In 1904, the horse du jour was Clever Hans, widely reputed to be so much smarter than his brethren that he could do math, tell time, and even read and spell. Word spread fast by word of mouth, and eventually the occasionally gullible The New York Times reported that Hans was so smart that he "can do almost everything but talk". Ask Hans what 12 plus 13 is, and he would stamp his feet 25 times. People were amazed, and paid good money to see him. Turns out the horse knew no math; it had solved the arithmetic problems--all of them --in a different way.

On June 22nd, visitors to Slate Star Codex, a long-standing blog of considerable influence, discovered that the site's cerulean banner and graying WordPress design scheme had been superseded by a barren white layout. In the place of its usual catalogue of several million words of fiction, book reviews, essays, and miscellanea, as well as at least as voluminous an archive of reader commentary, was a single post of atypical brevity. "So," it began, "I kind of deleted the blog. The farewell post was attributed, like virtually all of the blog's entries since its inception, in 2013, to Scott Alexander, the pseudonym of a Bay Area psychiatrist--the title "Slate Star Codex" is an imperfect anagram of the alias--and it put forth a rationale for this online self-immolation. "Last week I talked to a New York Times technology reporter who was planning to write a story on Slate Star Codex," the post continued. "He told me it would be a mostly positive piece about how we were an interesting gathering place for people in tech, and how we were ahead of the curve on some aspects of the coronavirus situation." In early March, Alexander had suggested that his readers begin to prepare for potential catastrophe, and his extensive review of the available medical literature led him to the conclusion that, despite the early guidance by the Centers for Disease Control and Prevention to the contrary, masks were likely to prove more useful than not. A month later, he looked back at his forecast and awarded himself a "solid B-"--not perfect, but at least more accurate than the news media, which, with some notable exceptions, he wrote, "not only failed to adequately warn its readers about the epidemic, but actively mocked and condescended to anyone who did sound a warning." Journalists, in his view, were guilty of an inability or a refusal to weight the possible outcomes. As he put it, if there was even a ten per cent risk of a ruinous pandemic, shouldn't that have been the headline? Alexander, who prefaces some of his own posts with an "epistemic status," by which he rates his own confidence in the opinions to follow, thought the media, too, should present its findings in shades of gray. The final post went on, "It probably would have been a very nice article.

Gwern has answered my prayers and taught GPT-2 poetry. GPT-2 is the language processing system that OpenAI announced a few weeks ago. They are keeping the full version secret, but have released a smaller prototype version. Gwern retrained it on the Gutenberg Poetry Corpus, a 117 MB collection of pre-1923 English poetry, to create a specialized poetry AI. I previously tested the out-of-the-box version of GPT-2 and couldn't make it understand rhyme and meter.