Begin at the moment of triumph, because that is when the procedure starts. Stop building the machine while it is still winning: the decision to end Saturn V production was taken in 1968, before Apollo 11 had even flown, so the shutdown was already signed while the triumph was still on the pad. Cancel the last missions while their rockets exist; the final three Apollo landings were struck from the schedule in 1970, and hardware built for them went to museums in Florida, Texas and Alabama, where the great stages lie on their sides to this day, the most expensive lawn ornaments ever manufactured. Note carefully that none of this was folly. A war was devouring the budget, the public had stopped watching by the third landing, and a government that kept producing Moon rockets nobody wanted to fly would have been accused, fairly, of building monuments to itself. Every step was defensible on the day it was taken. That is what makes this a procedure rather than a blunder, and what makes it repeatable.
Next, release the people, gradually, so that no single day ever looks like the day the knowledge left. At its peak the Apollo effort employed some four hundred thousand people across the agency and its contractors, and within a few years of the last landing the great majority had been laid off, reassigned or retired: welders who knew the feel of a sound seam on an engine chamber, seamstresses who had sewn the pressure suits by hand, inspectors who knew which tolerances were sacred and which carried quiet exceptions, and why. Each of them held a portion of the machine that existed nowhere else, the thousand small accommodations between drawing and metal that no drawing records. The severance paperwork accounted for their salaries. It had no line for what they carried out of the building.
Keep every document, and this instruction is sincere, because the documents are what will let you believe the ability is merely in storage. The story that the Saturn V blueprints were lost is a myth, and the myth gets the lesson backwards. The drawings survive, complete, on microfilm in Huntsville, and they were never the problem. In 2013 a team of young engineers at Marshall Space Flight Center took the gas generator from an F-1, the engine that lifted the Saturn V, out of museum storage, mounted it on a test stand and fired it under full instrumentation, because the drawings told them what the old builders had done while staying silent about why, and the only way left to recover the why was to interrogate the hardware itself, like archaeologists excavating their own grandparents' competence with the complete plans in their hands the whole time. They learned a great deal, those engineers, and every lesson was a measurement of the distance between a record and a practice.
Then wait half a century, decide you want the thing back, and bring money. The return has cost, by the inspector general's accounting, some ninety-three billion dollars through last year, at more than four billion per launch of the new rocket, sums nobody would have dared request as maintenance in 1975 and that were paid, for want of any alternative, as reconstruction fifty years on. And the bill is not denominated only in dollars. Reid Wiseman, Victor Glover, Christina Koch and Jeremy Hansen, the four now out beyond the far side, were all born after the last Saturn flew; the men who once walked on the Moon are in their nineties where they are still with us at all; and between those two generations there was no handover, only archives. A handover moves the why along with the what. An archive moves the what alone, and the why has to be bought back later at whatever price the decade demands.
I have written the procedure out because it is not aerospace history; it is standard operating practice in rooms with no rockets anywhere near them. Retire the platform that works, because the roadmap points elsewhere. Archive the documentation thoroughly, so the decision feels reversible. Accept the resignations one at a time across a quiet year, each departure individually unremarkable. Some years later, commission the replacement, and discover that the archive answers every question except the ones that matter, that the people who could have answered those are gone past tracing, and that the organisation is buying back, at compound interest, a thing it used to know in its hands. The defence is dull and has to be purchased in advance: decide which of your competences are load-bearing, and treat the practice of them, not the record of them, as the asset under protection. The four astronauts crossed into the Moon's shadow and came round the other side, and their flight is two documents in one, a triumph and a receipt. I hope the receipt gets framed somewhere prominent, because every organisation I have ever seen is, at this moment, somewhere in the middle of writing its own.