[bho]

One idea i've really been thinking about is traffic analysis over time.

Where I live, traffic is highly variable. If I leave 5 minutes later one morning, there might be no traffic at all, but if I left another 3 minutes after that, I would be stuck in heavy traffic. If there was some way we could track the rate of traffic over time (say, plot out the mph between 7am-8am) every morning, I bet after a month we could figure out the ideal times to leave for work.

Google maps already uses crowdsourcing for its traffic data, so maybe that's the way to go. I would need a large sample size though. I researched a few cities and Houston, TX has sensors that use the toll RFIDs to determine traffic conditions. There are tons of other applications for this data, too. Think of shipment or delivery companies (UPS...), among others.

[danohuiginn]

Partner with somebody making the in-car direction/GPS boxes? They feed you the data on where there users are at each moment, you extrapolate route and time, and model traffic patterns.

[do those GPS boxes tell a central server where they are. can they? If it's "just" a privacy issue, the benefit of traffic monitoring would be enough to get people to sign away their location data]

Or just partner with a taxi or delivery firm: their dispatchers already do a fair amount of this monitoring manually, and they already track everything they can about what their drivers are doing. Plus B2B, so /much/ easier to charge serious fees.

[revorad]

From my few years of experience analysing London traffic data, it was clear that traffic follows an extremely predictable hourly pattern day after day, week after week, year after year. Except for specific holiday periods.

Look on the air quality section of the US EPA site - they might have some traffic data.