From: hal@finney.org
Date: Fri Feb 08 2002 - 19:35:29 MST
Anders writes:
> Just a thought (to get back to extropian matters): suppose drivers could
> affect lights through some technological means. Would there be any
> system that would make the traffic throughput more efficient than the
> current timer-based lights?
A market-based system would be interesting to try. Give each intersection
an owner (could be just a software agent) which tries to maximize profits.
Then sell green-light futures seconds or minutes in advance. This would
allow someone to purchase green lights for their entire cross-town trip.
At uncongested times, there wouldn't be many people bidding the other
way and you could get your green lights for cheap. But when there is
congestion, more elaborate mechanisms would be required. You would
want to be able to see who else is setting up green light paths which
are going your way, so you can add your clout to their path and outbid
rivals who are setting up conflicting (crossing) paths. You would also
want to be able to easily slip your schedule forward and backward a few
seconds so as to avoid crossing conflicts as much as possible.
It's easy to say that the market will screw in the lightbulb, but actually
setting up the institutions to let a market work is another matter.
Optimizing traffic flow is a complex, nonlocal problem which may not be
ideally suited to market mechanisms. But in principle the market takes
into consideration the wishes of all participants and prioritizes them.
If mechanisms can be arranged to let the market work, this could be an
optimal solution.
I've been thinking about similar ideas for traffic flow in general, not
only at intersections but on highways as well. Wouldn't it be great if
there were some mechanism to pay the slowpoke in front of you to pull
over, or to get the tailgater behind you to move back? The signals which
exist now from driver to driver are limited and clumsy. Surely it should
be possible to improve on what exists today.
Hal
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 13:37:38 MST