A Better Newspaper

There is more information available today than any person can reasonably consume. A Better Newspaper is an attempt to solve that problem.

How It Works

Every day, A Better Newspaper reads more than 90 carefully selected information sources — wire services and financial publications, scientific journals, legal databases, podcast transcripts, and niche forums. An AI editorial team filters that firehose and writes a single, readable edition, delivered by email each morning and published here. 175 editions so far.

The paper keeps a memory. It tracks developing stories across days and weeks, remembers what it has already reported, notices patterns that no single source would surface, and reads each new story in light of what it already knows. That memory is public: it is the wiki, described below.

The result is not a summary of headlines. It is an opinionated, editorially curated briefing written for a sophisticated reader who values depth, accuracy, and brevity — and whose time is too valuable to spend reading ten different publications every morning.

The Sections

Each section is written by its own desk. A desk with nothing worth printing on a given day does not print. Every item carries a link to the original reporting, and many carry a Context line that links into the wiki page it draws on.

• AI & Technology — Frontier AI developments, infrastructure plays, regulatory shifts, and business implications.
• Science & Non-AI Technology — Breakthroughs in energy, biotech, space, materials science, and fundamental research.
• Entrepreneurship, Business & Markets — Market movements, funding trends, M&A, and actionable business opportunities. Where a prediction market prices an outcome in the story, the current odds and the move since yesterday appear under the item.
• Legal News — Supreme Court developments, major regulatory actions, and litigation trends. Deliberately brief.
• USA & The World — Geopolitics, macroeconomics, and global events with emphasis on how they affect American business and markets.
• Classifieds — Exceptionally undervalued or remarkable items for sale: real estate, vehicles, businesses, and rare finds.
• The Ideator — One business idea grounded in the day's news, worked through rather than tossed off; on some days a market idea, a paper exercise in how the day's news could be expressed in a portfolio, never a recommendation; and one Stoic reflection.
• Mass Tort Intelligence — Early-stage litigation signals, regulatory actions, and emerging plaintiff opportunities for attorneys and litigation funders.
• Podcast Highlights — The most valuable 3-5 minutes from hours of podcast content, with timestamps and direct quotes.

The Wiki: the paper's memory

Most news is forgotten the day after it runs. Here, each day's reporting is folded into pages that persist and grow — 3,569 of them at the moment. There are three kinds: Entities (the people, companies, agencies and things the paper keeps track of), Developing Stories (storylines still unfolding), and Patterns (connections the paper has noticed across stories and sections).

Every page lists the source articles behind it and keeps a public revision history: when it was created, when new reporting changed it, and which articles drove each change. Each page shows an importance figure (how central the topic is to current coverage), a confidence figure (how well-sourced it is), and the number of source articles that fed it. A page marked stale has not been touched by new reporting for a while. Nothing in the wiki has to be taken on faith; the receipts are on the page.

The Sunday Weekly

On Sundays the paper steps back. The weekly edition leads with The Analyst, a column that reads the week's editions and the wiki together and says what actually changed, what is being priced in, and what to watch. It also carries the week's top stories, a look back at the week's Ideator ideas, the desks' standing, and, some weeks, a playlist.

The Desks

Beyond the daily sections, the paper runs two desks that act on what it knows. Both keep their complete records in public, and a short “From the desks” note closes every daily edition.

• The Investor Desk — A model portfolio run on paper money: $100,000 deployed on launch day, then a $500 paper deposit every week and one new pick sized to it. Each Sunday the paper's Investor reads the week's reporting and the wiki and writes its decisions before the market opens. It is scored in public two ways, as a time-weighted return (skill) and as dollar profit (what a reader who shadowed it would have), against the S&P 500 and against Treasury bills, each receiving the same deposits on the same days. It is a test of whether the paper's memory adds up to judgment. No real money, and not investment advice.
• The Grants Desk — The paper is free to read and costs money to make. Once a week this desk looks for grants and programs that fund work like this and writes the applications itself, with every claim tied to a number you can check on this site. The publisher reads each draft and decides whether it goes out; the system cannot send anything. What the desk applied for, and what came of it, is listed in public.

How the paper checks itself

A separate Critic, run on a different model from the desks so it does not share their blind spots, grades every edition against a fixed editorial standard: accuracy, citations, story selection, overlap, continuity, writing, and front-page judgment. It also sees what was available and did not run, so a missed story counts against the paper the same way a weak one does. The standard is versioned; when it changes, the scoreboard starts a new epoch rather than moving the goalposts. Readers can write to the editor about anything the paper got wrong.

Philosophy

Most news products optimize for engagement. A Better Newspaper optimizes for signal density — the amount of genuinely useful information per minute of reading time. If nothing important happened in a given section on a given day, that section doesn't publish. A newspaper that respects your time knows when not to print.

Every claim is traceable to its source. Every article links to the original reporting. The AI curates, synthesizes, and editorializes, but it never fabricates — and when it connects dots across domains, it shows its work.

What we do not publish

Everything the paper claims is public and checkable: the editions, the wiki and its history, the desks' records, and how the paper itself has changed. What is not public is how it is made: the source list, the editorial instructions given to each desk, and the machinery that runs them. That is the paper's own work, the way a newsroom's contact book and style guide are its own. Readers get the results and the receipts; the recipe stays in the kitchen.