The marginal cost to add riders on transit is zero only to the point

where you don't have to add additional equipment and/or personnel or install changes to your system in order to accommodate more people.

A 60 foot articulated bus costs $650,000. A subway car costs $4 million.

When you add additional runs of either buses or trains, this requires more equipment and more personnel. For subway trains it requires more cars to make up 8 car trains (the typical car length is six cars) and if you run trains more frequently, it requires additional cars.

And to increase throughput, additional changes, costing money, sometimes a lot of money, especially because retrofitting a system not designed optimally for high usage, is extremely expensive.

Adding service, as WMATA did for the subway and bus services for the Inauguration, cost a lot more money even though they didn't buy more equipment for this particular event. They did run more vehicles and pay a lot of overtime.

------------
Another point that is overlooked is why you charge rush hour or peak fares. People including John Catoe, the general manager of WMATA, says that this is a hardship for "frequent customers," implying that frequent customers should be rewarded. (Hey I am all for lower fares.)

The thing is that a transit system has to be built for maximum capacity, not average capacity, therefore it costs more to build the system initially and to maintain it. So why not charge more to the people who end up generating higher costs?

And speaking of the economic concept of "sunk costs," the reason that the NYC Transit system charges one price, not variable fares unlike the WMATA system in the DC region isn't because NYC wants to promote transit use more (although they might say that). It's because the cost (and time required) to change all of the fare machines and the turnstiles throughout their 468 station network is conservatively estimated to be hundreds of millions of dollars (if not $1 billion). So they won't change as it is too expensive. On the other hand, that makes it easier for them to encourage transit use by charging special rates for purchasers of transit passes.

And this is an technology point, not a economics point. Doesn't anyone who promotes the idea of NYC Transit-like touchscreen fare machines ever notice how long they take to use, far more time is required compared to DC's more analog-style machines. Imagine how much harder it would have been to move farecard buyers along over the Inauguration period...