There is a particular sleight-of-hand at the heart of modern LLM reasoning that, the more I look at it, the more it bothers me. The argument goes like this: Transformers are shallow. A 70-layer stack is fixed depth ' it sits in complexity classes like AC' or TC', which is a polite way of saying it cannot, in a single forward pass, solve problems that fundamentally require sequential computation. So we paper over this by making the model think out loud. We give it a scratchpad. We call it Chain-of-Thought. We celebrate. But CoT is not reasoning. CoT is the model renting depth from its own output tokens. Every reasoning step has to leave the residual stream, become a discrete token in a vocabulary built for human communication, and come back in through the embedding layer for the next step. It is, mechanically, an absurd way to do internal computation ' like a CPU that must spill every intermediate register to disk in plaintext English. Sapient Intelligence's bet, made first with the original Hierarchical Reasoning Model paper last summer and now extended into the language domain with HRM-Text, is that this is fixable. Not by making the model bigger, not by training on more CoT traces, but by giving the architecture the one thing it doesn't have: variable, internal, depth. Reasoning that happens in the latent space, not in the token stream....
For a long time, Representative Thomas Massie confidently defied an ironclad law of modern Republican politics'that to oppose President Trump was to start a ticking clock on your electoral career. 'I'm not worried about losing,' he told me last spring inside the Capitol, as he explained to a group of reporters the strength of his support within his Kentucky district. Massie had already angered Trump just a few months into the president's second term, after clashing with him during his first. Massie voted against government-funding bills, criticized the president's tariffs, and would soon become one of the only Republicans in Congress to oppose Trump's One Big Beautiful Bill Act, which the fiscally hawkish Massie deemed irresponsible. Trump lashed out at Massie and vowed to find a primary opponent to defeat his bid for an eighth term; as early as last summer, the president's allies stood up a political-action committee to run ads attacking Massie in his district. Still, Massie refused to fall in line. Over the next several months, he condemned Trump's military adventurism, including his unilateral attacks on Iran, and he helped lead a remarkably successful bipartisan effort to force the administration to release its trove of files on the disgraced financier and convicted sex offender Jeffrey Epstein. Massie, an iconoclast to his fans and an ineffective gadfly to his detractors, had always gone his own way in Congress. Maybe he believed he was uniquely positioned to withstand a Trump-backed barrage. Or perhaps he knew he was toast and had resolved to go down on his own terms....
My senior seminar on the reactionary right has now come to an end. The honest truth is that I'm not very happy with how the class unfolded. It was a mixed bag. Some of the readings were illuminating, as were some of our discussions in class. But at least half of the readings disappointed me, and I would say the same about how the students responded to those readings. I can't say I blame them. This was a collection of grim, angry, embittered authors and texts often motivated by imperfectly processed resentments. In what follows, I will begin by laying out the conceptual assumptions behind my choice of readings and themes for the course. Then I'll list the authors and texts I assigned. From there, I'll discuss those that really failed to impress me'that is, those I would not teach again. Then I'll close by highlighting the best things we read and thought about in the class'and how that has changed the way I assess the reactionary impulse in politics....
One of the things I want to explain in my forthcoming book on Leo Strauss and his influence on the American right is how some of his students (and students of students) ended up embracing the figure of Donald Trump and advocating for him or another likeminded Republican president to rule as a 'Red Caesar.' I hope to show both that this way of thinking about present-day politics partially follows from certain assumptions they inherited from Strauss (along with certain moral convictions they erroneously thought his writings validated) and that Strauss would not endorse such judgments himself. The term 'Red Caesar,' which journalists covering the American right began to cite in the aftermath of the insurrectionary uprising of January 6, 2021, comes from a 2020 book by Michael Anton with a characteristically measured and not-at-all-hysterical title of The Stakes: America at the Point of No Return. Anton is, of course, the author of the notorious 'Flight 93 Election' essay from September 2016 that advocated, in pretty incendiary terms, that all conservatives should enthusiastically vote for Trump. Nothing less than the continued survival of the country was at stake, Anton claimed. By 2020, after serving as a national security official during the first Trump administration, Anton had begun speculating about situations in which a breakdown in the capacity for self-governance in the United States had proceeded so far that a dictatorial seizure of power would become necessary....
