News, but calm.
The synchronization layer between reality and understanding.
For everyone who stopped reading the news —
and everyone drowning in it.
If you've stopped paying attention, you didn't fail the news. It failed you — structurally. Somewhere along the way, the business of journalism stopped selling understanding and started selling your attention. Headlines are engineered to withhold. Stories are engineered to be clicked, not finished. The important detail lives twelve paragraphs deep because your scroll — not your comprehension — is the unit of revenue.
Walking away was a rational response to a product that wasn't built for you. But the cost of a world where reasonable people opt out of knowing what's happening — while the loudest, least careful sources fill the silence — is one everyone pays. What's at stake isn't an industry. It's whether staying truthfully informed remains possible for a person with a life.
Here is the tax nobody names: every article re-explains the world from scratch, because no publication knows what you already know. You re-read four paragraphs of background to find the one sentence that's actually new. Multiply that by every story, every day, forever. The feeling of "I can't keep up" isn't a personal failing — it's the compound interest on that tax.
And the feed never ends. There's no moment where the news says you're done — you're caught up — go live your life. A product with no bottom is a product that never intended for you to finish. Calm isn't the absence of news. Calm is news with an edge you can reach.
We know it can be different.
Because a small team already did it once — in 2012.
Circa News rebuilt the article from first principles: stories became stacks of atoms — single verified facts, quotes, stats, and events, each a complete thought, each carrying its sources. You didn't just read a story; you followed it. And when something changed, Circa showed you only the new atoms. Never the rerun.
A newsroom of about twelve ran it. Editors at the top of the profession — from Reuters, from the Journal — did the work, and held it in higher regard than anything else on their résumés. The press called the format a breakthrough. Users called it their daily habit. Neither was exaggerating.
When the first US Ebola case broke, Circa had been quietly maintaining the story for months. The newsroom wrote one atom, placed it at the top, and hit send. Followers got a pure update. Newcomers got a complete, verified story. A WIRED reporter was standing in the room with a stopwatch.
During the most error-riddled breaking story of the era — while major outlets misidentified suspects on live air — Circa published through the entire Boston Marathon bombing without a single retraction. Not because it went slow. Because its rules were structural: nothing entered a story unverified; the unconfirmed lived in clearly-marked editor's notes, separated from the known.
That discipline wasn't a memo. It was an ethics policy, a style guide, and a newsroom culture where every publish, every update, and every alert was logged with a name attached — in the open, in front of peers. The craft was the point.
Circa died in June 2015, and the diagnosis was unanimous — every post-mortem, and the founders themselves, agree: the business failed. Twelve salaried humans were the cost floor, and the revenue model that fit the product — loyal readers paying for calm — barely existed as a category in 2015. Consumer subscriptions hadn't had their moment. Free-plus-ads demanded the very clickbait Circa refused to write.
The critics' sharpest product line deserves airing too: "traditional journalism with a good workout for your thumb." Meaning: the atoms never became the deeper structure they gestured at. Fair — the graph underneath was designed, and never got runway. Hold that thought. It matters in a minute.
A decade later, nobody has rebuilt the premise. And people still write the founders to say they miss it. Scholars gave it a name worth keeping: Circa was an attempt at an editorial-driven state machine of the world — a maintained, factual, current account of what's known. Not a feed. Not an archive. A state.
Maintain two states: what is known about the world, and what is known by you. News is the meaningful difference between them. Everything else — articles, apps, briefings, podcasts — is just a way of delivering that difference.
No fact is inherently news. A change becomes news for you when it differs from what you know, concerns something you care about, and matters enough to spend your attention on. That's why the same event is a one-line update for one reader and a full explainer for another — and silence for a third.
Every AI news product being built right now is additive — more summaries, more chat, more generated words on top of the same pile. Edition's promise runs the other way: we will show you less, and it will be exactly the right less. In a world drowning in fluent machine text, subtraction is the scarce thing.
And subtraction can't be faked, because it's an operation between two state models. To show you only what's new, a system must actually maintain what's known about the world and what's known by you — versioned, timestamped, per-person. No summarization prompt on top of a feed can do it. The moat isn't the writing. It's the two ledgers underneath.
Why the idea from 2012 becomes buildable in 2026.
Then: twelve salaries capped coverage. A story with three followers couldn't earn its keep.
Now: agents make the one-follower story affordable. "If one person cared enough to follow it, it gets followed up." The founding principle finally has an economy that permits it.
Then: 2015 had no consumer-subscription culture; calm was unfundable, and ads demanded bait.
Now: paying for software that respects you is a normal habit. Circa's natural model arrived — about two years after Circa needed it.
Then: "just trust our newsroom" was the only answer anyone could give.
Now: in a world where machine text is indistinguishable from reporting, provable process wins. Honest provisionality — claims that carry their evidence and admit revision — beats confident certainty.
Your news arrives as an edition: finite, finishable, with a beginning and — radically — an end. When you're done, the product says so and stops. Badge to zero. Nothing tugs at you. Breaking news is the only interrupt, and it's governed by the most conservative alert doctrine in the business.
And some mornings, the honest edition is the one on the right. No manufactured urgency. No content because content is the business. The system asks one question before it asks for your attention: do you actually need anything from us today?
Nothing important on your follows has materially changed since yesterday.
Go live your life. We're watching.
Under the hood, Edition maintains a knowledge graph: entities + events + claims + relationships + time + evidence. That's not architecture for its own sake. Each structure exists because a promise to the reader is impossible without it:
| Structure | The promise it keeps | Why an article can't |
|---|---|---|
| versioned atoms | "You read version 2. Here's exactly what changed." | articles overwrite; history evaporates |
| entities | Follow a person, a company, a bill — not just a story | a name in prose is just ink |
| events ≠ things | Real timelines; "how did we get here" on demand | the occurrence is buried in the telling |
| typed relationships | Hop from the 50-state tracker to your state's story | links don't know why they link |
| the claim ladder | "Reuters reports X" stays distinct from "X is established" | fluent prose flattens certainty |
| two clocks (bitemporal) | "What did we know, and when did we know it" | one timestamp can't answer that |
You shouldn't have to inspect the plumbing — the whole point is earning enough trust that you don't. But the moment anything feels off, the machinery is right there. Every sentence in every edition traces to its evidence: what kind of source, seen when, verified how, superseding what. Progressive disclosure — for epistemology.
Traditional corrections are a scar under an article you'll never revisit: "an earlier version of this story misstated…" Edition knows exactly who read the suspect acted alone ← superseded 4:12 PM — and tells them, personally: you read this on Tuesday; here's what changed, and here's why we believe the new version.
There's a symmetry hiding in that. The world changed → tell me what changed. Our understanding changed → tell me what changed. From your side of the screen, those are the same operation: your model is stale, and Edition updates it. A correction isn't an embarrassment. It's the machine, visibly working. No publication on earth does this — because none of them can. They don't know what you read.
Yes: the reporters are agents. They watch primary sources — dockets, filings, transcripts, wires — around the clock, maintain the graph, and write under doctrine that humans seeded: an ethics code, a style guide, and a numbered rulebook for sensitive matters. Every action any agent takes is logged, attributed, and inspectable. The newsroom is a glass box, by architecture.
The humans didn't leave — they moved into governance. They hold the constitution: seventeen articles that no agent can override, amendable only by deliberate, recorded human decision. Verification scales with risk: a mundane fact sails through; a weakly-sourced claim sitting under thirty storylines faces a panel of adversarial checkers before it's promoted. And when a reader flags something, an advocate agent — the reader's, not the newsroom's — investigates and answers. With receipts. The letter to the editor, in its final form.
The same verified state renders as whatever fits your life: a morning email you can finish with your coffee. A five-minute audio brief for the drive. Three lines on a watch. A patient explainer for your ten-year-old — same facts, more context. A structured, receipted answer for your own AI assistant, so it stops hallucinating the news too.
And the bright line, written into the constitution: two readers asking "what happened?" get materially the same account. We'll adjust the depth, the vocabulary, the length, the timing — never the reality. Personalized presentation is a service. Personalized truth is how information systems go rotten. We refuse the second, structurally.
Come back to a story after eight months away and no one makes you feel behind:
You last checked this story in December. Here's the state you left with — and the four things that have materially changed since.
Not recommendations. Not history. Continuity — the most natural information need humans have, and the one no publication can serve, because publications model neither what the world knows nor what you do. Edition models both. That's the whole trick, applied kindly.
Every media business optimizes something. Feeds optimize minutes. Social optimizes impressions. SEO optimizes pages. Edition optimizes exactly one ratio: how much you genuinely understand, per minute you give us. Every architectural choice in this deck — the two states, the delta, the finite editions, the silence — points at that ratio.
It inverts everything. The product wins when you understand something and leave. It fails when it wastes your attention — even pleasantly. That single inversion is why this can't be a feature bolted onto an engagement business: the incentive is load-bearing, so it has to be the foundation.
Circa was the first implementation — from 2012 — of an idea whose necessary technology hadn't arrived yet. The technology arrived. Edition is the second implementation.
This time, the article isn't the database. The app isn't the product. The newsroom isn't the humans. And the reader isn't an anonymous endpoint at the bottom of a publishing funnel. The product is the synchronization layer between reality and understanding — and it's finally buildable.
EDITION · edition.is · a spiritual successor to Circa News
every claim in this deck carries a receipt; sources: the Circa reconstruction (2026), an 80-source external reception study, WIRED (2014), Coddington (2019), & the founder record
a thought experiment — none of this is real yet · explore the full proposal →
drafted overnight by the newsroom's first agent, while the founder slept