Predictions

Yesterday's post include the prediction that the founders of the Dartmouth Artificial Intelligence Conference were GI Generation - that is, the Civic / Hero generation that consisted of young adults during the Great Depression and World War II. (Which means they were born between 1901 and 1925.)

Out of the four founders, two were GI, two were Silent (children during the Great Depression and World War II) . Is that a successful prediction, or not?

First, is two enough to say it is successful? One could say that a split down the middle is insufficient, that the saying "these guys are GI" should mean "mostly GI, so 50% plus 1." In which case, nope, unsuccessful prediction, no way around it.

Next - well, it's been a while since statistics courses, so it's quite possible that the following will miss some important nuances. But:

If one can accept that two GIs fulfills the requirements of the prediction the question becomes if it's significant, in a statistical sense - that is, could it have happened due to chance alone? It's unlikely to flip a coin and get 10 heads in a row, but if you have one thousand people flipping you will almost certainly find one person who can do it.

Considering the people who chose to be founders as a random event,  in 1956 we could have had one of each then-extant adult generations: Silent, GI, Lost, Missionary. (The oldest Boomers would have been 13.), That it was instead only members of two would be statistically significant, indicating that it was a successful prediction, inasmuch as the predicted generation was included.

Old dogs like the Lost and Missionary generations aren't likely to have been supporting a radical new concept, though. Maybe some of those old Prophets would have wanted a final grand crusade, but  it's hard to imagine that for Reactives in a post-Crisis era. Either way, if we consider that it should have been composed of three generations, and it was only two, the prediction is still unlikely to have been correct on chance alone.

That is, it would have been correct about 40% of the time, if having at least 2 of the four is sufficient. Significant, but not shockingly so. Next time, it may be worthwhile to decide more completely -  ahead of time - what the success criteria are of for such challenges.