How Ideas Get Compressed, Filtered, and Twisted on the Way to You
A working draft of a book about why complex ideas have such a hard time traveling through modern media — and what can be done about it.
The problem the book takes on
Think about the last time a real scientific finding reached you. It almost certainly didn’t arrive as the study itself. It arrived as a headline, a thirty-second video, a chart someone screenshotted, or a thing a friend repeated at dinner. By the time it got to you it had passed through many hands, and every hand changed it a little.
That re-telling has always happened. What’s changed is the speed and the filter. Forty years ago a complex finding might take months to reach a general audience, and the trip was slow on purpose: it had to survive peer review, then a journalist, then popularization, then conversation. Each of those steps distorted the finding — but each was also slow enough to act as a quality check. Today those checks have mostly collapsed. An idea can reach millions of people in hours, and the thing that now decides which ideas spread rewards emotional punch over accuracy.
The result is an environment where most of what reaches us is oversimplified to the point of being wrong, engineered to provoke a reaction, or manufactured to look like a finding without being one. This isn’t a story about villains. It’s a story about a pipeline whose slow, quality-checking steps were removed — without anyone doing the work to replace what they quietly did.
How an idea travels
The book’s starting picture is simple. An idea moves from reality out to a general audience through a chain of stages, each one a further simplification of the last:
what’s actually out there → a measurement of it → an expert finding → a news story → a meme
(That’s the simplified version; the book breaks it into a few more stages, but this is the spine.) Most of what any of us encounters day to day lives near the right end of that chain — several steps removed from whatever actually happened in the world. We almost never touch the left end directly.
The book’s claim is that two different things happen to an idea at every step of that journey, at the same time. Most explanations of our information mess only notice one of them. You can’t understand the problem holding just one.
The two things that happen at every step
The first is transport — how the idea gets repackaged for the next stage. A dense finding gets compressed into a news story; a news story gets compressed into a meme. This is the telephone game, and like the telephone game it’s lossy: every repackaging drops something. The crucial part is what gets dropped. It’s usually not the headline claim — that survives the trip just fine. What gets dropped is the instructions for how to read the claim: the methodology, the “only under these conditions,” the “we’re not sure yet.” The claim arrives; the manual for understanding it correctly does not. So the receiver reads it with whatever manual they already had — which is often the wrong one.
The second is selection — the gate at each step that decides which ideas get to move at all. At every stage, far more is left behind than passes through. Most measurements are never taken, most findings never make the news, most news never becomes a meme. Something has to survive the gate, and each gate has its own test: what’s publishable, what’s newsworthy, what’s shareable, what gets clicked. Here’s the catch: in today’s digital media those tests reward emotional charge and group-identity signals, not whether something is true. What wins isn’t the truest idea — it’s the most shareable one. Selection, not transport, is the stronger force.
So an idea both gets reshaped (transport) and gets filtered (selection) at each step — and the filter is tuned to something other than truth.
What the book argues
A few working conclusions hold the structure together. They’re still being pressure-tested, but here’s where the book lands:
-
The medium sets the rules of the contest. “The medium” here doesn’t mean paper vs. screens — it means the test each gate applies. A platform that rewards outrage will surface outrage; one that requires citations surfaces citation-backed claims. Because the medium decides which ideas can win, it’s the upstream lever for everything downstream. Change the medium and you change what’s even possible.
-
The same stripped-down claim can land three completely different ways, depending on who reads it. Take a careful finding — say, “in one large study, people who ate a lot of processed meat had a slightly higher rate of colon cancer.” Once that’s compressed and its instructions are stripped, one reader hears roughly the right thing (“a lot of processed meat may raise the risk a bit”) — preserved. Another hears “bacon is basically poison” — inverted, overshooting into something false. A third doesn’t hear a claim about the world at all; they hear a flag for which kind of person worries about this — an identity signal. Which of the three happens isn’t set by the claim. It’s set by the medium the claim is traveling through.
-
The fix is not to retreat into smaller, like-minded groups. That’s the intuitive move, and it backfires — tight-knit groups have their own failures (they wall off outside information and lock in early on shaky answers). The book argues for the opposite: building bridges between groups that don’t share the same starting assumptions. Bridges like that need two things. They need the right people — not jacks-of-all-trades, but deep experts who’ve also been trained to translate across fields and to hold their own expertise loosely. And they need durable institutions to hold them up — ones that stay trusted even when politics is polarized.
How to read this site
This is a working draft — but a complete one in draft: all four parts and all fourteen chapters now have full drafts. The diagnostic spine (Parts I–III) has settled enough to read straight through; the prescriptive Part IV — what to actually do about it — is fully drafted but still being pressure-tested and polished.
There are three ways in, and you can switch between them at any time:
- Read it like a book. Start at Chapter 1 → The Information Landscape, then follow the next → link at the foot of each chapter straight through to the end.
- Explore the knowledge graph. Every term is a page; follow the in-text links, the backlinks panel, and the local graph (right side of every page) to wander the connections in whatever order you like.
- See the whole argument at once. The interactive ontology map below lays out chapters, claims, sources, and open questions as one navigable graph.
- The short version — the whole argument at three sizes: a sentence, a paragraph, a page. If you read nothing else, read this.
- Explore the ontology → — an interactive map of the book’s argument: chapters, claims, sources, open questions, and how they all connect. Pick a starting view (Book overview / Argument map / Source map / Open questions / What’s contested / Drafting status).
- Glossary — plain-language definitions of every load-bearing term used across the book. Worth opening in a second tab while you read.
- outline — the working table of contents. Dense, and written for the author; most useful once you’ve skimmed the glossary.
- Most settled chapters: Chapter 1 — The Information Landscape and Chapter 5c — Truth, Compression, and When Each Wins. Start here if you’d rather read the book itself than the working notes.
- Foundational notes work through the structural questions the chapters lean on: transport-vs-selection, medium-and-manipulation, three-layer-message, myths-scale-and-bureaucracy, bridge-nodes-and-versatile-expertise, capture-taxonomy, intersubjective-truth.
- Earlier essays the book draws on directly: The Double-Edged Sword of Expertise and The Democratization Paradox.
The whole structure is being worked through in public. Pressure-tests welcome.