I accept the technical correction. But for P close enough to zero, the probability remains nearly linear up to large T. So while you are technically correct I do not think you have refuted my statement as a very close approximation for a typical session lasting an hour or several. Given that, and given the relative difficulty of thinking about an exponential function, I prefer my formula for conveying the essential point in a comprehensible way.
It would be better to make the use of approximation explicit, and to specify the domain in which the approximation is a good one, like “don’t forget probability of accident per second is multiplied by time to get approximate total probability, given small total probability”.
No no, I am saying I erred outright. I simply used a heuristic and forgot it was a heuristic. You corrected me. That’s why I had not identified it as an approximation initially. But that said, I find the correction to make the reader’s task difficult, and my heuristic retains approximate validity, so I pointed that out in my reply.
I accept the technical correction. But for P close enough to zero, the probability remains nearly linear up to large T. So while you are technically correct I do not think you have refuted my statement as a very close approximation for a typical session lasting an hour or several. Given that, and given the relative difficulty of thinking about an exponential function, I prefer my formula for conveying the essential point in a comprehensible way.
It would be better to make the use of approximation explicit, and to specify the domain in which the approximation is a good one, like “don’t forget probability of accident per second is multiplied by time to get approximate total probability, given small total probability”.
No no, I am saying I erred outright. I simply used a heuristic and forgot it was a heuristic. You corrected me. That’s why I had not identified it as an approximation initially. But that said, I find the correction to make the reader’s task difficult, and my heuristic retains approximate validity, so I pointed that out in my reply.