All desks

A tested Go proxy routes Claude Code by model and effort, while accounting for unfinished work, cache rebuilds, explicit overrides, and hosted-data boundaries.

A history of neighborhood socialist organizing: tenant struggles, political education, and Palestinian solidarity built trust that an election campaign alone could not.

When agents make code cheap, review and integration become the constraint. Anthropic's migration shows both the opportunity and the unanswered questions.

A feed reader's launch brought persistent bot load, not just readers. The discussion adds countermeasures and the cost of blocking legitimate visitors.

A live test finds real failures in 100 leasing-assistant conversations, but premature test design, brittle grading and a poor review interface undermine the workflow.

A practical framework for asking more of AI: bigger outcomes, deeper methods, and wider disciplines—with independent tests instead of confidence as the proof.

A Laravel/Filament plugin turns plain-English statements into row filters, with content-keyed caching, bounded requests, and an honest synthetic-label comparison.

A reproducible English hate-speech benchmark compares local decision models and small supervised classifiers, while keeping hosted Jev and legal judgments outside its claims.

Grotesque greeting cards expose the gap between making everyday life artful and making art accessible.

A new biography follows Debs’s political education—and asks why personal anti-racism never became an adequate party strategy.

Peter Yang compares code-built scenes, games, storytelling, websites, and writing—and finds different winners for different tasks. These are selected demos, not a controlled benchmark.

Meet Cornell researchers in a Brooklyn community garden for a free conversation about beneficial insects and an on-site bee nesting study.

A synthetic caller-question study favors whole-page selection with a none option over Sayora’s production embeddings, but missing details in summaries can hide the right answer.

SAP reports stronger ticket-assignment accuracy from its tabular model than Jev on the same recent examples; the latency comparison uses a direct H200 run, not a deployed service.

Jameson Durall on how expert walkthroughs distort level reviews and invite the wrong difficulty changes.

A practical case for separating the model you use from the company you trust—with HN pushback on privacy, cost, and coding quality.

Behind the honorable-outlaw fantasy: cinema, invented tradition, strikebreaking, and organized crime's alliances with state power.

Tuesday’s nine picks: jazz and a UWS movie club tonight, Hamilton Heights watercolor tomorrow, a free Friday horror screening, and ongoing art and city walks.

Tuesday pairs a major hosted-model release with practical reminders that agent reliability depends on the surrounding system, not just the model. Meta’s Muse Wonder targets sustained coding and personal-agent work, but its memory features …

A substantive unofficial .NET client generates question sets at compile time, separates protocol handling from policy, and offers offline samples—but its calibration language needs evidence.

An inspectable delegation plugin keeps recommendations, actual execution, and accepted code quality separate—and refuses to call its retained-window report an accuracy benchmark.

Getting a result is not the same as acquiring the skill that producing it would have taught you.

The useful divide is not tools versus automation, but informed use versus delegation without understanding.

Reporting points to an early FTC review of sportsbook advertising, with no public agency confirmation and prediction markets reportedly outside its scope.

BetMGM’s quarterly figures show fewer active customers, higher sportsbook revenue per player and a larger role for parlays and online casinos.

An agent-fleet research repository publishes failed classifier experiments, conservative sampling bounds, and explicit boundaries between recorded evidence and live validation.

Atul describes scoped code and screenshot reviews inspired by Jev—but production uses Perplexity’s decider, and model scores remain hints.

A forecast that AI will turn paid thinking into a leisure pursuit—and a discussion that asks who can afford that leisure.

Branko Marcetic argues that deterrence without dialogue risks a wider war, while postponing peace further devastates Ukraine.

Hamel Husain's seven eval pitfalls: imaginary failures, generic scores, unvalidated judges, opaque interfaces, rating scales, the wrong reviewers, and over-automation.

A Google DevSite RCE writeup shows how readable session state, credential reuse, and pickle deserialization collapsed a trust boundary.

A small Go library separates typed decision handling from model setup, rejects inconsistent distributions, and runs bounded background comparisons.

An inspectable Kev/Qwen MCP project includes threshold fitting and honest small-sample results, alongside safety claims its evidence cannot support.

Silas Xuereb argues that fighting inflation by weakening workers is a political choice, not a neutral law of economics.

Unpredictable model output does not erase deliberate choices about tools and internet access. The HN discussion challenges the leap from foreseeability to criminal intent.

How an abandoned Richard Corben adaptation helped Miyazaki find Nausicaä—and a new kind of storytelling.

Cole Medin argues that stronger coding agents increase the value of clear intent, concrete references, and human validation—not passive forwarding of AI output.

See live birds of prey up close at Brooklyn's Salt Marsh Nature Center on Saturday, October 17. Free, with no registration required.

Antique architectural ornaments from Tottenville's Atlantic Terra Cotta Company form a free, self-guided sculpture garden at Biddle House.

Browser-agent experiments separate completion checks from routing—and show why fewer executor tokens are not the same as a cheaper task.

A vendor case study reports cheaper first-stage screening with similar shortlist recall; projected savings and human-selection labels need careful interpretation.

AI does not have to replace an experienced engineer to persuade management to eliminate the role.

A documentary fandom turns disaster analysis into reassurance—and sometimes an uncertain belief in its own expertise.

Monday brings a Barclays concert, tomorrow’s Astoria jazz, and a last chance to see MCNY’s handmade city model; Brooklyn Museum reopens Wednesday.

Monday’s useful AI news is about removing bottlenecks without mistaking local improvements for better outcomes: GraphTok moves exact tokenization onto the GPU, targeting work that prefix caching leaves behind rather than promising faster …

A Berkeley mathematician on AI-assisted research, the limits of quantum speedups, and why foundations still matter when answers are cheap.

A neighborhood simulation makes remote NPC reactions optional, bounded and visibly distinct from local rules—without establishing live Jev throughput.

An inspectable Chrome extension separates network rules from Jev page-element judgments, with local fallback, reversible hiding and important privacy limits.

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.