The Wired Car in the Wired World

Philip E. Agre
Department of Information Studies
University of California, Los Angeles
Los Angeles, California 90095-1520

Version of 2 September 2001
4200 words.

Note: Since I wrote this article in June 2001, a few of the links have gone bad. I have not removed the bad links, though, since I intended the article as a guide to the links, and not as something that stands on its own. If you can somehow infer citations for any of the articles that the bad links once pointed to, please do send them along.


As the news about new technology goes by in its daily doses, it's easy to lose track of the patterns. So let us take a moment and connect some dots. Let us take a single, familiar site -- the car -- and trace the amazing variety of institutions that intend to wire that site in one way or another. Most of us, especially if we are Americans, drive a car all the time, and we recognize that the car is a point of intersection for many institutional agendas. At any given time, most of us have at least one bureaucratic hassle in progress relating to our cars, be it buying, selling, registration, repair, insurance, law enforcement, parking, you name it. Every one of those hassles is organized by an institution, and every institution has more or less the same components: several layers of law, a sprawling map of interested parties, an information infrastructure, nonobvious interactions between governments and markets, people who specialize in working all of the angles, and a culture of popular resistance and vigilantism.

Institutions exist largely to solve informational problems, and as information technology changes it stands to reason that institutions will change as well. Institutions, however, are not driven in any direct way by technology. Institutions are political, and they change through a complicated and diffuse sort of collective bargaining among the various stakeholder groups. Some groups are more organized than others, and the new technologies create opportunities that each stakeholder group can either seize or fumble. Once the smoke clears, each institution will settle into a new form -- a form that could have turned out differently if, for example, unorganized stakeholders had gotten their act together to shape the technology in some different direction.

As the example of the car illustrates, however, institutions do not evolve independently of one another. Because every person is enmeshed in a wide variety of institutions, and because the institutions all operate on the same planet, all of the various institutional change processes intersect and overlap. One good way to see the intersections and overlaps is to focus attention on a single object -- an object on which several changing institutions are all laying claims. Such as, for example, the car.

We can start someplace obvious, with law enforcement. Now, a common pattern is that new technologies fit into existing institutional niches, not so much changing anything as amplifying things that were already going on. For example, law enforcement organizations already have several kinds of incentives to enforce traffic laws. But as the constituent technologies of various automatic enforcement schemes -- video cameras, data communications, image recognition -- become cheaper, automatic traffic enforcement can become widespread. Of course, political backlash by drivers is a countervailing factor, and it is an empirical question (both for theorists and for the parties involved) which technologies produce which kinds of backlash in which kinds of political settings. Radar enforcement of speed limits, for example, is regulated to a degree by drivers' complaints to elected representatives. But representatives experience many pressures, and they get elected on bundles of issues, not just speed traps. Now we have a new generation of enforcement technologies:,3604,425344,00.html

Britain is the leader here, as it is in many areas of advanced privacy invasion. Deep-running traditions of deference, combined with the Thatcherite legacy of centrally organized obliteration of traditional institutions in the name of decentralization, combined with the New Labour strategy of occupying the entire political spectrum while stifling internal dissent, have made the UK into a free-fire zone of surveillance. Of course, that wouldn't matter if technology were stable. But now the UK is going to be a testing ground for a new generation of surveillance technologies. Cars are only one area, of course; the UK is also a leader in video cameras in public places, and is competing with the US in the area of financial surveillance.

But law enforcement is not the only institution that has an interest in exerting control over cars and drivers. I've long predicted that insurance companies will institute "lower" rates for people who agree to have their cars tracked (i.e., higher rates for people who don't, and since the people who drive to dangerous places will rationally opt out, privacy-lovers will be lumped with this high-risk group), and at least one company has experimented with GPS tracking of cars:
That experiment has ended, but we can expect others. It's probably not worth doing as a stand-alone system, but once cars come equipped with GPS and data uplinks for other reasons, the incremental cost of tracking cars for insurance purposes will be minimal. There are plenty of actuaries who would love to write the algorithm which uses your locational information to produce a rating based on historical experience with insurance payouts to other drivers who have gone to similar places. It will be possible to print risk maps showing that driving past location X at time Y costs $0.02 in insurance, where driving past some other location at some other time costs $0.09. Where will the red zones lie? Notice that these red zones need not measure any intrinsic danger in the place itself. Rather, the statistics will pick up all kinds of correlations whose underlying causes may not be obvious. If people who leave work at the factory early tend to get a drink on the way home, then the increased accident rates of the drinkers may spill onto other people who happen to drive past the factory gates at the early hour. Of course, in theory it is in the insurance company's interest to separate out the variables. But their data won't be infinite.

It's not just insurance companies that have an interest in tracking cars, of course. At least one rental car company is already doing so:,4586,2778752,00.html
We're not just talking about theft recovery systems, which have been around for a long time, and which do not need to be activated unless the car has been reported stolen. We're talking about continuous monitoring of the car's location, and potentially of many other aspects of the way you're using it. The rental companies are in much the same position as the insurance companies; they want to charge their customers based on the risk they present. They can easily put language in their form contracts that lets them bill your credit card automatically whenever the GPS device says you're speeding.

The market does create countervailing forces. Customers will shop for a bundle of features among different car rental companies, and included in that bundle will be contract terms about tracking and fines. In the magical version of the market that economists theorize about, the result will be an equilibrium in which customers decide how much tracking they are willing to put up with, in exchange for how much of a discount on the rental rate. But in the real market, things are more complicated. Form contracts take time to study and education to understand. It will be nearly impossible to compare the form contracts side-by-side. Customers may assume that they have legal protections that they they do not really have. The tracking information will probably have commercial value, and customers may not understand the consequences of that information circulating out in the world.

Another institutional question is where roads come from. One answer is that people pay taxes and then politicians and bureaucrats put roads down, with interest groups working both sides of the issue all the way. In the idealized world of the economists, however, roads should be paid for by the people who use them. Technology makes this vision more feasible by making it easier to collect tolls automatically, and the politics of automatic toll collection takes on an endless diversity of forms. In Singapore, authoritarianism with high-tech characteristics allows the government to impose variable pricing on all of its roads, which in turn requires every car and driver to be integrated both legally and technologically into the market-making bureaucracy. In California, various state and regional governments have been struggling with a plan to finance roads by letting the contractor collect tolls until they've made their money back. This plan has generally been a disaster on account of the many devils in the details. One such devil is usability:
Simply put, drivers don't use the toll roads because they find them confusing. The roads don't have consistent rules about things like exact change, and people routinely get big penalty bills in the mail because they took the wrong exit or couldn't figure out how to use the toll system for the first time. Now multiply this situation by dozens and you have the usability nightmare for all of the independently developed electronic systems that impinge on cars -- and that's just cars, not even counting all of the other machines we deal with all day.

So far I have been emphasizing institutions that have bad reputations. But my point is not that electronic wiring of cars is necessarily all bad. A lot depends on how it is done. Let us consider, for example, proposals for digital systems relating to traffic accidents. I have mentioned that some theft-detection systems are unobjectionable since they are not activated unless the car is reported stolen. Likewise, it is not so bad to have a device that incorporates an accelerometer, and that is activated automatically (only) when the device detects an acceleration of sufficient magnitude that injuries are likely. In those situations, the device could alert appropriate authorities and upload full details of the car, its position, and the identities and medical records of the car's owners. Of course, one would want to be certain that the information was encrypted so it couldn't be intercepted by people who you don't want knowing about your traffic injuries. Nonetheless, at the moment when the accelerometer trips, your desire for privacy normally declines rapidly relative to your desire for relevant people to know about the situation. Here's a government report on the idea:
But then there's reality. Reality, once again, is that these things tend not to happen as stand-alone systems, and stand-alone systems tend over time to be absorbed into more general architectures. So consider the case of GM's OnStar. When you get into an accident, OnStar reports the accident to the authorities. But where the USDOT was writing from the idealized perspective of physicians and highway safety experts, OnStar is trying to be a generalized platform for the delivery of electronic services to GM drivers.
So, recently, a GM driver got into an accident and decided to make a run for it, only to be tracked down using information that his OnStar had provided:
Of course, nobody likes hit-and-run drivers. Once these systems are in place, however, their function starts to creep, and we tolerate function creep because at each step we can identify a bad guy who is going to get his comeuppance as a result. But who else is OnStar telling on, and where does it stop? The main tradition of computer system design is inherently privacy-invasive in the sense that it assigns identifiers to everyone and everything in sight, and once a system is built this way it's very hard to stop it from sliding down the slippery slope of ever-broadened functionality.

As electronic systems proliferate, moreover, security attacks become more likely. Most system designers live in an idealized world without hackers, and they haven't gotten it through their heads that any security hack can be packaged in a black box and disseminated by the thousand. So take the case of wireless "keys" that unlock cars at a distance. Those things are just asking to be spoofed, and history shows that designers always underestimate the likelihood of such attacks.
Imagine the consequences as your car goes on the Internet. We're used to viruses in our desktop computers and we've heard about viruses in our palmtops. Next we'll have viruses in our cars, and then we'll have them in our pacemakers. Wireless communications is especially asking for it, and a public-spirited lawyer once mailed me a package of documents from a California Air Resources Board plan to equip all new cars with a device that would upload the car's identification number and emissions equipment status in plaintext whenever it was pinged by a roadside transponder. Wrong!

Lots of money is at stake in the internal workings of your car, and the role of electronics, which is already enormous -- from globally networked teams of designers using CAD tools to anti-lock brakes and computerized fuel-injection -- will only grow. The laboratories of advanced experimentation are in auto racing, where computerization has grown so extreme that cars routinely "crash" in a software sense:
A car whose operating system has crashed is not only a joke, but a serious danger to other drivers. And that's not even an environment where competition has gone as far as the sorts of electronic warfare that are taken for granted in military contexts, or that ought to be taken for granted in a world of script-kiddie thieves. Electronically hyped race cars can also be dangerous to their drivers as a result of their sheer speed -- race car drivers have recently started blacking out on sharp curves.
The politics of automotive technology is especially hard-fought in the context of environmental issues. When Al Gore made the commonplace observation that the internal combustion engine's days are numbered, Republicans twisted this into a generalized attack on the family car. The fact that these reptiles expected such a lie to work indicates the magnitude of the interests and the depth of feeling around basic issues of automotive technology. Digital control will be central to whichever technology succeeds internal combustion, but no alternative will be possible until a substantial infrastructure is built. This has been one of the vulnerabilities of electric cars, which California has been compelling the auto companies to build. It hasn't worked, of course, because the auto companies have hundreds of ways of subverting the very long learning curve required to establish economies of scale and complementary services for the new technology. Passive aggression is quite enough.
They can thus complain that the technology is "expensive", which of course it will be so long as electric cars remain a small percentage of the cars on the road.

Now let us move to the world of electronic gear in the interior of the car. There is a huge amount of action in this area:
I've mentioned that GM wants to make OnStar into a general platform for delivering services into cars. As with the marketing of many new technologies, they are introducing the services first into their high-end cars. So it is not surprising that stock trading is high on their list of extended OnStar services. Some of these services will be standard equipment, so to speak, whereas others will be offered as options or by subscription. That dividing line will be determined by marketing considerations and by the economics of the service. The important thing is the generalized platform, which will rapidly come down in price and itself become standard equipment on all cars. And so naturally one part of their long-term picture is to push advertising into cars. I'm not making that up:
And that's not even including advertisements on electronic billboards along the side of the road:
These networked advertising systems will be technological wonders. Think of banner ads in your car, alongside the highway, everywhere. Of course, banner ads on the Web are one of those technologies that was overpromised because lots of people couldn't tell the difference between the idealized market and the grotty realities of markets as they exist right now. But in the automotive environment, huge amounts of personal information really will be available for the customization of advertisements. How about having ads pop up for every business you approach along the roadway, each of them tailored to your own demographics? If your car knows you've been driving for ten hours straight, then you can guess what kinds of ads will pop up. Now imagine the automatic capture and aggregation of the demographics of the drivers along a given roadway. The demographics of that moving audience could be posted automatically on the Internet, and then advertisers could bid in real time to put their own presentation on the roadside billboard. One-minute intervals would be about right. You could then have ads that appear on successive billboards as your little cohort of drivers makes its way along.

It is well-understood that these ads can be distracting, especially when they are bright and animated, and the relevant public relations departments are gearing up their "issues management", for example with advertising campaigns encouraging people to drive responsibly and pay attention. These ad campaigns, pioneered by cell phone companies, are aimed as much at regulators and legislators as they are at drivers; they are part of an elaborate professional practice of looking like you're doing something about a problem without sacrificing any of the profits that you enjoy from it.

The ads are one part of a big picture. America is the land of disastrous automobile-driven urban planning, or non-planning as the case may be, and nobody is surprised that cars are increasingly turning into apartments with wheels. The pioneers were the mobile workers, like salespeople and truckers, who work out of their vehicles more or less full-time. Wireless data networking kicks the "car office" into high gear:
The next step is the in-car entertainment center, especially for children who would otherwise be constantly asking "are we there yet"?
The first generation of these in-car technologies are more or less transplanted versions of the home technologies, but we can expect that future versions will be designed specifically for the car environment.

Now, from one perspective these technologies represent the next step in an already well-established trend whereby everyone becomes isolated in their own personal media zone. It's "bedroom culture" generalized to the back seat of the car. Shouldn't everyone be sitting around the fire having deep conversations and teaching their grandchildren the folktales that they learned from their grandparents? Well, that's hard to do when you're in a car. Having your kids in the back seat of the car while you're driving was already an impossible situation: enough togetherness to be annoying, but not enough to be useful. Children, after all, feed on their parents' attention, and attention is precisely what a driver doesn't have. Back-seat entertainment systems aren't destroying any quality time; instead they are bringing out the inherent logic that already existed in the situation: the car as the technological reification of the atomized society, with everyone knitted into a welter of institutions that pick them up and haul them from one place to the next with no special concern for geography or relationships.

Again the picture is not all bad. Bill Mitchell observes that the same digital networks that allow people and things to be connected at a distance also allow people to strengthen connections that do require geographic proximity. People are somewhat more free to live where they want. Some visionaries had predicted that this would lead to an increased gentrification of rural areas, and some of that is surely happening. Just as important, though, is a renewed gentrification in industrial-world inner cities. Urban density has the great virtue of creating critical mass for services that require both geographic proximity and economies of scale, and we can only hope that digital technologies will be put to use in making dense urban centers more livable, for example through computerized vibrational analysis that makes for quiet compressors. Then maybe we Americans can be more like Europeans and get rid of our cars altogether, and Europeans can stop becoming like Americans and get rid of their brand new cars before it's too late.

In any case, we have hardly begun to reckon with the huge world of institutions that want to get their electronic hooks into our cars. A guy who designs atom-bomb powered X-ray lasers at a US national laboratory once mock-ruefully told me that "the biggest problem with the modern world is that we now have institutions that are built on derivatives" -- that is, institutions with a vested interest in changing the world as much as possible, not randomly of course but in ways that create a continued demand for their own particular kind of work. Ordinary people's lives are a hassle in large part because of all the experts and social climbers who get ahead only if they are changing things whether they need changing or not, and now there is a whole world of intersecting institutions that have an interest in changing the technology of transportation. Here we have one version of the standard sales-pitch:
Endless hundreds of white papers and PowerPoint presentations have been devoted to disseminating this sales pitch, which is the symbolic foundation of numerous companies and careers, particularly in this relatively early phase when "intelligent transportation" is generating little actual revenue. Intelligent transportation, so-called, really began as a push to find civilian uses for military technologies at the end of the Cold War, when everyone assumed for some foolish reason that society would be demilitarized. Alright, it was more complex than that, as all of the various interests converged from their own angles into a loose, baggy coalition. But the vision was very much driven by the traditional transportation bureaucracies, such as the ones who have recently built grandiose traffic monitoring command post headquarters for themselves. Whole institutions are devoted to collecting and archiving information about traffic for purposes like maintenance, planning, and safety:!.pdf

Much of this information is anonymous, of course, but that's because it is collected with the primitive old analog methods of data capture. Modern data comes in fully identified digital records, and most people hardly realize how established the interests in transportation data-collection are. Just wait until identified data becomes the rule rather than the exception.

The world of "intelligent transportation" is enormous. Here is the main US organization:
Here are some directories of projects:\JPODOCS\CATALOG/13464.html\JPODOCS\REPTS_TE/13338.pdf

Here are some of the conferences:

(The ITS America conference happened in Florida earlier this month.)

And here is a summary of conclusions from ten years of federally-funded projects:\JPODOCS\REPTS_TE/@9W01!.PDF
A problem with most of these projects is that they are starting from a technology base that's too primitive, and so they aren't informed by the decentralized ethos of the Internet. In the end these 1980's technologies aren't good for all that much, and the emphasis has shifted from the roadways to the vehicles.
A road is a road, pretty much. But as information technology platforms become standardized and universal, the world will change a lot. Cars and their drivers are embedded in a tremendous variety of institutions, and new electronic technologies will allow every one of those embeddings to become more present, more continual, more informationally complex, and more integrated with one another. The web of institutionally organized relationships into which we are all knitted will become more tightly woven. None of these relationships are all bad, or else it would be relatively easy to get rid of them. But neither are they inevitable in their design. The fact is that few if any of the functionalities I have described require people to be identified; nearly all of them, in other words, can be provided anonymously.

The point generalizes. Every institutional relationship, whether buyer/seller, insurer/insured, legislator/citizen, or cop/driver, has its architecture. That architecture is a product of history -- that is, of conflict and compromise. The architecture is partly legal, partly technical, and partly cognitive and cultural. Institutional architectures rarely spring full-blown from nowhere; most often they are shaped incrementally, and the power of each stakeholder group to shape an institution tomorrow will be influenced by the way the institution works today. We keep hearing that computers are revolutionary, but the opposite is more nearly true: the conventional way to design a computer is to start with the existing institutional order and inscribe it into machinery. But we don't have to design our technology that way, or our lives. The technology really is value-neutral to a great extent. It's a matter of choice.

What values will we choose to inscribe into the technologies and institutions of transportation in the future? We'll express our choices in many ways: through politics, through the market, by subverting the system, and in the ways we imagine our options. Contradictory as it seems, we need a form of imagination that is grounded in reality. Look closely at any of these incipient technological-slash-institutional projects and you'll see a lot of reality that isn't imagined in the beautifully simplified forms of imagination of the past.