Good point. The OP is OCD about connection faults, not about failures in the cable body. So for that measure of concern cord-years is good enough.
A further refinement in yet another direction is cord-connector-years since many such cords have a three-female connector on the distal end. A cord with 3 female ends has 4 potential connection failure points, not just two that a single-female cord does.
One can further refine by cord-connector-in-use-years, since a cord end with nothing plugged into it is far less likely to suffer connector failure leading to a fire. By that measure my lifetime experience is vastly reduced as the vast majority of my connector ends are unused.
A full risk model would include all these factors as well as your suggestion of cord length. As watchwolf49 has pointed out, his cords suffer risk proportional to length (and bystander stupidity). For indoor extension cords deployed behind furniture that particular risk is very near zero. But accumulating dust bunnies and enclosed spaces impeding cooling, and perhaps vermin chewing unnoticed are all risks we ought to account for to arrive at the idealized perfect risk model.
Despite all the silliness above, it’s interesting that there is a real-world direct analog to this silliness with real-world consequences: What is the relevant measure of risk for commercial air travel?
Is it per aircraft mile, per aircraft flight, per aircraft time, per passenger mile, per passenger flight, or per passenger time? Are different risk measures applicable to different constituencies? How & why?
To a first approximation, all the risk is in takeoffs and landings. In other words, at the “connector” between flight and not-flight. Which are, by definition, one each per flight cycle, with enroute time or distance or passenger count immaterial. Much like the risk of extension cord failure is mostly at the interface ends, not so much in the middle.
OTOH, people don’t go for flights as entertainment. They fly X hundred miles from A to B in lieu of driving or taking a train the equivalent distance. So *relative *risk ought to have some distance or time factor applied.
etc.
