Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
An Algorithmic Failure Beneath the Secret Ballot
Princeton CITP fellow Max Springer shows that, using AI tools and only public records, he reconstructed the order of 1.5 million ballots (98.9% of in-person ballots) in Georgia's May 2026 primary by reversing a deterministic shuffle flaw in ballot scanners disclosed in 2022, and urges jurisdictions using Dominion scanners to install version 5.17 or newer.
7 min · 1,688 words
There's no point at which turning your brain off will work
Dan Luu argues that turning off critical judgment while using LLMs fails in practice: when people let models take actions or produce analyses without verification, errors compound — there is no safe point at which you can stop thinking.
12 min · 2,831 words
GenRec: Towards LLM-Native Recommendation at Netflix
Recommendations sit at the heart of the Netflix experience. Our current production models rely on thousands of hand‑crafted features over users, items, and interactions, along with specialized architectures for sequence modeling, feature interactions, and multi‑task objectives. This stack has evolved over many years to support diverse content types (movies, series, games, live, podcasts) and product surfaces, but its complexity makes it costly to onboard new use cases: adding a content type or surface can require significant feature engineering, architecture change,...
12 min · 2,813 words
Meet Stripe's Knowledge AI Platform
Stripe's Knowledge AI Platform is our versatile AI agent platform built to handle diverse non-coding knowledge work, from quick queries to complex, multi-day projects. By connecting employees to over 1,000 internal tools and skills, it enables secure, enterprise-scale productivity across the organization.
8 min · 1,870 words
The Best Code Review Says Less
You open a pull request and there are 40 inline comments waiting for you, all from the AI reviewer. It has opinions about a variable name. It wants you to extract three lines into a helper. It found a null that can’t actually occur, on a path that never runs. It’s suggesting a micro-optimization on code that executes twice a day.
5 min · 1,185 words
AI and Math in 2026: a non-mathematician's read
xlr8harder synthesizes recent AI math results for non-specialists: real progress with a different strength profile than humans, formalization cliffs, and why “math nearly conquered” remains overhyped.
3 min · 750 words
Expenditure Horizon: Measuring Optimization Ability, with an Application to NanoGPT
METR proposes expenditure horizon—a measure of an AI agent’s optimization ability—and illustrates it with the NanoGPT speedrun, including human expenditure estimates judged with an LLM.
41 min · 9,521 words
Self-generated prompt injections in compaction summaries
Research on aligning AI with human values and intent, and reports documenting model failures.
6 min · 1,350 words
From Scrum to Shape Up in the AI Era
When AI made coding the fastest part of delivery, two-week sprints started to burn the Intrepid Developer team out. Chris Vanderplank describes shifting to a Shape Up-inspired model with longer cycles and appetite-first design.
10 min · 2,350 words
Companies hire more after AI adoption
Ramp lead economist Ara Kharazian summarizes the Ramp Economics Lab and Revelio Labs working paper on 21,000+ US firms: heavy AI adopters grew headcount about 10% and entry-level headcount 12% in the two years after adoption, low-intensity adopters saw no significant change, and gains were uneven across firms.
5 min · 1,148 words
Helping build shared standards for advanced AI
OpenAI argues for U.S.-led shared technical standards for frontier AI—including evaluation, incident reporting, and cautious treatment of recursive self-improvement—via the Appia Foundation.
3 min · 708 words
The National Design Studio open sources Rampart, a 14.7 MB on-device PII filter for AI chat that runs in the browser. It pairs validated regular expressions for structured data like SSNs and card numbers with a small MiniLM model for names and addresses, and reports 98.4% private-term recall across seven languages.
4 min · 841 words
Reviewing code in the agent era
Reviewing code in the agent era Last week we launched Diffs, a new way to review PRs directly inside Linear. Since then we’ve had a lot of questions about it, so we wanted to share more about why we built the feature and how we use it ourselves today. Like a lot of people who build software, I f...
4 min · 870 words
"Vibe coding" is the new "Internet dating"
An analogy between early-2000s internet dating stigma and today's vibe-coding skepticism: both start as jokes, then quietly become how a lot of people actually meet their goals.
1 min · 199 words
AI-generated posters don’t have to be horribleA practical workflow for event posters that don’t look like every other AI flyer
John Hartnup shows how to stop AI event posters looking identical: constrain composition, type, and color with concrete prompts and layout rules so the result still looks designed rather than generically generated.
6 min · 1,414 words
AI Now Writes as Many Online Articles as HumansGraphite’s Common Crawl sample finds primarily AI-generated articles plateaued near 50%
Graphite’s Five Percent research averages three AI detectors across tens of thousands of English articles and finds primarily AI-generated pieces have plateaued near half of new articles since early 2025—after a steep rise following ChatGPT’s launch.
8 min · 1,859 words
Put a price on research breakthroughs
Alex Wang argues that AI neolabs are being asked to pull off two moonshots at once, a research breakthrough and a venture-scale business, and proposes a pharma-style model where frontier labs put explicit prices on breakthroughs such as cheaper training runs so neolabs can focus on research and sell or be acquired.
8 min · 1,767 words
Giving Your Home AI Agent Real Tools: MCP Servers on a Mac mini M6
A walkthrough of the MCP servers James M runs on a Mac mini M6—filesystem, email, calendar, notes, home automation—and the permission choices that keep an always-on home agent from becoming a liability.
8 min · 1,743 words
The fall of the theorem economyHow AI could destroy mathematics and barely touch it
David Bessis argues AI is exploiting mathematicians' honor-code incentives: when theorems become cheap to produce, prestige-driven theorem economies collapse even if mathematical understanding itself is barely touched.
47 min · 10,783 words
We Audited 10 Popular Open-Source Robot Datasets. Here's What We Found.
Traceplane ran automated quality checks on ten widely used open robotics datasets and found structural or semantic issues in every one—arguing trajectory data needs ingest-time QA like every other data-intensive field.
11 min · 2,503 words