Log in

How to create jobs and fix traffic problems - sparr [entries|archive|friends|userinfo]

[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

How to create jobs and fix traffic problems [Nov. 16th, 2012|11:34 am]
Computerized traffic control gets better every year. My parents remember when inductive sensors under the pavement were new. I've seen computer-vision assisted traffic control devices. Eventually cars will have transponders in them, and later humans will stop driving. However, until that happens, traffic still sucks, and there are so many small changes that could make it better. I'm going to present here two ideas, one practical, immediately possible to implement with minimal effort, but unappealing to some people, and the other rather less realistic but nonetheless at least as effective.

First, I'm going to address mis-use of intersections. Allowing cars to pull forward into an intersection that they can't clear is an example of the tragedy of the commons (http://en.wikipedia.org/wiki/Tragedy_of_the_commons). You, as an individual, benefit from pulling forward and blocking the intersection. It means you get home 30 seconds sooner. However, by blocking the intersection for 10 seconds, you make a dozen other people get home 5 seconds later, and the entire traffic system encounters 30 car-seconds more load than it would have. You have momentarily benefited yourself at the detriment of everyone else. However, what goes around comes around, and you will be stuck at a half dozen other intersections because someone else blocked them, and they will be stuck because of someone else, and so on. Everyone ends up getting home slower than they would if no one blocked intersections, despite blocking an intersection being an apparently good idea to any individual driver. The solution to this problem is simple: write tickets. New York City changed their laws a few years ago (http://www.nytrafficticket.com/blog/index.php/2009/03/21/blocking-the-box-no-longer-a-moving-violation-in-new-york-city/) to allow traffic enforcement (not just "real" police) to write citations for "blocking the box". If the local enforcement regime started writing a few hundred such tickets a day, I do not think it would take very long at all for a neighborhood or entire city to get the message that blocking intersections (and, worse, causing gridlock) is no longer acceptable behavior.

Next, a more fanciful and less penal idea. To illustrate this, I need to digress for a moment and describe something analogous.

There is a computer program called Folding@Home (http://folding.stanford.edu/English/HomePage); it lets your computer spend idle cpu time doing brute force physics/chemistry calculations trying to solve problems related to curing cancer from a biological/chemical standpoint. However, there are a lot of steps that can be taken to solving those problems that humans, with their complex minds and perceptions and insights, can do more creatively than a computer. Thus was born the game foldit (http://fold.it/portal/), which allows people to play what appears to be a video game, complete with tutorials, levels, leaderboards, competitions, and prizes, while they are actually learning to manipulate proteins, and later actually solving real world problems while also producing solution method processes and data that can be fed back into the automated computer program.

I want to apply this approach to intersection management and traffic signals. Virtualize a wide variety of traffic scenarios and let real people on the internet try to manage them as efficiently as possible. Produce ratings and scores, set goals and achievements, inspire competition and exceptional performance. Let them trigger signal changes, handle arriving vehicles and pedestrians, view statistics on the intersection, the locale, other players' performance in that scenario, etc. In other words, make it a game, with the same sort of competitive online community that other such games have. And then, perhaps without telling people when it's actually happening, occasionally give the best players control of real world intersections. The ones with the worst traffic, or the most complicated traffic patterns, or even ones that are simply up for re-programming soon for which additional handling data would be useful.

How often have you arrived at an intersection just before it turned red, annoyed at the timing all the more because there is no one yet waiting to use the newly green light in the other direction? While waiting at a red light, have you had the realization that a gap in cross traffic is more than large enough for an entire cycle of the light to let you go through without delaying the other people? Have you ever cursed when you are stuck with a red light while oncoming traffic has a green left turn arrow that they don't need? Can you imagine being part of a pack of cars released by one intersection just in time to get caught at the next? These are all situations that it is difficult to handle with relatively simple computers programmed with less than perfect traffic detection systems and heavy pre-determined assumptions. They are all situations that an average high schooler with their hands on the switchbox could solve immediately. Such a straightforward approach would certainly be ripe with potential and actual abuse, but if you implement the qualification and monitoring layer of a game community as a gatekeeper then I expect you would see exceptionally positive results with very few negative outcomes. Constraints on actions of the "players", such as hard maximum signal cycle times and wait times, would eliminate most of the feasible remaining avenues for problems.

I have little hope of seeing the second idea implemented any time soon. The first, however, is something I do look forward to. I expect that eventually some municipalities will use red light cameras to enforce existing laws against blocking intersections, but even if they don't then someone else will have to wise up to NYC's approach with letting "meter maids" write citations for these other cars stopped where they shouldn't be.

[User Picture]From: vortex
2012-11-16 07:03 pm (UTC)
Hey! ~ I remember when inductive sensors under the pavement were new!

I don't use Folding@home...I use Seti@home (since 1997) and Rosetta@home (since about 2005)...
(Reply) (Thread)