The Press

Part of the pageleft series.

The scribes did not lose a war. They just stopped being hired. For a thousand years, copying text by hand was the bottleneck and the business model. Then Gutenberg made the bottleneck irrelevant, and the monasteries emptied. Their skill became expensive relative to the alternative, and the alternative got cheaper each year.

But the printing press created its own intermediaries. Publishers, distributors, bookstores, libraries: a new industry built on the transaction cost of binding, shipping, and shelving physical pages. The book itself was a transaction cost on information. The internet compressed that layer. Compilable prose compresses what’s left: the interpretation layer.

Vibelogging: blog posts are build instructions. Canon: copyleft propagates through the compilation chain. Open Prose: the category has a name. This post is about the press. The forme only holds if the chase is open.

Page at a time

PageLeft grows linearly. One person writes. A few more join. Every post adds mass. Every link is a quoin. Every derivative tests whether the loop works.

From the outside, it looks unserious. Weirdly formal blog posts. Cross-links. Manifestos. Repo links. A person writing as if essays were source files and footnotes were dependency management. People glance at it, half-smile, and file it under personal internet hobby. They see a niche writing habit. It’s not even funded. They do not see an insurgency.

That is how these things look before they tip.

Software engineers adopt first because they have the clearest pain. Code rots. Specs drift. README files lie. The first serious vibelogs become reference implementations for prose itself: what precision is enough; what structure survives compilation; what review standards separate source from slop. The first authors impress the grammar for everyone else. Convention growth.

A linear system with accumulating network effects builds pressure. One constraint ruptures and the slope inflects.

Cost inversion

Ideas move at human speed. Compilation moves at machine speed. One person with a coding agent produces copyleft implementations far faster than new ideas can be learned. The canon fills faster than the frontier advances.

Every public implementation becomes prior art, narrowing the room for proprietary exclusivity and giving the next agent more material to compile from. Copyleft gets cheaper because the base grows.

But copyright gets more expensive. The more canon is copied left, the harder it is for holdouts to sort through an ever expanding provenance of prior art, to defend an ever thinner claim to novelty. And they hire humans to rebuild for thousands what the canon already piled for pennies.

The holdouts hold till holding costs exceed switching costs: epsilon-Nash, the same equilibrium that keeps credit card fees at 2%. But as the copyleft platen drops, so does the cost for switching. So the equilibrium flips.

Both cost curves accelerate. Each impression widens the spread.

Talent squeeze

Labor follows price.

Early converts trade code translation skills for prose composition. If prose is the durable artifact and code is the compiled output, the scarce skill shifts upstream. The valuable engineer is one who can see clearly, structure precisely, and write prose an agent can compile without guessing. Set the type or the type sets you.

The open side becomes more productive; the closed side’s cases empty. Crossing the chasm is pragmatic, not idiomatic. Harder provenance. Better workflow. Faster iteration. Stronger reasoning.

On the closed side, the parasitic industry lingers. Consulting firms, integration vendors, proprietary tooling: an ecosystem feeding on the friction of opacity. The fog becomes the business model. Its beneficiaries resist canon for the same reason adtech resists transparent auctions: they milk revenue from friction. The press squeezes at the margins of rent-seekers.

Holdouts do not stay noble. They become extortive.

Contamination

The legal risk turns operational.

Engineers at proprietary companies will read copyleft prose because it’s the best documentation available. PageLeft makes it inevitable. If the canon is where the clearest explanations live: then engineers will traverse its corpus; routers will inscribe its entrails; infrastructure will register the corporate proof.

If compilable prose is source material, exposure is not without breach. The company does not get to say “we only read the prose, not the code” if the prose already compiled to code. That is the whole argument of Canon. The question is whether the shipped implementation derives from the prose’s copy-protected selection and arrangement, not whether someone copied a file line by line.

The question gets sharper as the prose writes better. Traffic from proprietary companies to copyleft canon is detectable. Why are their copyright engineers reading copyleft ideas at work? Once exposure is established, the burden shifts. Ink is on their dabber.

You can firewall engineers from the canon. Some firms will. That strategy fails for the same reason information quarantines always have. It imposes an internal productivity tax to avoid an external legal risk. Compliance by ignorance, paid with velocity. They are in a bind.

The press makes its impression.

Coasean collapse

Every economic argument from Transparency Is Irreversible applies here. But there is a second turn.

The software that copyleft prose produces is itself a tool. Those tools lower transaction costs in every adjacent domain. Each tool removes opacity. Each removal is irreversible. The newly transparent domain becomes a market where the same economics unfold yet again.

Aaron Swartz understood this before anyone. In 2008 he said it plainly: the world’s scientific heritage is being locked up by a handful of private corporations, and sharing is not immoral. It is a moral imperative. Academic publishing charges institutions to access publicly funded research, then locks the papers behind paywalls. He tried to free them and was destroyed for it. The gatekeeping survived him, but it will not survive the canon.

Academia falls right after software because academic literature has every property that makes software vulnerable to the press: it is precise, it is compilable, it is cumulative, and it is already written. Whether it’s published or not, the papers exist. The synthesis is extractable, and the tools to apply them are buildable. Copyleft indexing makes them searchable and forkable.

Swartz wrote a manifesto; so did we. His was a plea; ours has a compiler. The journal paywall is the same fog, and the same economics lift it.

This pattern is not new. Coase (1937) predicted: when transaction costs change, market structures change with them. Wallis and North (1986) measured it: by 1970, nearly half of U.S. GNP was the transaction sector. It already happened to stock trading (commissions went to zero), travel agents (fare comparison became software), classified ads (Craigslist killed the newspaper), and insurance pricing (comparison sites compressed broker margins). Every time transparency progressed, intermediaries compressed. These are next:

Every one of these is an industry built on a foggy moat. Gutenberg emptied the monasteries. The winds will empty the intermediaries.

The press does not print just once. Each print is a Coasean collapse, the same cliché struck in a new domain. Copyleft prose collapses the cost of building software. That software collapses the cost of gatekeeping in domains that have nothing to do with code. The press is on all information gatekeeping we know.

Sustained acceleration

More than velocity. Acceleration. Sustained acceleration.

Each strike fastens the loop. Canon pares the cost of its own scion. Prior art burdens closure. Converts deepen the talent pool on one side and teeter the other. Corporate exposes contamination risk. Tools collapse transaction costs for the next issue to be released.

None of this reverses. Canon does not lose utility as it expands. Prior art does not hide. Talent does not deflate for holdouts. Transparency does not become optional after the fog clears.

The press is not one explosive moment. It ratchets. Once it fastens, none can opt out.

Written with Claude Opus 4.6 via Claude Code. Claude drafted prose; these are the words I chose.