The automated metro is sweeping these old equations aside Greg. There are
no driver costs, therefore you don't have to factor driver productivity.
You can move huge numbers of commuters with no drivers at all and move them
very quickly while stopping at every stop. Although underground metros are
proving more flexible in terms of reach than old suburban surface lines,
there is of course a cost limit to how much dense coverage you can achieve,
so you certainly still need buses as feeders and infill. What is being
squeezed out in this equation is trams and what is encouraging them to be
squeezed out is the poor cost benefit (high costs, low productivity) of all
those poorly procured and run new light rail systems. There are a lot of
places they will work well, notably smaller cities, but in super cities
like Sydney they're not the right mode. They cost too much in relation to
the limited extra capacity they provide.

Legacy systems that are already in place are generally fine, especially if
the city has already developed around them and they are well-run as in
Central Europe. They have managed to find a good balance with metro systems.

In Sydney, with hindsight (if it had been known that the CSELR project
would turn out as it has), it would have been better to proceed with the
original 1990s plan to create a CBD circulator tram shuttle (CQ-Central)
linked to the IWLR and build a metro out to the SE including linking
Central, Moore Park and UNSW. Too late now and, in any case the CSELR is
valuable as the CBD shuttle (its most successful component), the
Central-Moore Park shuttle (even though short on capacity) and the
Central-UNSW shuttle. As a suburban commuter service it's useless and the
further it goes the worse it gets, so it shouldn't be extended beyond where
it is, though they seem to be thinking that it has to be extended to
Maroubra Junction to link up with the future SE metro.

Tony P