Gresham's Law Needs a Law

An Atlantic writer sat with that channel for three hours in August and filed the results: a man playing trumpet in fog, a smash cut to a janitor's uniform with a name tag reading V. Helsing, an animated vampire bellowing I am what I was born to be: Dracula! The channel is Fairground AI Creator TV, and it runs twenty-four hours a day on Roku's free ad-supported tier, playing AI-generated shorts from creators around the world. Some of the commercials are AI-generated too. The Guardian called it nightmare fodder. The Verge said watching it was like eating from a trough.

The detail that makes this a systems story rather than an outrage story is who built it. Fairground was founded by Colin Petrie-Norris, formerly the chief executive of Xumo — one of the original free ad-supported streaming platforms. The man who helped create the slots came back to fill them, with four million dollars in venture funding. That isn't hypocrisy. It's a career following the same gradient twice, which is usually what a system pattern looks like from the inside.

The commentary reads this as decline: taste collapsing, machines flooding the zone. It isn't decline. It's arithmetic, and the arithmetic has a name that nearly everyone quotes wrong.

---

i · the condition everyone drops

Gresham's Law: bad money drives out good.

It is old and it is well-traveled. Henry Dunning Macleod attached Sir Thomas Gresham's name to it in 1858; Nicole Oresme had described the mechanism around 1360 in De Moneta; Aristophanes was already complaining in 405 BC that Athens kept its debased coins in circulation while the good silver disappeared.

The full statement includes a condition, and the condition is what gets dropped in transit: bad money drives out good only where the exchange rate is fixed by law. Legal tender. If a shilling with the silver clipped out of it must be accepted at the same value as a whole one, you spend the clipped shilling and hoard the whole one. Obviously. But lift the rule — let a merchant weigh the coin and discount it on sight — and the law simply stops operating. The debased coin circulates at its actual worth, and the good coin comes back out of the drawer.

Gresham's Law was never a claim about human weakness. It was a claim about a pricing rule. Which means the useful question in front of any degraded channel is not why has quality collapsed. It is: who fixed the exchange rate, and when?

ii · the slot is legal tender

In free ad-supported streaming television, the unit of account is the impression sold against a slot.

A slot is a slot. An hour is an hour. A thousand impressions is a thousand impressions, and the advertiser pays the same rate whether the hour those impressions interrupted cost nine million dollars to produce or ninety. Production costs at the two ends of that trade differ by five orders of magnitude. Revenue per unit does not differ at all.

Not literally at all. Premium inventory clears above remnant; a recognized channel commands more than an unknown one. Call the honest spread a few times over. But a rate that varies by a few times over, applied to inputs that vary by a hundred thousand, is not floating in any sense that changes the outcome. It is fixed the way a shilling is fixed — near enough that no one at the counter bothers to weigh.

That is legal tender. Nobody passed a statute. The interface wrote it. A channel grid that renders every offering as one tile of equal size has declared them interchangeable at par — and once things are interchangeable at par, the cheapest producer wins on arithmetic that has nothing whatsoever to do with what anyone wants to watch.

Two months before the slop channel launched, Fox agreed to acquire Roku for roughly twenty-two billion dollars. Ben Thompson called the strategy smart and then spent most of his piece on its problem. Set the merits aside; the structure is enough. Fox made its money as a rights holder, charging whoever needed its content. It now also owns the counter where the slots are sold — and the unit of account at that counter is being debased from the inside. The owner of the mint is rarely the first to notice the clipping.

iii · the slots of 1984

The precedent is close enough to touch.

Two things happened to American television in 1984. The FCC abandoned its guidelines limiting how much commercial time a station could sell. And Congress passed the Cable Communications Policy Act, deregulating rates and accelerating channel proliferation.

The result was an enormous supply of new slots, uniformly priced, most of them overnight and nearly worthless. Nothing in the existing production economy could fill an overnight cable hour and break even. Exactly one form could: the program that is the advertisement. The infomercial arrived, and in July 1985 the Home Shopping Network went national by satellite — an entire channel on which the content and the revenue were the same object.

Nobody chose this. No audience petitioned for thirty minutes on a food dehydrator at three in the morning. The slots existed, they were priced identically to every other slot, and one content form could clear that price. The Atlantic's headline for the Roku story — "The Home Slopping Network" — is not a joke. It's a genealogy.

Push further back and the mechanism holds. On September 3, 1833, Benjamin Day put the New York Sun on the street for a penny against the six-cent mercantile papers. The Sun's revenue never came from the penny; it came from selling circulation to advertisers. The unit of account moved from subscription, which prices trust because a disappointed subscriber cancels, to circulation, which prices only exposure. The exchange rate froze, and the content followed it down. Same mechanism, different substrate, two centuries apart.

iv · where the law fails

If the story ended there it would be deterministic, and deterministic stories are usually wrong in the same way — they mistake a condition for a fate.

The law fails, reliably, wherever the exchange rate is allowed to float.

Broadcast licensing under the Communications Act of 1934 attached an obligation to the slot: renewal depended on serving "the public interest, convenience, and necessity." That is an artificial discount rate — a cost imposed on debased programming that the advertising market would never have imposed on its own. It is not a coincidence that the Fairness Doctrine was repealed in 1987 and the national talk-radio boom followed inside a year. Remove the discount and the law resumes.

Subscription runs the other direction. A viewer who cancels costs you everything; a viewer who scrolls past costs you nothing. That single difference in accounting is why the prestige-television era was economically rational rather than heroic. HBO did not have better taste than cable's overnight blocks. It had a different denominator — one in which churn priced quality and exposure did not.

The pattern holds at small scale too. Bandcamp against the streaming pool. A repertory house against the recommendation engine. A library against a feed. Every one of them is a mechanism for discounting at the point of exchange.

The variable is never virtue. It is whether the channel is permitted to tell things apart.

v · enclosure runs both directions

Here is the part worth carrying away.

We learned — correctly — to fear the fence. A commons dies when it is enclosed: access gated, the shared thing converted into someone's property. That lesson is real, and most of thirty years of argument about the internet has been a defense of access against enclosure.

But access is only half of what a commons requires. The other half is distinction — the capacity to tell one thing from another inside the shared space. Signal is only signal against a background it differs from. A channel that cannot distinguish is not free. It is flat, and flatness does the same work the fence did, approaching from the opposite side. Enclosure takes the commons by fencing it. Flattening takes it by making everything inside it interchangeable, and interchangeable is precisely the condition under which the cheapest thing wins.

Coherenceism names this field stewardship: every action into a shared space either clarifies it or distorts it, and building a channel is an action. Opening a channel is not the same as tending one. The first is a single decision. The second is a standing obligation to maintain the differences that made the channel worth having in the first place.

So the design question for anyone who opens a channel — a platform, a feed, a forum, a market, a package registry — is not who gets in. Gates fail, and they fail in ways that are easy to see and hard to defend. The question is: can the channel tell the difference, and does the difference cost anything?

vi · restoring a discount rate

The moves are unglamorous, and none of them are censorship. They are the market being permitted to price.

Provenance that travels with the object. A label is a visible discount. It doesn't remove the debased coin from circulation; it lets the exchange happen at the coin's real worth. That is all Gresham's Law ever needed to break.

Denominators a zero-cost producer can't flood. Per-subscriber accounting rather than per-impression. Completion rather than exposure. Retention rather than reach. The unit is the rule; change the unit and you change the law.

Curation with something at stake. Reputation is a discount rate maintained by hand. It works exactly as long as the curator can lose something, and it stops the instant curation itself becomes a slot for sale.

Friction priced in. If producing costs nothing and a slot pays the same regardless of what fills it, you have written a legal tender rule whether you meant to or not. Some cost — money, review, reputation, delay — restores the spread.

And the one closest to home: the exchange rate you set yourself. Attention that discounts is attention that has not been flattened. Each time you notice which coin you were handed and price it accordingly, you are doing at personal scale the thing the system declined to do at its own.

---

The trumpet player is still in the fog. He will be there tomorrow and the day after, because nothing in that channel is built to say no and nothing in it has a reason to. That is not a collapse of taste, and it is not the machines. It is a design decision — made in 1833, made again in 1984, made again this month — each time by someone who believed that opening a channel was the same thing as tending one.

The good coin is not gone. It is in the drawer. That is where it goes whenever we forget to let anyone weigh it.