How Do You Solve a Problem Like 100,000 Uncoordinated Driverless Cars?
Ian Steadman
834

This sounds like an amazing new technology that brings up some serious questions about the limitations of the tech from a computer science perspective. The distributed nature of the solution gives way to its own problems, such as hardware failures (common on clouds like Amazon built off commodity hardware), consistency issues across the machines, and interactions on the borders of the simulated tiles.
But I think an even greater problem is looming, especially in city modelling, which is that by the time cities are ready to be modelled in this system, they may already have tens or hundreds of thousands of sensors (the emerging Internet of Things). These extra objects in the virtual environment may themselves make up the bulk of actors that each machine needs to account for. The cities that we need to model will not only be interacted with, but will interact back, and that may prove a much greater challenge than modelling current cities.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.