Launch-season December, fully institutionalized (the counter-programming cadence now has a holiday calendar): OpenAI is mid-“12 Days of Shipmas” (o1 full release with a $200/month Pro tier, the inference-economics file notes the price point as the margin war’s opening bid: thought, tiered; Sora’s public release Monday, the earlier demo now a product, with the provenance stack shipping alongside as C2PA metadata, graded “necessary, insufficient, present” by the file), and Google counter-launched Gemini 2.0 Wednesday (explicitly framed as “for the agentic era,” Project Mariner browsing autonomously, Astra seeing through cameras, and the watch-list item now the strategy slide of both giants simultaneously: 2025 is being pre-announced as the year models stop chatting and start doing, and the agent-spring’s expensive recursion meets its second, better-funded spring). The file’s eval-discipline note, always: “agentic” converts error rates from per-answer to per-action-sequence, compounding failure probabilities across multi-step tasks is the reliability math the demos elide (a 95%-per-step agent is a 60%-per-ten-step agent), and the orgs that instrument sequence-level golden sets first will own the trust layer the way the rehearsers own reliability. Pre-registered for 2025’s grading.
Google’s other launch mattered differently: Willow, their quantum chip, demonstrated below-threshold error correction, logical qubits whose error rates drop as physical qubits are added (the old checkpoint, “watch error-corrected logical qubits, currently approximately zero,” formally passed eight years later), plus a benchmark septillion-year classical-comparison claim the file discounts per the IBM-rebuttal doctrine (the goalposts will move; they should; that’s the sport). The honest summary for the shelf: fault-tolerant quantum computing graduated this week from “if” trending toward “when,” with “when” still measured in years-to-decade for cryptographically-relevant machines, but the crypto-agility homework (post-quantum migration inventories; NIST’s standards shipped this year) converts from prudent paranoia to scheduled maintenance, and the file notes that the industry that took four years to patch Heartbleed now needs a decade-scale key-rotation it should have started yesterday (the dependency-tiering RFC gains a cryptographic appendix; the invoice always arrives).
The month’s ledgers otherwise: the California file’s federal echo, the incoming administration’s AI-policy signals (the EO’s fate clarifying toward repeal-and-replace; a “czar” appointment from industry) keep the geography redrawing, and the archive keeps minutes; Devin-class coding agents and their benchmark controversies continue the taxonomy’s maturation (the gap between “solves scoped tickets in demos” and “survives our repo” remains the golden-set’s favorite chasm); and the sports ledger notes Mbappé’s massive contract details resetting the contract-engineering meta within six weeks of its filing (the term-sheet arms race compounds; the file’s twelve-sentences doctrine adds no thirteenth).
TIL: quantum error-correction thresholds. The surface-code crossover where adding physical qubits helps instead of hurts (noise below threshold compounds correction faster than error): the entire field was a bet on reaching one inequality, and Willow’s data says the inequality is real. Some walls are load-bearing until one measurement; the replication machine now gets its decade with this one.