(courtesy xkcd)
Common misperception: replication isn't about the mere act of repeating things. The point of replication is to make a convincing case that a systematic rule exists, such that we can be confident that a given set of actions will produce a given set of results. How many times do you need to stick your finger in a power outlet to convince yourself that it's a reliable source of electric shocks? Twice? Maybe three times? More is just masochism...
When you experience something, you've experienced it; there is no question about that. The problem lies in two questions:
- Was what you experienced a one-off fluke or aberration, or something you should treat as a real phenomenon?
- Can you convince other people that you actually experienced it, or are you going to sound like a loon talking about it?
In order to answer either question, you need to be able to repeat the experience so that you or others can see it again. If you've seen a ghost, you've seen a ghost. You might be imagining things, or outright hallucinating, but you've still had the experience. Can you set up the conditions to produce the ghost again, so that you can see it, and others can see it? If you can, then you've shown that the experience is replicable, and that by itself is enough to add credibility to your claim.
In the academic world people can get obsessive about replication. They will test something dozens of times, in different variations and conditions, in the hopes that they can 'break' a theory. That's good and useful in that environment, because breaking a theory in academia is an important part of constructing new and better theories. But in the real, practical world we don't need to repeat something over and over and over to convince ourselves that a rule is in play. Common sense dictates otherwise.