Dave Alpert has some interesting thoughts on improving bus service. I'm a big advocate of route consolidation in which a city like Washington that has a lot of bus routes would pare them down to a smaller number of routes that are more frequently serviced. The benefits of frequent service for transit are, I think, hard to overstate. If you're trying to get into a bus that only runs once every 30 minutes then if you want to get anywhere on time you need to be paranoid about not missing the bus and usually wind up showing up too early and wasting time. What's more, if a bus happens to be a minute or two late, panic sets in that you've missed it or that some incident has taken the bus out of service and maybe you need to scurry off and find another way of getting where you're going.
That kind of stress and hassle winds up making the bus a "must avoid" transit method, and helps perpetuate the bad bus branding where you constantly meet carless young professionals who've lived in DC for years and know almost nothing about bus routes. Then low ridership means you can't justify frequent service, leading to inconvenience and fewer riders. Of course I'm a fanatic who'd happily say we should spend the money to just increase frequency on all our routes, but working within realistic budget constraints it's better to pare the number of parallel routes down somewhat and increase service frequency.
Meanwhile note that this isn't just an urbanism issue or an environmental issue (though better buses will make our cities more livable and sustainable), it's also an important equity topic. Buses play a much larger role in the transportation bundle purchased by poor people, so better service can dramatically improve quality of life for the working poor, make it much easier for people to find and keep jobs, etc.
Photo by Flickr user intangible used under a Creative Commons license