Earlier quoted context omitted.
There's only about 8k people who are 1 in 1k crazy and 1 in 1k capable, assuming independent traits. If you restrict that to even 1 in 20 violent, then you're talking 400 people world wide. Factor out kids, elderly, etc and you're talking 200 really scary motherfuckers. Figure 75% are employed by governments, and you're talking 50 rogue badasses -- which seems vaguely in line with the number of really dangerous terro…
Can you explain how you came up with these numbers? Intuitively I have always thought they were low, due to the small incidence of terrorist attacks, but I am curious to where you come up with your base numbers.
But they're reasonable bounds -- the number of terrorists are very low, we know this for a fact. So we should suspect that the contributing traits are relatively rare. And if we want to talk about people who really scare us, then it should be people who are real outliers.
An NFL player is a one in a few hundred thousand, and an NFL star is around one in a million. So if we're looking for the Tom Brady of terrorists, it would make sense that we don't want a fraction substantially higher than the number of Tom Brady's in football players.
So I cut off a segment of the population that would have the traits required and estimated how large it was -- 99.9th percentile capable, 99.9th percentile crazy, 95th percentile violent. (Intuitively -- if the smartest and craziest kid in your high school were the same kid (1 in 1000 each trait), then you picked the most violent in 20 of those kids.)
This turns out to be ~400 people, which seems ballpark right. I'd certainly expect the answer to be in the 120-1,200 person range.
As long as your estimates aren't too far off and you aren't systemically wrong (eg, always rounding down), estimating is usually a reasonably accurate method to get a feel for the scale of something.