When can a military technology remain secret?

Have closely guarded military technology secrets ever worked?

Yes, when a sufficiently large and committed institution preserves them, though every preservation method introduces vulnerabilities.

Written copies can be stolen, while trained specialists can defect, so secrecy requires both redundancy and loyalty. Broad institutional buy-in can make democracies surprisingly effective at protecting secrets. Autocracies may sacrifice technical competence by selecting politically dependable personnel instead of the best-qualified specialists. Successful secrecy is an institutional design problem: the system must preserve enough expertise without creating more vulnerabilities than it can manage.