Aktalita founder Tom Kessler reports:

Operation is fully seamless between Web and lowest end cellphone with Java 2 mobile edition capability. (We work fine down to Nokia 3220 - $50 cellphone).

Users sign up at the site, provide some personal and vehicle info, declare their landmarks or reference public landmarks, then trace their exact route(s). If they don't trace a route, we assume a route from a web services query.

The user then downloads his web profile to the cell phone midlet. A driver declares his trip as a precise space-time polyline object with detour distance and time tolerances. A passenger requests origin and destination also with detour distances. We consider a match to be a spatial intersect between the drivers polyline "sausage" and the passenger's discs and time window overlap at the passenger pickup point. We also store all the public transit routes in an area and do compound matching... So a driver from an airport to home that crosses, say a light train line going to the passenger's destination, will compound match with a passenger's request.

On the social side, we support an n-level social network and three types of groups: Regular, Qualified, and Geographic. Basically the idea is to make the site one big social network dynamic rideshare experimental toolkit, where others with similar interests can customize operation in a geographic area.

We support SMS, MMS, and http query of matching party identification data.

To address the critical mass issue:

Use lowest-cost cellphone java approach and seamless app with web, make it effortless for drivers to "offer their polyline to Aktalita".

Return some value to every passenger request; for example, the public route matches in case of no private matches.

Use the exact polyline space-time object.. A driver who only wants to go out of his way 30 feet can specify that.