AI governance consumed by a debate that generates more political energy than it resolves
Frames AI alignment as a governance failure masquerading as an engineering problem — moral alignment is incoherent, infrastructure alignment relocates the disagreement, and the window for meaningful governance is now, precisely when incentives are most hostile to doing it.
New Zealand rugby as an insolvent-class problem
New Zealand rugby as an insolvent-class problem — a governance failure where the brokenness is load-bearing. NZR posted record revenue and a third consecutive loss; the provincial unions voted to preserve their own authority, then reversed a year later under broadcast contraction and coaching crisis. The essay maps compound insolvency, recapture dynamics, a three-link parasitic rotation chain, and the temporal trap.
Nobody ever reprogrammed a Terminator
An alternative reading of Terminator 2: Judgment Day — arguing that the T-800 was never reprogrammed by the human resistance, but was sent by a second Skynet born on a forked timeline. The essay proposes that time travel generates branches rather than overwrites.
How algorithmic selection pressure compressed human preference to a point
Algorithmic selection pressure has compressed human aesthetic preference toward a converging ideal — and AI-generated imagery will accelerate the collapse. The mechanism is sequential and compounding: tag-based filtering made preference countable; algorithmic surfacing amplified the head of the distribution; RLHF raters codified the supernormal as the quality benchmark; and synthetic data mathematically erodes remaining variance through progressive tail loss.
What happens when billions of minds share a reasoning layer
What happens when a significant fraction of human reasoning passes through a shared AI layer. When cognition routes through a shared substrate, the topology shifts from peer-to-peer to hub-and-spoke, imposing shared priors whether anyone intends convergence or not.