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.
snapshot 25 Sept 2026, 18:53 UTC · 67/78 feeds live · 17,141 tracked since 12 Jun
- 1 Commodified Intelligence herecomesthemoon.net · 30m
- 2 Applications open: Winter 2027 AFFINE Alignment Seminar (due Nov 22) lesswrong.com · 30m
- 3 We need a better theory of polarization, because it's failing to predict the AI debate lesswrong.com · 54m
- 4 Show HN: Reliopt – Pareto-frontier, contract-gated optimization for LLM programs github.com · 2h
- 5 Show HN: Doom or Bloom, map your AI worldview with Jev doom-or-bloom.com · 2h
566 stories entered the watched feeds in the last 24 hours — 287 in papers, 8 in repos, 242 in community, 29 in analysis. The scored board skews to Community (38 of 40). 2 stories are heating across independent sources, and the median lead across the ledger's 25 observed fires is 12h — measured, not estimated. Freshest receipt, 5h ago: caught in arXiv cs.AI 6h before Show HN carried it. The oldest survivor has held the board for 11h.
- Papers → Community +6h lead arXiv cs.AI first · then Show HN 5h agoHard Stop: Kernel-Level Preemption and Containment for Rogue Agentic Execution
- Community → Analysis +8h lead Hacker News front first · then Dynomight 18h agoWhy is the human body so crap except for the liver?
- Repos → Community +7d lead GitHub Trending Python first · then Show HN 2d agoMG1937/ASC
- 1Commodified Intelligenceherecomesthemoon.netLive Signal Community ·Business · 30m old · via Lobsters
- 2Applications open: Winter 2027 AFFINE Alignment Seminar (due Nov 22)lesswrong.comLive Signal Community ·Policy & safety · 30m old · via LessWrong
- 3We need a better theory of polarization, because it's failing to predict the AI debatelesswrong.comLive Signal Community ·Policy & safety · 54m old · via LessWrong
- 4Show HN: Reliopt – Pareto-frontier, contract-gated optimization for LLM programsgithub.comLive Signal Community ·Tools · 2h old · via Show HN
- 5Show HN: Doom or Bloom, map your AI worldview with Jevdoom-or-bloom.comLive Signal Community · 2h old · seen in 2 sources
- 6
- 7Show HN: PromptPad- A lightweight editor for long coding promptsgithub.comLive Signal Community ·Tools · 2h old · via Show HN
- 8Show HN: Cafe(s) and the Context You Can't Write Downmetalbear.comLive Signal Community ·Practice · 2h old · via Show HN
- 9Show HN: Built an open-source AI browser assistant remembers when you scrollgithub.comLive Signal Community ·Tools · 2h old · via Show HN
- 10Show HN: Web search agent for finding leadsagentjesse.aiLive Signal Community ·Tools · 3h old · via Show HN
- 11U.S. appeals court upholds designation of Anthropic as supply chain riskcnbc.comLive Signal Community ·Policy & safety · 3h old · via Hacker News front
- 12Classified Estimates Show the NSA Is Paying Billions to Test AI Modelswashingtonsun.comLive Signal Community · 3h old · via Hacker News front
- 13Show HN: CiteRank, track whether ChatGPT and Perplexity cite your siteciterank.thecompound.techLive Signal Community ·Tools · 4h old · via Show HN
- 14Show HN: Misheard turns "recognize speech" into "wreck a nice beach"wordlab.rickinto.placeLive Signal Community · 4h old · via Show HN
- 15Show HN: Xtriever – offline RAG retrieval on a phonegithub.comLive Signal Community ·Tools · 4h old · via Show HN
- 16Show HN: Voice cloning for a TTS whose style encoder was never releasedgithub.comLive Signal Community ·Tools · 4h old · via Show HN
- 17Proaction boosts sales 60% and saves 75+ hours with Codexopenai.comAI Firehose Analysis ·Practice · -6m old · via OpenAI News
- 18Microsoft Abandons Personal AI Chatbot Race with Copilot Rebootbloomberg.comLive Signal Community ·Business · 5h old · via Hacker News front
- 19Show HN: Worktable, an open-source workspace for you and your agentsgithub.comLive Signal Community ·Tools · 6h old · via Show HN
- 20
- 21Show HN: I made a WWI dogfight game that runs in the browsergamedev.plLive Signal Community · 2h old · via Show HN
- 22
- 23Show HN: Xyp RSS – hold to play, swipe to skip, highlight a word for graphxyp.appLive Signal Community · 2h old · via Show HN
- 24
- 25CommonGrid: The open source registry of the US power gridcommongrid.infoLive Signal Community · 3h old · via Lobsters
- 26
- 27Show HN: Choreoglyphs – our new, invisible writing systemchoreoglyphs.comLive Signal Community · 3h old · via Show HN
- 28Show HN: Draw something once a day with friendsjustdraw.peerdiem.comLive Signal Community · 3h old · via Show HN
- 29Gravity Seems Holographic. What Does That Mean for Reality?quantamagazine.orgLive Signal Community · 3h old · via Hacker News front
- 30A Type Stronger than the Sum of its Componentsschneems.comLive Signal Community · 4h old · via Lobsters
- 31Show HN: HocusFocus – Pomodoro timer and Lofi Music player for your Mac's notchgithub.comLive Signal Community · 4h old · via Show HN
- 32Show HN: TandemCode – practice coding interviews together in one live editortandemcode.spaceLive Signal Community · 4h old · via Show HN
- 33
- 34
- 35
- 36What happens when you analyze your favorite college football team like the CIA?cultivatelabs.comLive Signal Community · 5h old · via Hacker News front
- 37
- 38
- 39Pentium II at 600Mhz with Voodoo 3 Emulated on 86Box with M6 Mac Mininyaa.shLive Signal Community · 11h old · seen in 2 sources HEATING ×2
- 40Show HN: Hamilton – a no-internet Android health dashboard, named after my dogplay.google.comLive Signal Community · 6h old · via Show HN
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 classifier. Two judgments on this page are made by a model, not read from the feed log. A small typed classifier (TypeSafe's Jev) answers one question per story — is it about AI? — and stories it judges not to be about AI move to the bottom of the board; nothing is removed. It was checked against 232 hand-labelled titles before it shipped and agreed with 97.8% of them. The same classifier assigns the topic shown on a story and used by the filter above; a topic appears only when the classifier is confident, and at that setting it matched the labels on 115 of 116 stories. Stories without a confident topic are listed under All.
The honesty rule. Every other value on this page is an observed fact from the feed log — which source carried a story, when, and how many. Beyond those two labelled judgments, 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.
Questions
- What is the lead-time gradient?
- The path a story travels through the AI ecosystem: it appears first in papers (arXiv, Hugging Face) or repos (GitHub), then surfaces in community discussion (Hacker News, Lobsters, LessWrong), then gets interpreted in analysis (operator newsletters, essays). Radar ranks by where a story sits on that path and how far it has travelled.
- What does the lead-time spectrum show?
- Every story that entered Radar’s watched feeds in the last 24 hours, bucketed by the gradient station where it first appeared — papers, repos, community, or analysis. It is the day’s flow through the whole watched ecosystem, not the composition of the scored top 40 below it: a paper counts in the flow the moment it is observed, even while its signal score is still too low to earn a board row.
- How is Radar ranked?
- By a recency-decayed signal score: each source tier has a weight and a half-life, a story scores higher when independent source families corroborate it, and a small bonus applies when a story has visibly climbed the gradient. Stories past their tier max age expire from the board. Popularity on a single site is not corroboration.
- How often does Radar update?
- On every ingest run of the pipeline behind it — several times a day, on no fixed clock. Each run re-renders this page, so the stamp under the headline is the exact time of the snapshot you are reading, and the live board shows how old it is. A frozen snapshot is archived at a dated permalink each day, so the record is both current and citable.
- What does "heating" mean?
- A story has surfaced in two or more independent source families — not two feeds of the same site. It is corroboration, not popularity.
- What is the lead time reading?
- When the same story is observed first at an earlier gradient stage and later at a higher one, Radar records the gap in hours — for example, seen on arXiv 14 hours before it reached the Hacker News front page. That gap is your lead time. The fires ledger is rolling: a fire ages out with its story, roughly 45 days after the story’s last sighting, so the median lead always describes recent motion.
- What sources does Radar watch?
- Around 78 verified feeds across five tiers: live signal (Hacker News, Lobsters, GitHub trending, arXiv, Hugging Face papers), AI labs and firehose blogs, operator newsletters, contrarian essayists, and primary analysis. The registry is curated; aggregators are used only to corroborate a story, never to be the first to report it.
- Is Radar the newsletter?
- No. Radar shows what is moving; the newsletter tells you what to do about it. Radar is the free, public funnel; Vlad’s Newsletter is the weekly operator read.
- 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 →