Monday Night Links

• Let’s suppose that the Gospels really are a historical account of real people and events, and not some later invention. Well the Gospels are full of named individuals, so we should expect the statistical distribution of their names to match that of the society from which they were taken. Sure enough, the fit is very good: if you go and tabulate the names found in first century Judaean ossuaries, the most popular ones by far are Simon and Joseph, and there are 8 distinct Simons and 6 different Josephs mentioned in the New Testament. The most popular female name at the time was Miriam (Mary). None of this is something that somebody making up a story centuries later would have known, and once again we find that the various apocryphal and gnostic gospels are full of weird names and names that were popular in other times and places. Because you see, the thing about names is they wax and wane in popularity very fast! The most popular names in first century Judaea were not the most popular names in third century Judaea. They weren’t even the most popular names in first century Alexandria! Most of the wealthy and assimilated Jews of the later Roman Empire, the ones that somebody trying to fake the Gospels would have known, were Alexandrine. But the name distribution in the New Testament fits the name rankings of cosmopolitan Alexandria very poorly (common Jewish names in Alexandria included Sabbatius and Dositheus), and that of backward and isolated Palestine very well. You can push the onomastics angle even further. Imagine that you have ten friends named Simon and one friend named Thaddeus. When you’re writing to somebody about Thaddeus, you might just call him Thaddeus, and when you’re writing about one of the Simons, you’d include extra information to disambiguate him. Every bit of the Gospels, down to the random side conversations, fits this principle perfectly. [Mr. and Mrs. Psmith’s Bookshelf] 
• The first pattern that you notice is that popular baby names change over time just like all fashion. What seems to happen is that certain sounds or phonemes become popular and that drives name choices with those sounds. For example, Ava, Emma, and Anna all seem to trend together in part because they share similar sounds. But if any name becomes too popular, it stops being used as much. This leads to shifts in popularity over time. [Explorations in Personality]  
• After that first visit I read the Essay on Development and found, as he did, that “to be deep in history is to cease to be Protestant.” In a gap year I read the Grammar. Back at Oxford, I made the Littlemore pilgrimage each feast day. I drifted from my D.Phil. topic in history into theology. On an Easter retreat at Littlemore, I wrote an essay about my conversion in the library. Then doors opened. [The Lamp]
• The most significant development is that the “scaling law” appears to be breaking down – more compute is no longer delivering proportionately meaningful gains in model performance. Indeed, it is even possible future models start to get worse on account of AI “pollution” of the training data set (discussed more below). Moreover, evidence is also emerging that LLMs have fundamental limitations in their capacity to reason, and in contrast to early speculative optimism, it appears they do not in fact have internal models of the world and are instead simply sophisticated imitation engines. Unreliable output, or “hallucinations”, are proving persistent, and may in fact be an incurable feature of LLM architectures, rather than merely temporary nuisances. To the extent this proves to be the case, LLMs may be a dead end and genuine breakthroughs in AI/AGI may require us to go “back to the drawing board” with RL and/or entirely new and more targeted architectures, potentially a tougher grind and setting us back decades relative to prior expectations. [Lyall Taylor]
• Management was recently authorized to increase Aztec’s buyback program to accommodate additional privately negotiated block purchases. This brings the total dollar amount of authorized buybacks to $8,750,000. Aztec will have repurchased a total of approximately $7,800,000 in shares since December 2024. Subsequent buybacks will be considered on a case-by-case basis. [Aztec Land & Cattle Co., Ltd.]
• BlackRock-owned Global Infrastructure Partners is in advanced talks to buy utility group AES people familiar with the matter said on Wednesday, a deal that could be one of the largest ever involving a U.S.-listed power company. [Reuters]
• [H]e was unhappy that, in two places in the piece, an editor had changed the word “but” to “however.” He made his case for a page and a half, and concluded, “But is a hell of a good word and we shouldn’t high hat it. . . . In three letters it says a little of however, and also be that as it may, and also here’s something you weren’t expecting and a number of other phrases along that line.” [The New Yorker]
• The other night, struggling to sleep, I was visited by a fantasy: The 2028 presidential election is coming to its conclusion and the candidates are Republican Sen. John Thune of South Dakota and Democratic Gov. Josh Shapiro of Pennsylvania. President Trump has quietly—I told you this is a fantasy—announced his intention to retire to the golf course. JD Vance, driven out of the campaign by disastrous polling, will caddy for him. [WSJ]
• Kennedy got into the habit, according to a person familiar with the matter, when he was living in Bedford, N.Y., an area sometimes described as the epicenter of Lyme disease. He liked to start the day by taking his dogs or hawks (Kennedy is an avid falconer) for a hunt or a hike, then hit the gym afterward. Wary of ticks, he wore jeans for his outdoor adventures and then just kept them on for his workout. [WSJ]