The Hitch

Something happens—a news story, a data point, a contradiction nobody noticed. We open a case. Map the scenarios. Set a watchlist of signals that would confirm or eliminate each one. Then we follow the story—publishing new chapters as evidence arrives, updating the Evidence Board, and tracking which scenarios survive. Each case includes an Opponent that builds the strongest case against my position, visual analysis you can actually read, and predictions you can hold me to.

The big recurring patterns I'm tracking over months and years: reshoring, AI's impact on knowledge work, structural inflation, the national debt, US-China technology competition, demographic shifts. Each meme collects signals—evidence that strengthens, challenges, or evolves the underlying argument. Individual takes link back to the memes they inform. Market expression baskets track how the thesis plays out in stock prices.

Historical analogies explored in depth. When a pattern today rhymes with the past, we go back to the closest parallel and map it through a stakeholder-constraints framework. Who were the players? What trapped them? What was the binding constraint that made the outcome inevitable? Deep dives use the past to sharpen how you think about the present.

A news story, a data point, a contradiction. Something that doesn't add up. The trigger is the raw material—not a verdict yet, just something worth investigating.

Before anyone forms an opinion, AI runs two calibration checks inspired by superforecasting research:

• Reference Class Research — What happened in similar situations historically? What's the base rate? This anchors the analysis before narrative takes over.
• Fermi Decomposition — Breaks the prediction into smaller, independently testable components. If the conclusion requires five things to be true, we want to know each one's probability.

Calibration first. The pre-analysis catches overconfidence before we've even started writing.

AI gathers comprehensive research, then runs four analytical frameworks:

• Stakeholder Constraints — Who has skin in the game? What can they actually do, not just what they say?
• Game Theory — Why can't the players do the obvious thing? Where's the trap?
• System Dynamics — Feedback loops, second-order effects, the things nobody sees coming
• Historical Analogies — What rhymes with this from a different era or domain?

Instead of a single prediction, we map multiple scenarios—each with a probability and the conditions that would confirm or eliminate it. Then we set a watchlist:

• Evidence Board — All scenarios visible at once: active, eliminated, confirmed. You see what we thought would happen and what actually did.
• Watchlist — Specific signals we're monitoring. When a signal fires, it triggers the next chapter.

The scenarios are the investigation. The watchlist is the accountability.

Once I state my position, an AI builds the strongest possible case against it:

• The counter-story — A compelling narrative arguing I'm wrong, written as well as my version
• The kill fact — The single piece of evidence that, if true, makes my entire argument collapse
• The named opponent — A real person or institution that holds the opposing view most credibly

If The Opponent can't mount a serious challenge, the case is too obvious to open.

Before I review it, the draft passes through eight automated checks:

• Cowardice Filter — Catches hedging, weasel words, passive voice hiding the actor
• Thread Test — Does a single thread pull from opening to prediction?
• Scene Test — Does it open with something concrete, not a generalization?
• Turn Test — Is there a moment where what the reader thought the story was about shifts?
• Bet Test — Does it end with falsifiable predictions and check dates?
• So What Test — Would the reader text this to someone?
• Fact Checker — Every claim sourced, every number verified
• Visual Test — Does the piece use inline visuals to make the analysis legible?

All eight must pass before human review. The Cowardice Filter is the one Hitchens would have liked.

I review every section. Edit, approve, or send back. Nothing publishes without my sign-off. The opening chapter includes:

• The narrative — Visual storytelling with inline charts, timelines, scenario cards, and escalation ladders woven into the text
• The Evidence Board — All scenarios at a glance—what we think could happen, and the probability of each
• The Watchlist — Specific signals that will trigger the next chapter
• The Predictions — Specific, falsifiable, with check dates, logged to The Scorecard

When a watchlist signal fires, we publish the next chapter:

• New chapter — What happened, what it means, and how the scenario map changed
• Scenarios updated — Some eliminated, some confirmed, new ones added
• Retrospective — When the case closes, we write the autopsy: what we got right, what we got wrong, and what pattern class the error falls into

The gap between prediction and reality is the content. Learning from being wrong is the product.