So basically a docked system minus the kiosk and the bays that lock the bike up. Not a bad idea by any means. It sort of shifts the cost from the kiosks to the bikes which would need to be able to detect if they’ve been locked in the proper area. It would be interesting to see a study on which is more cost-effective at large-scale.
Part of why I mentioned the cost of stations above is that, somewhat cynically, I wonder if resistance to removing street parking is a bigger limiting factor on docked systems than the cost of docks. If so, virtual docks don’t really solve the problem.