The comparison is against a sort of system optimum so you don't really need to account for such historical benefits. This isn't a bank account!
At no point in my coverage of the externalised costs have I indicated that 'no car use' is the answer. However it is clear from every metric we can measure that we are considerably beyond any peak benefit to society of the motor car and are busy plummeting down the other side. Obviously this varies by area. The IFS and others have done studies on this previously and concluded that some driving is actually slight net contributors to the public purse (generally rural and quiet times of day) whilst as mentioned, driving in a city at peak times is in the £s/km range of subsidy, hence road user charging is probably the best solution here.
Without that though, a direct subsidy/cut of petrol/diesel prices is not a great way to spend public money.
Even urban transit systems are best when supported by government investment. As Kahn keeps pointing out in response to the ongoing wrangles with DfT over TfL's funding, every other city in the world comparable to London has a transit system receiving much more public subsidy, including New York. We can't even manage what the usually more economically right wing USA is willing to fund. Our public transport is rubbish across the country.