The Technium

Class 1 / Class 2 Problems


There are two classes of problems caused by new technology. Class 1 problems are due to it not working perfectly. Class 2 problems are due to it working perfectly.

One example: many of the current problems with facial recognition are due to the fact that it is far from perfect. It can have difficulty recognizing dark skin tones; it can be fooled by simple disguises; it can be biased in its gendering. All these are Class 1 problems because this is still a technology in its infancy. Much of the resistance to widely implementing facial recognition stems from its imperfections. But what if it worked perfectly? What if the system was infallible in recognizing a person from just their face? A new set of problems emerge: Class 2 problems. If face recognition worked perfectly, there would be no escaping it, no way to duck out in public. You could be perfectly tracked in public, not only by the public, but by advertisers and governments. “Being in public” would come to have a different meaning than it does now. Perfect facial recognition would probably necessitate some new varieties of public commons, with different levels of disclosure. Furthermore, if someone could hack the system, it’s very trustworthiness would be detrimental. A faked ID could go far. We don’t question perfect tech; when was the last time you questioned the results of a calculator?

Another example: Self driving cars. Self-driving cars don’t self-drive very well. They are getting better, but for the next several decades their problems will be Class 1 problems of imperfect function. We will demand near perfect results from robot-drivers (a higher standard than we demand from human drivers), so all the hard problems of detecting edge cases, acts of god, and the weird behavior of human drivers will prevail. Eventually, the tech will be perfected, and then we will encounter its Class 2 problems. In the Class 2 regime, driving a car yourself may be outlawed as too dangerous. The imperfections of human drivers may be incompatible with perfect robot drivers. When the system fails (say from a solar storm) its perfection may not permit it to degrade gracefully to accommodate less-than perfect drivers. A well-functioning robot car infrastructure might lead to more intersections with pedestrians; we might become more comfortable walking alongside silent automobiles that never crashed — until they did.

Class 1 problems arise early and they are easy to imagine. Usually market forces will solve them. You could say, most Class 1 problems are solved along the way as they rush to become Class 2 problems. Class 2 problems are much harder to solve because they require more than just the invisible hand of the market to overcome them.

Take cell phones. The first versions of consumer cell phones were too big, they only worked in some places, they had frustratingly short battery life, and their rings and talking on them were disruptive. Most importantly only the rich could afford them, in a new inequality. At the time many saw these problems as inherent in the technology. Yet years of intense market forces fixed most of those problems, making smartphones that silently vibrated, and had quiet text, and became so cheap and ubiquitous every adult on the planet has one. Unlike computers, they rarely crash, are easy to operate, and are extremely reliable. They just work. The cell phone quickly jumped into Class 2 problems.

Whereas once the problem was “not everyone has this technology that doesn’t work very well” now the problem is “everyone has this technology that works very well.” We now contend with a technology that is present everywhere, all the time. Billions of people around the globe are connected 24/7, which allows all kinds of information, ideas, as well as rumors and disinformation to ricochet and touch everyone in an intimate way. The technology can suggest, recommend and “guide” us through the billion-eyed cacophony of everyone talking at once. Mob fears and beliefs can take over. Whispers are amplified and distorted as they cascade through friends of friends.

The difference between Class 1 and Class 2 problems is that Class 2 problems cannot be solved by the market alone. Entrepreneurial spirit and the profit-mode are perfectly capable of solving most Class 1 problems. But Class 2 tech has already been perfected, and is ubiquitous — it works and everyone has it. What can the market do in this case? Making it better and selling more aren’t options anymore; those are saturated. What can the market do if facial recognition works perfectly and is everywhere? If robot drivers are the default? If everyone is connected to everyone all the time? These kind of system challenges require a suite of extra-market levers, such emerging cultural norms, smart regulation, broad education, and reframing of the problem. These are soft, slower moving forces that are currently not given the attention they deserve.

To deal with ubiquitous accurate facial recognition when it comes (and it will come) requires a societal consensus on what it means to have a face that is both personal and public, to re-evaluate what public or private even means, to ensure symmetry between watchers and the watched, and to encourage expansive ideas around the very notions of identity of any type. A lot of this work is beyond the realm of dollars, and will take place in schools, courts, forums, communities, tweets, congresses, books, and late at night. When technologies reach the state that they work extremely well and become ubiquitous, their problem domain shifts from the realm of quick cycles powered by money, to the slower cycles of cultural imagination. To solve the problem of perfect facial recognition demands an expanded imagination, society wide, with new and different ideas about our face and identity.

The latest fashionable tech is crypto. While the math behind blockchain is utterly reliable the implementations so far have many Class 1 problems. Crypto is hard to use, easy to trip up, biased to early adopters, an energy hog, and of marginal utility except to make money. But all these problems will be overcome by entrepreneurs. Someday blockchain will be ubiquitous and boring. It will be perfected and its wide-spread adoption will enable many thousands of new types of organizations and relationships that we can’t even imagine today. Blockchain tech could unleash collaborations of several million members working on one project in real time, or orgs that are far more leaderless than today. When crypto succeeds that way, it will graduate to Class 2 problems. At that point, entrepreneurs alone won’t solve those. These new problems will require a social imagination to revision what orgs do and what they are for, to re-imagine what transparency in a group means, to re-evaluate the role of money, or even the meaning of money. These in turn will launch new social expectations and norms of behavior, and in turn as a consensus forms, new legislation to codify the norms.

Class 1 problems are caused by technology that is not perfect, and are solved by the marketplace. Class 2 problems are caused by technology that is perfect, and must be solved by extra-market forces such as cultural norms, regulation, and social imagination.




Comments