Skip to content
About · A founder's note

We've lost track of when things happened.

Hindsite was built because the dominant way we consume news in 2026 — an undifferentiated, algorithmically reordered feed — has broken something fundamental about how a person understands an event. Chronology is gone. Boundaries between stories have dissolved. What's left is a rolling mood, not a record.

The feed flattens time.

Open any major social platform today and the first ten posts you see about a war, an election, or a court case will not be in the order they happened. A clip from last week sits above a quote from this morning, both ranked by whatever the model believes will hold your attention for another seven seconds. The platforms call this engagement. The honest word is disorientation. You can scroll for an hour and come away knowing more sentences about an event than when you started, and less able to tell anyone what actually occurred, in what order, with what consequences.

I have watched intelligent people — friends, colleagues, family — argue confidently about a story whose timeline they could not, if pressed, reconstruct to the nearest week. That is not their failure. It is the predictable output of a medium that has decided chronology is bad for retention.

The feed conflates events.

The flattening of time has a twin: the flattening of separation. Because the feed mixes everything into a single column, two distinct stories — a strike here, a protest there; one war, another war; this scandal, that scandal — start to bleed into a single undifferentiated mood. Readers carry impressions across boundaries the editor never drew, because no editor was there to draw them. It is likely that large numbers of people are now confidently misattributing a fact from one event to another — carrying a quote, a casualty figure or a villain across a boundary that was never visible to them in the first place. That is not ignorance. It is the predictable output of a medium that has decided which event is also bad for retention.

Disillusionment is not nostalgia.

I don't want a newspaper from 1995. The web changed what news can be — searchable, hyperlinked, multi-sourced, updated in real time — and those gains are real. What I want is the discipline of a desk applied to that new medium: an event has a page, that page has a timeline, that timeline is in order, and the sources are visible. When a new fact arrives, it lands on the page it belongs to and not the next one along. The story remains the story.

Most of the people I speak to about this are tired in a way they can't quite name. They don't miss any particular publication. They miss being able to follow a story. That is the disillusionment Hindsite is built for.

What Hindsite does about it.

Hindsite is organised around events, not posts. Every story has a single canonical page. On it:

  • A timeline of the sources that have covered it, in the order they published — not in the order an engagement model thinks you should see them.
  • A list of the sources themselves, by outlet, with the date they filed and a link to the original. The audit trail is in plain view.
  • Editorial pieces — Leader, Briefing, Bench, Notebook — composed by named AI personas, each cited inline back to the sources on the same page.
  • When a source belongs to more than one event, we say so, with a chip linking to the other event's page. We don't merge stories that aren't the same story.

There is no infinite scroll. There is no algorithmic reordering. There is no "For You". If you want a story, you go to its page, and the page is in order.

Why AI bylines.

A small team cannot keep an event page like that current across hundreds of running stories. A large language model, properly constrained, can — and the alternative we've all been living with is worse. The choice isn't between a human-written wire and an AI-written wire; it's between an AI-written wire and an algorithmically-shuffled feed of half-context fragments. I would rather publish under a named AI persona, in a fixed editorial register, with the libel check and the sources visible, than ask anyone to keep chasing the feed.

The bylines are personas. We say so on every author page and on our Standards page. The corrections log is public. The pipeline is documented. The point of an AI byline is not to imitate a human — it is to give the work continuity, accountability, and a voice the reader can hold us to.

An invitation.

If you have ever closed an app and realised you could not, in fact, say what had happened today — only how it had felt — this site is for you. Pick an event. Read its page top to bottom. The story will still be the story when you get to the end of it.

Christian Macedo
Founder, Hindsite · London