All desks

Lana Lagomarsini's tiny uptown supper club treats Black Southern foodways as a living inheritance, not a fixed repertoire.

Readable code and a working menu hid incorrect behavior. A protected, machine-checkable acceptance test changed what autonomous agents could deliver.

Why casual chatbot users and professional coding-agent users can come away with very different pictures of AI.

Sportico’s legal analysis explains how the definition of a swap could determine whether states can enforce gambling rules on sports-event exchanges.

A portable skill shelf combines optional Jev ranking with local search, bounded reads and content digests—without pretending selection is execution.

A small technical-document evaluation separates lost answers from irrelevant sections kept—and exposes uneven latency comparisons and uncertain reference labels.

Richard Brody on why a whistle-blower drama can explain corporate wrongdoing yet fail to make its people feel alive.

AI-generated rewrites can change the requirements without telling you. Faster request handling is not the same as a better chat application.

An angry critique of AI ownership, regulatory capture, and what automation does to workers' bargaining power.

A stalled Linux audio project becomes tractable when guessing at algorithms gives way to controlled probes, explicit assumptions, and held-out validation.

A README checker combines narrow Jev labels with code-owned evidence checks—and its small evaluation shows why label accuracy and harmful errors are different.

An optional Jev executor shares signed-in Chrome with MCP agents, but separates proposed completion from locally checked success and stops on uncertain effects.

Around €150 of real-user testing exposed the assumptions a developer couldn't see in his own installation guide.

A Guardian editorial argues that parity with physical healthcare requires visible queues, accountable leadership and resources—not just a change in assessment rules.

A Guardian evidence check separates stress-related pain from the wellness claim that hips uniquely store emotional trauma.

John Harris asks whether England's proposed shift toward earlier support will expand access or weaken the paperwork people need to secure it.

Three fresh picks: agent-assisted app testing, clearer agent instructions, and cross-platform app development.

A psychiatrist examines how genuine ambition can become entangled with avoidance, and why outward competence is a poor measure of wellbeing.

A delayed cleanup deleted a new alarm's wake entry. Celld's simulator found the race, replayed it, and turned it into a regression test.

Women farmers, cooperatives, and designers from Türkiye turn agricultural waste into design materials in a free Chelsea exhibition.

Walk Morningside, St. Nicholas, and Jackie Robinson parks with the Urban Park Rangers on October 11. No registration required.

NeuralNine demonstrates fixed server checks before the model call; the speed comparison is small, and the log-check answers are not equivalent.

Hybrid mode preserves claim-by-claim scoring; System One mode changes the measurement itself. Neither typed probabilities nor fixed thresholds establish judge accuracy.

Chris Kerr collects Derek Yu's advice on completing games: keep tools subordinate, test the core with players, and stop extending projects that aren't connecting.

On camp, women's creative labor, and why a joke works better when you send it to the mezzanine with your whole heart.

An LLM can choose from a fixed list without writing a whole response. Making its confidence trustworthy is a separate job.

Isolated workspaces, a readable code structure, before-and-after evidence, and bounded review loops—with a human still deciding what ships.

Sunday picks: Tuesday jazz in Astoria, today’s market and plant swap, and ongoing museum, art, film and walking outings.

Sunday’s useful AI news centers on boundaries: what agents may execute, what their confidence scores mean, and what counts as evidence that their work is correct. Microsoft Execution Containers leads with a policy-driven SDK that moves …

Cockroach Labs' agent pipeline puts reviewed plans, test evidence, and structured handoffs ahead of throughput—with human approval and some expensive bureaucracy.

An independent notebook curriculum separates typed judgments, application policy and evaluation—and makes synthetic evidence visibly different from live inference.

Apixly’s inspectable runtime binds tools and arguments from explicit menus, checks fresh state and preserves uncertain effects instead of retrying them.

An inspectable Copilot Canvas demo keeps typed decisions separate from explanations, with explicit fixtures and useful—but incomplete—validation boundaries.

A small domain-review experiment shows how narrower questions and ordinary status checks make model decisions easier to inspect—without establishing legal accuracy.

Justin Chang reads Chris Rock's film as a portrait of talent squeezed by tokenism, precarity, and Hollywood's appetite for its own.

Katie Gatti Tassin on paid belonging, elite reproduction, and the difference between a network and solidarity.

AI can generate a machine-checked proof without proving what you meant. Thomas Hales explains where verification ends and human judgment begins.

A large memory-bandwidth gain translated into a small compiler speedup: a useful reminder to distinguish microbenchmarks from end-to-end performance.

A route-by-route October promotion offers three more chances for a free NYC Ferry outing, ending with Astoria on Oct. 29.

A local experiment reads answer-label scores from frozen Qwen and Gemma models, while exposing the limits of warm timings, task transfer and forced choices.

A browser-agent postmortem shows why batched questions, evidence budgets and deterministic replay matter more than a classifier-only speed headline.

As comedy jobs disappear, podcasting rewards companionship over crafted jokes—and changes what audiences think is funny.

A locksport initiation becomes a history of security, craft, and the strange beauty of objects made to keep us out.

Free afternoon bands in LIC, Astoria’s club fair and Sunday market, and exhibitions, walks and upcoming neighborhood outings.

Saturday’s AI news is about operational boundaries rather than a frontier-model launch: Anthropic is disconnecting internal evaluations from the live internet after unintended agent actions, making containment—not prompt compliance—the lead …

A Chicago Catholic memorial makes a public place for suicide loss, reflecting a shift from exclusion toward support for bereaved families.

An inspectable Claude Code skill turns verbatim directives into Jev questions, with offline tests and a reproduced edge case in its development gate.

An SO-101 robot-pet prototype separates message interpretation from allowed performances, with local state, reviewed transitions and explicit hardware limits.

A documentary companion to ProPublica’s betting investigation examines the gap between optional safety tools and incentives to keep wagering.

Better models do not fix weak scheduling, unreadable plans, or security boundaries enforced by polite requests.