Free-Market Moloch? Paying to Make Red Lights Turn Green

Traffic timespace diagramI saw this on Twitter a while back, posted by someone who was attending the Wolfram Data Summit: “Data future vision: you're at a red light and can pay for it to go green.” Like a fragment from some lost Sumerian tablet, this cryptic comment is all we have. But it's enough.

Paying for a red light to go green is the sort of thing economist Tyler Cowen and blogging partner Alex Tabarrok (their blog is Marginal Revolution) like to file under a category called “markets in everything.” This seems to have been just a passing thought, but it lodged in my head and wouldn't leave. At first it seemed like the perfect example of the future as a technolibertarian nightmare – the future I fear and dread the most. Then I gave it a second thought, and a third.

And then, after much thought, I made up my mind. Yeah, it's a technolibertarian nightmare – although it's not nearly as big a change from today's reality as it first seemed. But then again, isn't that the problem?

How would something like this actually work? Would drivers buy something like an EZ-Pass that automatically provided preferential treatment at every traffic light? Would a prepaid device be sold with the car itself, perhaps included as a standard feature on larger, more conspicuously-consuming vehicles like Cadillac Escalades? Those ideas don't seem very imaginative – and they're not true to the original, shamanic vision: “you're at a red light and can pay for it to go green.” That seems to describe what economists and marketing types call a “point of sale” decision, not a premeditated bulk purchase.

Most people would probably want a point of sale system, anyway. There are times when it's worth money to get a green light – for some people, there are times it's worth the risk of running a red light – and there are other times when they'd rather wait.

So you'd need an electronic network that links cars and traffic lights. That's easy, since the network of things is well on its way. Better yet, why not have a link between the traffic light and the driver's thoughts? That's on its way, too, thanks to microchip implants and other new technologies that originally designed to control prosthetic limbs. Throw in a mechanism like PayPal that handles each transaction, and you've got your green light payment system.

What if I'm in a hurry and want to pay, but the same is true for the guy approaching in the other direction? Will it be a competitive bidding process? Then costs could vary dramatically depending on the wealth of the parties. Since my regular driving route takes me through downtown Beverly Hills, I'd probably lose most of my bids, causing my driving time to go up.

Competitive bidding would make the process even more complicated. There would be a lot of back-and-forth data traffic, with the light itself as the hub communicating with all the parties — and the drivers probably glaring at each other across the intersection.

Would drivers be able to band together and form purchasing alliances? After all, if a driver behind me or facing me added to my bid, we'd both benefit, and there would be more revenue for the “light.” But to execute this process successfully, the traffic light would have to become a kind of “digital auctioneer,” able to accept competing offers, encourage higher bids, and choose the right time to end the sale and execute the transaction. The light would become a robot intermediary, one that might be perceived as a form of artificial intelligence.

And this would be a powerful robot intelligence, albeit one with a small sphere of influence. That's a major shift in the way these devices have been presented to the public. “Intelligent” robot devices are usually portrayed as subservient tools performing service roles of various kinds for their human masters(1). But our Intelligent Traffic Light would be an authority AI, invested with all the power of the state. You offer it something, and it then waits to see if it has a better offer from another human before proceeding. It's “free” to “refuse” your offer, and if you proceed anyway you will be punished (after being recorded by its cameras).

That's a shift in the human/machine balance of power that might disturb a few people – and would piss off a lot of them. It's a role reversal: In some metaphorical sense, it would become the brain and we would become its prosthetic limbs. The intersection would be its domain now, not ours.

Come to think of it, libertarians may not like this idea any more than I do.

Robin Hanson thinks about this stuff a lot, but I couldn't find any citations from him on the topic of – well, what would you call it? I suppose if I were writing a paper about it I could call it something like “Green Light e-Purchasing: Authority-Figure Robots as Economic Actors in a Technologically-Enabled Nanoeconomic Environment.” (The “nano” in “nanoeconomic(2) environment” would represent both the small physical dimensions of the traffic intersection and the brief duration of the transaction.)

But I don't plan to write an academic paper, as fun as that sounds (and it does), so the idea needs another name. My anti-authoritarian streak wants me to call it “Automated Corruptible Officials: Little Traffic Light Robots That Rich Guys Can Bribe.” Except that, like so many political transactions, this corruption would be perfectly legal. Corruption can be legal,of course. Merriam-Webster's first definition of the word is “an impairment of integrity, virtue, or moral principle.”

Which gets us to the heart of the matter: Would this data future vision (I still don't quite get the italics) lack in integrity, virtue, or moral principle? Would it be wrong? My sense of fairness (or is it just a biologically-hardwired inequality aversion?) says yes. Why should the wealthier among us – who already enjoy so many advantages – get another way to spend their excess riches to make life easier for themselves and worse for everybody else? The public as a whole – that is to say, we – built the streets. We hire the police. We would even build and manage these “smart” traffic lights (or subcontract them to private companies – but it would be done with our collective authority.) Why, this idea is grossly unfair, undemocratic, unreasonable …

… And completely consistent with the way things are done today. We finance mass transportation (some would say we grossly under-finance it), but those with cars pay for greater ease of travel. We pay for the roads and highways, but some travel on them in greater luxury than others (while some never use them at all.) Many highways and bridges collect tolls, charging amounts that are insignificant for a few and a burden for many others. Great Britain has the National Health Service, but if somebody doesn't like the wait times they can join a private insurance scheme like BUPA or pay for their own care out of pocket.

But there's a difference: These examples are not direct, immediate zero-sum exchanges. Everybody waits in line to pay the toll (or pays the same amount but uses an EZ-Pass or its equivalent). Sure, there is always some competition for resources, so in some sense these transactions affect others. But the green-light effect would be immediate and undiluted: People going in one direction would sit and wait – in that very spot, in real time – whenever the people going in the other direction had more money to spend.

If my inequality aversion is triggered – peaceful guy that I am – then other people will have more severe and even violent reactions.

Consider New York City: What would happen at rush hour when a stream of traffic going uptown toward the well-to-do suburbs of Westchester County crossed another stream headed to the less-wealthy neighborhoods of Queens. Would the Queens-bound drivers have to wait until late in the evening, when the last of the Westchesterites has left town?

How long would it take New Yorkers to successfully vandalize a traffic-blocking, offer-weighing, driver-judging, money-extracting, traffic-delaying electronic monster, a red-and-green eyed Moloch sacrificing their time to the wealth of others? I'd give it 48 hours, tops.

That said, the green-light exchange would be an interesting – and rare – exercise in the so-called “free market” (a process that's rarely free and where the parties rarely have equal access to information). Both cars at the traffic light would pretty much know all there is to know about the transaction. We'd quickly learn the “real market value” of a green light be at different intersections, at different times, in different cities all around the world. The lights would generate a gold mine of data that would probably surprise us.

But they'd wreak havoc on the already-struggling art of traffic control, and the nascent field of Traffic Flow Theory would be shot to hell. Elegant models like the “time-space diagram for one way street operation” shown above, developed for the U.S. Department of Transportation, would become useless. They'd drown under an ocean of new variables like driver economic status, geographic distribution of income, and group dynamics.

Then there's the social question: How much should be “for sale” in a society with such grossly inequitable distributions of income – inequities that are themselves usually generated and reinforced by the engines of government? As with highway tolls (or gasoline taxes, or any other regressive tax), amounts that are insignificant to the few are burdensome to the many. How much should the well-to-do be able to inconvenience their economic lessers for what is to them a mere pittance?

You can take the idea as far as you like: If you can pay to turn red lights green, why not pay to hit pedestrians with your car, too? The traffic light could manage the transaction, restricting it to only those pedestrians who are willing to be struck – for the right price, of course. They're out there. After all, some people already get hit by cars so they can file an insurance claim.

Of course, nobody would take our traffic light e-market that far (we hope). But, as Robin Hanson points out, robotic workers will take on an increasing number of cognitive jobs currently performed by humans. If “robot traffic lights” can be paid off- er, “employed to facilitate a market in reduced waiting times” – what about robot doctors? After all, nobody likes to sit in the waiting room. How about digital schoolteachers? Imagine: I throw the “learning module” a few bucks and my kid gets tutored before yours. The there's the supermarket. If I just transmit a few cents to the robo-teller I'll be pulling into my driveway while you're still standing in line watching your ice cream melt.

Some people would say that the solution lies in finding the right balance between publicly provided necessities and private resources: Mass transit vs. private automobiles. Toll roads vs. surface streets. But life is a human right, and time is a dimension of life. Time is the real commodity for sale at our traffic light. The light is selling the time of all the drivers, and it's awarding it to the highest bidder. Is human time fungible? Can I sell you my waiting time at a green light – either actively, through a direct transaction, or indirectly by being unwilling to pay? Those are the same ways I “sell” you my time during a television commercial or at work, or when I use “free” software like Google.

Let's take our thought experiment out to the farther future, to the age of the proposed Singularity. (Some of us remain agnostic about the idea, but this is a thought experiment.)If people upload their “consciousness” to a computer, should the wealthy be able to purchase more storage space and faster processing time for their “minds” than the poor resident of the same central processing unit? There's a sci-fi story to be written there. (I think one has been, actually, although I can't think of it right now. Maybe if I paid for a few more processing cycles …)

Meanwhile, in the traffic-infested present, the city of London charges a congestion fee for use of its roads during the busy daytime hours. That's not so different from the “green light” proposal. The process id managed by a private company, and it was introduced under Ken Livingstone, the leftist mayor they once called “Red Ken.” There's a school of libertarian thought that says the government ought to get out of the public-thoroughfare business altogether and turn it over to the private sector. We're living in the world of “road socialism,” some libertarians say. A free-market model for green lights might be exciting to them.

But the privatization of college loans in the United States was a disaster, and privatization programs for schools and prisons have led to scandals, inefficiencies, and cost overruns. “Red” Ken's congestion scheme has its troubles, too. And there's the communitarian question: How will we continue to have shared experiences as nations and societies, how will we remain a common people, if new technologies eradicate those experiences we all share in common?

What won't change is human nature. If they start charging for green lights, people will find ways to avoid paying. Here's one way: Find a Cadillac Escalade and follow it. He'll pay for all the green lights and you'll get the benefits for free. Of course, that means you'll have to go wherever he wants to go. If that isn't a metaphor for the world we already live in, then what is?


(1) Except for military applications, of course – but readers and viewers of the popular media never imagine these devices might be used against them. Nor are they supposed to …

(2) Papers have been published on something called “nanoeconomics,” but they use the word to refer to the economic impact of nanotechnology. I object! As “macroeconomics” is economics on a large-scale, shouldn't “nanoeconomics” be extremely small-scale economics? Shouldn't the prefix refer to the word itself, not the commodity being traded? Sadly, there is no National Registry of Neologisms where I can file a complaint, nor is there a free market where I can bid on the word and buy it from those who I believe have misused it. So there is neither a statist nor a libertarian solution.