Bletchley does seem to be the best modern example of a major secret kept a long time by a lot of people. I’ll have to try to think of some others.
It seems to me Lost and The Prisoner fell apart for similar reasons. The responsible parties seem to contract some form of creator guilt and veer off allegorical. In Lost it seemed the writers did not want to let bad things happen to any characters. In The Prisoner it seemed there really weren’t characters to do anything or have anything happen to and the archetypes tried to throw off their cloaks and dance around but got tangled a bit.
With RPG mysteries I’ve had to make certain that a satisfactory conclusion for the party was possible and visible and that’s been my goal with key clues. I fee like the key clue should make the goals visible to the party (this is the murderer, the island possesses this mysterious quality, this is what the villagers want from number 6). If I can’t articulate that in a form that’s satisfying for me to run the game I need a different mystery. The rest of the clues may lead to putting the murderer behind bars but I want the players to choose between acting directly on a key clue or acting to find more clues.
I particularly liked Roger’s story of the group that realized they had evidence manufacturing skills. That kind of freedom to respond to a key clue seems a goal to strive for in a mystery game. The worst results for me have occurred if the players were confused what the mystery was or lacked the confidence to act on the mystery and kept asking me as GM “what are we allowed to do next.”