The other ingredient is that you don’t have a pre-defined set of “cool things”. Let’s say that if something has a one-in-a-thousand chance of happening, it’s pretty cool when it happens. But there could be thousands of such things, even if we choose only the “meaningful” ones (the ones which can leave a psychological impression of “something special happened” on us). So even if each one of them is individually unlikely, the “set of all unlikely things” is actually pretty likely.
But despite “something unlikely” happening rather often, if you happen to observe one unlikely thing and decide to reproduce this specific one, it does not work. The “set of all unlikely things” is likely, but the one specific thing you try to reproduce often remains unlikely despite your trying.
Confirmation bias is one ingredient.
The other ingredient is that you don’t have a pre-defined set of “cool things”. Let’s say that if something has a one-in-a-thousand chance of happening, it’s pretty cool when it happens. But there could be thousands of such things, even if we choose only the “meaningful” ones (the ones which can leave a psychological impression of “something special happened” on us). So even if each one of them is individually unlikely, the “set of all unlikely things” is actually pretty likely.
But despite “something unlikely” happening rather often, if you happen to observe one unlikely thing and decide to reproduce this specific one, it does not work. The “set of all unlikely things” is likely, but the one specific thing you try to reproduce often remains unlikely despite your trying.