Martin, was commissioned by John and Patrick Collison. I have no personal connection to Ahearne, but given its apparent relevance to my interests, I felt compelled to read it.
Essentially, Irish workers rank among the most productive globally. However, when disaggregated into multinational companies versus domestic enterprises, we observe a stark contrast: while the multinational sector is highly productive, the domestic economy appears lackluster, if not mediocre. This scenario renders the Irish economy particularly vulnerable; a mere handful of American firms departing could spell economic disaster. Thus, it would be prudent for us to foster more domestic entrepreneurship.
4. George Berkeley’s The Querist. I sometimes wonder if I’m the sole individual on the planet who hasn’t delved into Bishop Berkeley’s philosophy, yet has read his pamphlets on Irish coinage policy from the 1730s.
Berkeley’s renowned philosophical works on idealism were penned in his twenties. His journey culminated in becoming the Bishop of Cloyne in County Cork, in the southwest of Ireland.
The Querist is a true gem: a three-part pamphlet advocating for paper currency and a national bank in the context of Ireland’s development, all expressed through sarcastic numbered questions. Imagine if Wittgenstein’s Tractatus discussed banking and credit while also touching on agricultural productivity in 18th-century County Tipperary.
This book is so obscure that I faced challenges locating a copy, but I found the formatting of the Project Gutenberg edition to be perfectly acceptable. One day, I’ll complete my treatise on the Irish Enlightenment, but for now, you’ll have to take my word that The Querist is one of the most underrated contributions to economic thought.
Films and Videos
1. Jonathan Demme’s Stop Making Sense. If you haven’t heard of it, this iconic recording captures a Talking Heads concert held in a Hollywood theater in 1983. They perform hits from all their albums up until that point, including Burning Down the House and their rendition of Take Me to the River.
