Sixty Years on Hold
For sixty-one years the US held a space-nuclear capability without ever using it. In April 2026, someone finally cashed it in.
On April 14, 2026, a deputy national security advisor stood at a podium in the Broadmoor Hotel in Colorado Springs and read the text of National Security Technology Memorandum 3. The memo set a 20-kilowatt mid-power floor for space nuclear reactors and committed a fifteen-person program at firm fixed price.
Everyone covered it as an announcement. It was closer to an exercise: the cashing-in of an option the United States had been holding, without exercising, for sixty-one years.
Invisible Infrastructure #5 is a pair of essays about what was being held while nothing appeared to be happening. The first follows the hold itself. The second widens the lens and asks the same question of the entire federal government, treated as an inventory.
What Webb Kept Alive
The Webb essay runs fifteen sections and ends in an interactive mission Explorer. It moves through SNAP-10A flying forty-three days in 1965, NERVA's twenty-three firings that never reached flight, SP-100's decade without hardware, JIMO's $463 million cancellation inside two years, and DRACO's planned phase-out in 2025.
None of those programs succeeded in the usual sense. That's the point. The thesis is that sixty years of program failure was itself a structural accomplishment, sustained by a distributed apparatus of senators, civil servants, and engineers who kept the knowledge viable across political cycles that kept yanking the money.
The Explorer at the end lets you build a mission profile against the NSTM-3 tier baseline and see which historical precedent your build resembles. The argument isn't that the memo was the point. The hold was the point, and the memo only worked because the hold held.
The hold was the point, and the memo only worked because the hold held.
What the State Keeps
The companion essay turns the same lens on the federal state as a whole. It proposes treating the government as a portfolio of forty-nine unexercised capabilities, each with a strike condition, a premium, a decay rate, a switchability score, and a reconstitution difficulty. Then it runs a thought experiment: if Amazon bought the warehouse and applied its acquisition playbook, what gets kept, what gets cut, what gets monetized?
The essay embeds two interactives. A Playbook Simulator walks through six acquisition moves and shows how each one degrades the inventory along a specific axis. A Portfolio Explorer sorts the forty-nine options by reconstitution and switchability and reports how the highest-scoring options fare under acquisition logic.
The pattern that emerges is the argument. The options that matter most are disproportionately the ones most easily lost, because the very qualities that make them matter, diffuse beneficiaries, preventive outputs, retrospective value, are the same qualities that make them illegible to an optimizer.
Why Two Essays
The pieces are written to be read in either order. The Webb piece is a dramatized case; the State Keeps piece is the framework that case is an instance of. Webb shows what a sixty-year hold looks like when it works. State Keeps shows why that kind of hold is rare, costly, politically disorganized, and structurally mispriced by any evaluator who only scores observable output.
Both are interactive. Both run on the same design system: cream paper ground, ink borders, red and yellow accents, Archivo Black display, Fraunces serif body. And both argue, in different registers, that the work of holding something is invisible by construction. The memo gets the podium. The apparatus that made the podium possible does not.