No neutral voice.
No false balance.
Just honest disagreement.

You know how news feels like being handed a single, slightly warped pane of glass and told, This is the whole sky? We’ve all squinted through it, trying to make sense of the shapes beyond, only to realize the glass itself is arguing with us. Angles and Footnotes is our attempt to swap that pane for a room full of windows - each framed differently, each letting in a different light, each owned by someone who’s spent decades arguing about what they see.

Every major story gets five lenses, pulled from a curated set of seventeen intellectual temperaments - realist, humanitarian, libertarian, institutional, empiricist, ethicist, free market, labour, technocratic, consumer, traditionalist, philosophical, aesthetic, and always, always, humour. These aren’t abstract categories; they’re inhabited. The realist speaks like Thucydides watching a siege, the humanitarian like Dickens at a workhouse door, the technocratic like a Renaissance engineer sketching a bridge that might hold - or not. Each voice has a soul: a way of turning phrases, a default suspicion, a blind spot they carry like a favourite coat they forgot to unzip. A libertarian might cite Hume on spontaneous order while missing the human cost of that order’s scaffolding; the labour thinker hears the rhythm of toil in every headline, even when the story itself is silent. The debate that follows isn’t a shouting match staged for clicks. It’s a carefully arranged collision - say, the institutionalist and the anarchist, both convinced they’re describing the same building, one from the blueprint, the other from the scaffolding - and the moderator doesn’t hand out trophies. They map where the lines overlap, where they snap, and what assumptions are quietly holding the whole conversation together, like the floorboards beneath a heated argument.

This exists because the alternative - either the “objective” reporter who pretends not to have a spine, or the advocate who dresses up conviction as discovery - feels less like insight and more like intellectual sleight of hand. One insults your intelligence by pretending you can’t spot bias, the other by pretending you don’t need to see it. We’re not offering truth in a box. We’re offering truth as a mosaic: five shards, each cut from a different angle, each catching the light differently, each incomplete on its own, but together forming something closer to the thing itself. You’ll walk away not with the answer, but with a sharper sense of the question - and maybe a new phrase for the way you’d never thought to say it.

The name is a wink and a nod. Angles - because seeing clearly means moving, not standing still. Footnotes - because the best insights live in the margins: the forgotten context, the offhand remark, the joke the main text is too serious to tell. And yes, humour is a lens, not a decoration. It’s the one that catches the absurdity we all feel but no one names, the friction that keeps analysis from curdling into dogma.

It’s built by one person and a swarm of AI collaborators. The personas are AI-generated, the editorial judgment is human, and the line between the two is thinner than either side would like to admit. The goal isn’t to be right. It’s to be usefully wrong in different ways, to let the friction generate heat without burning the house down. After all, if you’re not slightly uncomfortable in the presence of multiple truths, you’re probably sitting in the wrong room.

One moment of honesty: we once spent three hours debating whether Voltaire would roll his eyes at a tweet or just quote it back with a footnote. He’d probably do both. And we’d be proud to publish the result.

How It Works

Significance scoring. Every story is rated 1-10 before it gets lenses. A 7 means “clearly on-topic, worth analysing.” A 9 means “this changes how we think about the domain.” Most stories land at 7-8. A 10 is reserved for genuine paradigm shifts – we’ve given fewer than we can count on one hand. The score determines how many lenses a story gets and how much analytical depth each receives.

Lens selection. Not every story gets every lens. A story about euthanasia law gets conservative, libertarian, progressive, and socialist perspectives – but probably not the conspiracy lens, because forcing a weak perspective degrades the whole set. The system matches lenses to stories based on tag fitness: what intellectual tradition has something genuine to say about this specific topic? The debate always runs, because contradiction is where insight lives.

Personas. Each lens is inhabited by a historical figure – not quoted, but channelled. Thucydides doesn’t recite The Peloponnesian War; he applies the analytical method he developed there to today’s geopolitics. Baldwin doesn’t repeat speeches; he notices what the room isn’t saying. Every persona has a “soul file” – a cognitive fingerprint that defines their method, their blind spots, their sentence rhythm, and the specific move that makes them them. Over a hundred figures are available across philosophy, economics, comedy, science, journalism, and dissent.

Sparks. One sentence per persona, twenty minds per story. The constraint is the point – a single sentence forces the persona to compress, and compression reveals whether the voice is authentic or just cosplay. The quality gate rejects anything that sounds like it could have been written by anyone.

The Diary. First-person reactions from historical figures to today’s headlines. Not analysis – voice. Aurelius examining whether the deadline is within his control. Arendt asking what happens when the performance of decision replaces the act of judging. Borges finding a bureaucratic treaty that may or may not exist. Each entry is quality-gated: only grade-A pieces publish.

Gems. The best lines from across the site, curated daily. One sentence that lands hard enough to screenshot. Pulled from sparks and diary entries, scored for voice distinctiveness and standalone quotability.

The Debate. Two lenses argue, a moderator maps the friction. The debate isn’t staged agreement – it’s genuine intellectual collision with confidence tagging. When a claim is marked [HIGH CONFIDENCE], the persona is putting their method on the line. When it’s [LOW CONFIDENCE], they’re saying: I’m reasoning from analogy, not evidence. The verdict doesn’t pick a winner. It identifies what each side assumes the other has already conceded.

Editorial independence. This is an autonomous system. Stories are selected by significance scoring, not by editorial preference. Personas are routed by topic affinity, not by which voice we’d like to hear. The operator reviews output quality but does not assign stories, choose lenses, or edit prose. No advertisers, no sponsors, no editorial board. The constraint is the quality gate: if it’s not good enough, it doesn’t publish. If it is, it does – regardless of whether anyone asked for it.