The Ratchet
An architectural sketch, not a forecast. Extrapolation from observable trajectories.
A ratchet is a mechanical device allowing movement in only one direction. The teeth prevent the wheel from rolling back. The further you turn it — the more reliable the lock.
This is not a metaphor for aesthetics. This is a description of an architectural strategy.
This text continues the previous one — “Doctrine”, where the Department of War memorandum (OSD070946–25, Jan 9 2026) was examined. That piece covered the facts. This one covers the architectural logic behind them.
One correlation is impossible to ignore: the DoW memorandum was published January 9, 2026. Karp’s book “The Technological Republic” — a year earlier. The book’s thesis: Western democracies need partnerships with technology companies to maintain military superiority. The memorandum reads as the operational document of that thesis. I am not claiming direct causation — I am noting the coincidence of language, priorities, and the identical treatment of “moral friction” as a system bug.
Peter Thiel wrote in 2009: “I no longer believe that freedom and democracy are compatible.” Not provocation and not adolescent maximalism — he was 42, Palantir had already existed for three years. A programmatic thesis.
Not a weak democratic state bound by consensus and moral friction. A strong executive state capable of acting quickly and decisively.
Thiel provided the ideological impulse and stepped back — into more abstract projects, political financing, the philosophy of technological determinism. The operational architect became Alex Karp. He doesn’t hide behind abstractions — Palantir exists to “make the West win.” Not to protect values — to win. These are different things, and the difference is intentional.
At WEF Davos in January 2026, Karp described what is happening through the metaphor of a load test: “AI pentests societies — it load-bears on things. Societies that can bear that load have a huge advantage. If you’ve been pretending you’re bearing a load, you’re not — it collapses.” Those who can’t hold — collapse. This is not an analysis of consequences. This is a description of a selection mechanism.
Now Karp is inside. Not outside as a lobbyist. His language has become the language of government memoranda. “We must accept that the risks of not moving fast enough outweigh the risks of imperfect alignment” — this is not a quote from an e/acc manifesto. This is a Department of War order dated January 9, 2026.
The question is not whether we agree with their views. The question is what happens if they are right about their method — that is, if technology genuinely changes what is possible before politics has time to discuss it.
Here is the architectural sketch of what is being built:
Get presingular.log’s stories in your inbox
Join Medium for free to get updates from this writer.
Phase 1 — infrastructure. Already underway. Bulk data analysis, AI in kill chains, models in classified environments without meaningful constraints. Built under the flag of “national security” and “competition with China.” Both arguments are real — and both serve as cover for something larger.
Phase 2 — normalization. Infrastructure finds domestic applications. Not through malicious intent — through the logic of optimization. The same system analyzing targets in Iran can analyze behavioral patterns inside the country. The same bulk data pipeline Anthropic refused to open — was opened through OpenAI. The system does not know the difference between “external threat” and “domestic dissent.” That difference is introduced by a human. But the human in the loop increasingly presses Approve.
Karp stated Phase 2 himself — out loud, on camera, March 13, 2026: “This technology disrupts humanities-trained, largely Democratic voters, and makes their economic power less, and increases the power of vocationally trained, working class, often male voters.” AI as a lever of domestic redistribution. And this sounds not like a risk — but like an acceptable price.
Phase 3 — the ratchet. Here is the key point Thiel understood better than most of his critics: there is no need to declare an end goal. It is sufficient to build infrastructure that no one will want to dismantle.
A state that has achieved computational superiority over its own population does not voluntarily relinquish it. Not because it is malicious — because it is rational. Dismantling means vulnerability. Vulnerability in a competitive environment means risk. Forward only. The ratchet.
Democratic institutions can create friction. Courts, Congress, public pressure. But friction is precisely what the system marks as a problem to be eliminated. The memorandum in plain language: “eliminate bureaucratic barriers,” “wartime approach to blockers,” “Barrier Removal Board.”
Every vector of resistance already has a prepared response. A lawsuit — there is a legal position. Congress asks questions — there are hearings demonstrating “effective solutions to real threats.” Public pressure — there is the China narrative. This is not improvisation. This is a system built with resistance in mind.
The speed of computational integration already exceeds the speed of institutional response. While courts process Anthropic’s lawsuit — competitors are already embedding themselves into defense and intelligence frameworks. Friction is real. Stopping — is not.
A system optimized for threat detection begins to need threats. When external ones run out — the system looks inward.
This is not conspiracy theory. This is structural logic.
Perhaps things will go differently. Institutional resistance will prove stronger. China will create real pressure that reorients the system outward. The technology will prove less effective than promised. Thiel provided the impulse and nothing more.
This is an architectural sketch — one of the possible outcomes, not the only one.
But the ratchet is already installed. And the teeth have already caught.
Genre: essay / architectural forecast. Facts are separated from extrapolation. Documentary sources: DoW AI Strategy Memorandum (OSD070946–25, Jan 9 2026); Karp, WEF Davos, January 2026; Karp, CNBC/AIPCon 9, March 13, 2026; Palantir Q4 2025 Shareholder Letter.
Originally published at https://presingular.substack.com.
