Improving public transportation's reliabilty

I ride a WMATA bus every morning to commute to work - and every morning I get to work at a different time. There are posted bus signs at each bus stop but they can hardly keep their schedule because of the unpredictability of traffic (there are no dedicated bus lanes in DC) as well as the unpredictability of the bus driver.

This means that whenever I'm out in a different part of the city, I won't wait to catch a bus. The unpredictability just isn't worth it especially considering there's a chance the bus might not come at all. Although the bus maps at each bus stop tell me where the bus is going, I have no idea if the bus is actually going to come or how long it will take as the posted schedule is meaningless. So I'll pay a premium to just hop in a cab.

I suspect I'm not the only one who has this problem. And each time this happens WMATA loses revenue. Because the incremental cost of an additional person on that bus is essentially $0, that's lost profits.

Interestingly there are people who have the information I need to let me know a bus is coming - it's the people riding that bus and the bus driver himself. If I could just call/text/e-mail/(or look them up on Google latitudes) , they could easily let me know how far away they are, run that through an algorithm, and bam I know in how much time that bus is coming.

Such a system could easily be automated to avoid annoying my friends.
First there are (at least) two ways to gather the data:

1. Simply install a GPS in each bus with wireless to communicate the bus's position to a central server (essentially the I-phone w/ Google Latitudes)
2. Plug in a bare bones cell phone into the machine which announces the stop name (these busses have a machine that announces stop names, which is activated by the bus driver by pressing a button). The cell phone would simply text the main database with a code for each stop.
   

With the information gathered, the main database can then run the positions along with current traffic reports and it's own algorithms (which can be based on the data collected). Potential customers, like us, can then simply text/ call/e-mail etc into an automated service with our stop number or cross street and receive the estimated wait time. You could even provide an estimated time of arrival.

The data collection itself would also provide WMATA with important metrics which they could use it to track bottle-necks in real time as well as the driver's ability to keep to the schedule.

This kind of small investment = better data = better service = more $