I think that you're trying to force metro travel patterns into a normalized format for comparison with other modes that doesn't match actual usage patterns. If

I think that you're trying to force metro travel patterns into a normalized format for comparison with other modes that doesn't match actual usage patterns. If we have to do that, then for my original post, the relevant factor is passenger minutes/hours travelled because the driver's cost scales with time not distance. My cost is not proportional to either, though. If I go to my nearest station and go the next tube stop or if I go five tube stops the fare is the same. If anything it's inversely proportional e.g. the longer I spend on the train, the less I contribute to the driver's hourly wage. The amount of time I spend and the amount of value I derive is entirely irrelevant to any of this. I'm not especially interested in any of this beyond my original point that the cost of the driver is trivial on a per passenger basis, my back of the envelope calculation gives the same result (doesn't matter) even if you 3x or 0.33x the outcome.