(And, by the way, I'm not claiming anyone was advocating a ban on writing down numbers in notepads on subways; the analogy to photography just seemed too good. And I believe in the past Schneier's answer to the question "how to stop it?" has been to hire curious, well-trained people who know what to look for, instead of instituting blanket bans and searches that require wasting a lot of manpower on showy but easily circumvented measures.)

"It is my belief that the London transport bombers actually rode the subways one year previously during the summer to count how many people were riding in different trains at different times.

"It is my belief because my mother visiting London riding the subways then saw a 20 or so year old man of Middle Eastern origin counting people and writing numbers on a small notepad."

That sounds extremely far-fetched to me.

But even if that were true, the solution in this case wouldn't be a ban against people writing numbers on notepads in the subway, because a) security people who could otherwise be doing something actually useful would be wasting their time enforcing the ban even in cases where the behavior was clearly innocuous, and b) even if the bombers did do that it's unlikely that it was really an essential step, so it's hard to believe a ban would have upset the plot. (In fact, if writing down numbers really were some kind of terrorist tell-tale, then a ban might lose you a useful clue.)

So it seems a better use of security people's time to let them take that behavior as one among many (small) clues, and have them follow up (politely and discretely) if they really think it's useful.

Seems to me that the hardest part of estimating this kind of risk isn't finding the accident data, it's finding the exposure data.

If intersection A has twice the accidents of intersection B, it may appear to be more dangerous, until you learn it also has ten times the traffic....