The Colophon

Not Made on Wix

The footer says this site was made on Wix. It was not. If that needed saying out loud, the rest of this page is going to be a long afternoon for both of us.

What it actually is

A reference index built on top of a comedy podcast. Thousands of hours of a man talking, cut into pieces small enough to find again, with every claim on every page pointing back at the second it came from. The whole thing exists to answer one question — you remember a bit and you want to hear it again — and everything else here is scaffolding around that.

What happens to each episode

Every episode goes through the same passes. Each one does a single job, reads what the last pass wrote, and leaves its results for the next.

Find it. Work out which episodes exist, when they aired, how long they run, and whether they are an episode, a clip or a guest spot elsewhere.
Get the words. Pull the captions, with a timing on every line. This is where most of the errors come from — they are auto-generated, and they mangle names constantly.
Cut it into stretches. Split the episode by subject, because he drifts and the drift is the material. A stretch is usually three to eight minutes: one rant, one ad read, one caller.
Mark the moments. Inside each stretch, a one-line note every 45 to 90 seconds saying what is happening. These are what search actually matches against most of the time.
Note who and what came up. Read each stretch and record every person, place, company and subject in it, keeping his own words as the evidence.
Make it searchable. Two indexes: one for the words themselves, one for what they mean, so "the bit about his Irish family drinking" finds the right moment even when it shares almost no words with it.

How the mentions get sorted

Everything he brings up lands in one of four buckets — people, places, companies, subjects — and each of those is split further, so a page is browsable instead of being one enormous list.

People split by what they are to the show: comedians, politicians, business, the Epstein cast, commentators, musicians, actors, influencers, criminals, royals, and the people in his own life.
Places split into countries, US states, cities, neighbourhoods, regions, and specific sites — the Bohemian Groves and Mar-a-Lagos.
Companies split into corporations, networks and studios, news outlets, social platforms, restaurants, comedy clubs, hotels and institutions.
Subjects split into shows, films, podcasts, conspiracies, politics, tech, money, health, crime and war, food, belief, and American life.

Those groups are not permanent. The first version was drawn up when a fraction of the show had been processed, and it was wrong in ways that only became obvious with all of it indexed — one bucket had swallowed half the entries, and another was holding seven different kinds of thing at once. They get redrawn when the material says they should.

The same goes for names. Auto-captions spell a name a different way every time, so the same person arrives as several — one producer showed up eight ways, and one guest never once appeared spelled correctly. Those get folded together, and the misspellings are kept so the next episode resolves to the right page instead of starting a new one.

What a machine decides, and what it doesn't

The reading is machine work; there is far too much of it not to be. The judgment isn't. Which phrases count as running bits, which episodes to hand a newcomer, whether two spellings are one person or two people who sound alike — a person decides those, and nothing gets to add itself to those lists on its own.

Some of what the machines produce is wrong. There isa whole section about that, because pretending otherwise would be worse.

What is not here

No audio, no video, no copies of anything. Press play and you are being handed to wherever the show is actually published; this site only knows where in the recording to point you. Nothing tracks you, because nothing here wants anything from you. There is no account to make, nothing to buy, and no revenue anywhere in the design.