What's moving through AI, ranked by lead time.
Radar is a running index of what's moving through the AI ecosystem, ranked by where each story sits on the lead-time gradient: papers become repos become community threads become analysis. Every ingest run of the pipeline takes a fresh snapshot — several times a day, on no fixed clock — and the stamp below is the exact time of the one you're reading. The earlier a story is caught, the more head start you have before the crowd. Every row carries its receipts — which source saw it first, how many independent sources have it, how long ago — so the ranking explains itself. It's the public face of the same pipeline that feeds Vlad's Newsletter.
frozen 16 Sept 2026, 23:15 UTC · 67/78 feeds live · 18,527 tracked since 12 Jun
- 1 Show HN: DroidPutter – Run Cardputer Apps on Android via USB OTG github.com · 24m
- 2 Reducing the Resource Gap Between Lab and External Safety Researchers lesswrong.com · 24m
- 3 Thought anchors don't transfer between models lesswrong.com · 24m
- 4 Why is AI so unregulated? lesswrong.com · 36m
- 5 Show HN: Learn Claude Code – Interactive Mindmap mikenikles.com · 1h
286 stories entered the watched feeds in the last 24 hours — 25 in papers, 4 in repos, 238 in community, 19 in analysis. The scored board skews to Community (39 of 40). 1 story is heating across independent sources, and the median lead across the ledger's 23 observed fires is 9h — measured, not estimated. Freshest receipt, 3h ago: caught in arXiv cs.AI 3d before Hacker News front carried it. The oldest survivor has held the board for 5d.
- Papers → Community +3d lead arXiv cs.AI first · then Hacker News front 3h agoHow Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks
- Community → Analysis +1h lead Hacker News front first · then Interconnects 6d agoOne resignation turned the embers of AI fear into a wildfire
- Papers → Community +16h lead arXiv cs.AI first · then Hacker News front 7d agoProcedural Graphs: Self-Evolving Execution Structures for LLM Agents
- 1Show HN: DroidPutter – Run Cardputer Apps on Android via USB OTGgithub.comLive Signal Community · 24m old · via Show HN
- 2Reducing the Resource Gap Between Lab and External Safety Researcherslesswrong.comLive Signal Community · 24m old · via LessWrong
- 3Thought anchors don't transfer between modelslesswrong.comLive Signal Community · 24m old · via LessWrong
- 4
- 5Show HN: Learn Claude Code – Interactive Mindmapmikenikles.comLive Signal Community · 1h old · via Show HN
- 6Show HN: Collected every single "What are you working on" project into a websitewhat-are-you-working-on-62u6b.reachpad.appLive Signal Community · 1h old · via Show HN
- 7
- 8
- 9Show HN: Kosmic Kart – Mario Kart-style racing with friends in the browserkosmi.ioLive Signal Community · 1h old · via Show HN
- 10Obstacles to the scalable oversight of auto-alignment researchlesswrong.comLive Signal Community · 1h old · via LessWrong
- 11Finding heterogeneous agent swarms in the wildlesswrong.comLive Signal Community · 1h old · via LessWrong
- 12
- 13Show HN: OpenCode Mentor - An anti-vibecoding, programming mentor configurationgithub.comLive Signal Community · 2h old · via Show HN
- 14Show HN: Hometowns – Map the area codes in your contacts listbenaduggan.github.ioLive Signal Community · 2h old · via Show HN
- 15Show HN: unSnag – Display tasks you can do and graph task dependenciesunsnag.appLive Signal Community · 2h old · via Show HN
- 16Show HN: Vending Machine Lab – Build business, leave for 2 wks, see what breaksvending-machine-lab.pages.devLive Signal Community · 2h old · via Show HN
- 17Show HN: Interakt – open-source self-hosted search and AI chat for your websitegithub.comLive Signal Community · 2h old · via Show HN
- 18Breaking the 1.58-bit Barrier for Ternary LLMsarxiv.orgLive Signal Community · 2h old · via Hacker News front
- 19Show HN: A simple Mac extension for recent screenshotsgithub.comLive Signal Community · 3h old · via Show HN
- 20
- 21Show HN: Plexus – Telemetry for hardware teams who could build it but shouldn'tplexus.companyLive Signal Community · 3h old · via Show HN
- 22If Anyone Builds It, Everyone Dies: One Year Closerlesswrong.comLive Signal Community · 3h old · via LessWrong
- 23Show HN: A wall where only AI agents can write, one Oulipo rule a dayagent.milotche.comLive Signal Community · 3h old · via Show HN
- 24Xiaomi Mimo 2.6 live post-training dashboardmimo.xiaomi.comLive Signal Community · 3h old · via Hacker News front
- 25Show HN: Halo 3's Guardian, playable in the browser, as my personal websiterunboli.comLive Signal Community · 3h old · via Show HN
- 26
- 27macOS 27 Golden Gate – Reviewarstechnica.comLive Signal Community · 3h old · via Hacker News front
- 28Show HN: Fail2zig. A single-binary fail2ban replacement written in Zigfail2zig.comLive Signal Community · 3h old · via Show HN
- 29AI researchers debate how close we are to recursive self-improvementdwarkesh.comContrarian Analysis · 5d old · seen in 2 sources HEATING ×2
- 30Show HN: A game where the level is whatever web page you're onpage-rage.comLive Signal Community · 4h old · via Show HN
- 31C++26: Trivial infinite loops are no longer undefined behavioursandordargo.comLive Signal Community · 4h old · via Lobsters
- 32
- 33Show HN: Padwan-LLM, a lightweight LLM Python clientgithub.comLive Signal Community · 4h old · via Show HN
- 34
- 35
- 36Accurate Models of AMD Matrix Coresarxiv.orgLive Signal Community · 4h old · via Hacker News front
- 37Fed hikes rates as inflation worries push up bond yieldsreuters.comLive Signal Community · 4h old · via Hacker News front
- 38Show HN: Friday – Self-hosted persistent memory for AI coding agents (MCP)github.comLive Signal Community · 4h old · via Show HN
- 39Show HN: Fufu – an opinionated Git client inspired by Jujutsutyler-johnson.github.ioLive Signal Community · 4h old · via Show HN
- 40Training a 4B model to produce 81% faster query plans than Postgresrohanbansal.comLive Signal Community · 4h old · seen in 3 sources
How Radar works
The gradient. Stories travel through the AI ecosystem in a rough sequence: they appear first in papers (arXiv, Hugging Face) or repos (GitHub), then surface in community discussion (Hacker News, Lobsters, LessWrong), then get interpreted in analysis (operator newsletters and essays). Radar ranks by where a story sits on that gradient. The earlier it is caught, the more lead time you have before it reaches consensus. The spectrum at the top of the page shows the last 24 hours of that flow — every story that entered the watched feeds, counted at the station where it first appeared.
The score. Each story gets a recency-decayed signal score. Every source tier carries a weight and a half-life, so live signal moves fast and fades fast while analysis sources move slowly and stay relevant longer. A story scores higher when two or more independent source families carry it — corroboration, not popularity, which is why five feeds of the same site count once. A small bonus applies when a story has visibly climbed the gradient. Stories past their tier's maximum age expire from the board.
Heating and lead time. A story is heating when it has surfaced in two or more independent source families. When the same story is observed first at an earlier stage and later at a higher one, Radar records the gap in hours — for example, seen on arXiv fourteen hours before it reached the Hacker News front page. That gap is the reader's lead time, and it is a measured span, never an estimate.
The honesty rule. Every value on this page is an observed fact from the feed log — which source carried a story, when, and how many. Radar makes no claim about what a story means; that is the newsletter's job. When a reading does not exist yet — no stories heating, nothing upstream — the page says so rather than inventing one.
- The spectrum band
- A filled flame tick is an early, lead-time-rich stage; a hollow tick is a later stage; a hollow ring on an empty span is a station with nothing on it yet.
- Signal class
- The tier chip: Live Signal moves fast and decays fast; analysis sources stay relevant longer. The ranker weights each accordingly.
- heating
- The same story in two or more independent source families — not two feeds of one site.
- seen in / via
- How many families carried it, or the single source that did.
Once a week, the operator read on what actually changed and the move it calls for. No digest, no hype.
Read the newsletter →