Open Maxis and New Frictionists

Fey Action II convened an all-star team of passionate civic tech builders. These are folks who have been reading from the same open source hymnbook for quite some time. But in a series of unforgettable conversations, an important debate bubbled to the surface.

To oversimplify slightly, there are two schools of thought. Let’s call them “Open Maxis” and “New Frictionists”.

The Open Maxis hold that we should push as much information as possible into the public domain to deflate power concentrations and maximize digital public goods. They want to redouble the push toward informational openness to ensure that AI’s benefits redound to all.

The position of the New Frictionists is for better and worse more complicated. Many of these people were Open Maxis in the Web 2 and/or Web 1 eras, but AI’s capacity to ingest and exploit information en masse has made them uneasy. Some worry that by making new information (art, data collection, etc.) fully open, we may disincentivize important forms of creativity and pro-social data collection. Others are frustrated by the open source movement’s difficulties in financing its work and maintaining the tools it builds. Consequently, they’re looking for new ways of establishing limited control over how information is used – friction in the system. New Frictionists tend not to be apologists for traditional IP. Instead, they are tiptoeing into a design space of new licensing schemes and technical information management systems.

It is far from clear who’s right, but I lean towards some version of New Frictionism. To see why, it’s necessary to rewind the clock at least to the middle of the 20th century.

In 1962 the economist Ken Arrow articulated an important idea that remains at the heart of information economics. He argued that in a competitive market economy, innovation receives far too little investment. This seems counterintuitive – doesn’t market competition drive innovation? Not really, says Arrow. Investors don’t like to put money into innovation, because they can’t predictably capture the gains it produces. The reason is that once something has been invented (or an idea communicated, or a text written), everyone can just copy it. The cruel twist is that the very fact that everyone can copy innovation is what makes it so valuable to society. In an ideally efficient economy, investors would prefer to invest in innovative ideas, technologies, etc. versus more easily protectable, proprietary assets like goods and services. Yet in competitive markets, the reverse is true. This has been called “Arrow’s information paradox.”

There is no perfect solution to the paradox, but there are, roughly speaking, three ways to mitigate it.

The first way is to attack the problem by non-market brute force. If a state decrees massive investments in innovative research, it addresses the markets’ inefficiency. In fact, when Arrow was writing in the 1960s, people were worried that the Soviet Union would outcompete the American economy by doing just that. The drawback with this method? It massively consolidates power, letting the state drive the economy. (You can also think of big entrenched tech monopolies as playing a statelike role here.)

The second way is to limit the competitiveness of the market economy. This is what intellectual property law does with patent, copyright, and trade secret protection. (Call it Old Frictionism.) These laws give investors a reason to invest in information production by granting them mini-monopolies. The drawback? It causes the information they produce to be wildly underexploited. A ton of potential social welfare goes to waste while patent or copyright holders hoard their rights decade after decade.

The third way, which is slightly miraculous, is to build a culture around open source. This happened to a significant extent in Silicon Valley during the late 20th and early 21st centuries. It turns out that people will do amazing intellectual work and give it away for free if they believe in the cause. And this unleashes enormous value. It is an important part of the reason why the US technology culture dominated the globe in those decades. The drawback? It does not lead to anything like a fair distribution of gains. Many people who create massive informational value end up with nothing. Others, who stand upon their shoulders by building proprietary ways of exploiting that information, get extremely rich.

So what’s the best way forward into the AI economy?

In many ways, it feels like the open source solution has run its course. The Silicon Valley economy that it spawned was neither fair nor broadly beneficial enough to make anyone excited about running the same playbook again.

The brute-force investment avenue feels blocked, too. States are pulling back from active roles in the economy, and consolidating power in worrisome ways. Statelike tech monopolies are still investing in innovation, but the more powerful they get, the less confident we can be that this will continue. And they are even less accountable to the public than states.

Meanwhile IP, or Old Frictionism, is broken. Patent and copyright holders have long been gaming the system to extend their monopolies beyond all reason. Their worries about competition from generative AI has caused them to dig their heels in further. Worst of all, the shape of patent and copyright protections no longer match the shape of modern information production. The most important breakthroughs are reached not by individual tinkerers in workshops who can file for a patent, but by huge networks of people building incrementally on one another’s work. Creative work has similarly evolved to be more gradual, diffuse, and collaborative.

But we can’t just throw up our arms. If we do, we will disempower all but society’s most powerful actors. We’ll lose the human culture of creativity and innovation, and ultimately slow down progress, because the little people simply won’t be reasonably rewarded for participating.

So we need to attack the information paradox in new ways. Here’s one line to draw: future versus past information production. Suppose we just drew a line under the year 2000. Everything from before then should be pushed, within the limits of the law, toward Open Maxing. Legal enclosures of cultural and intellectual heritage from the 20th century are frankly past their sell-by date.

But going forward, Open Maxing doesn’t feel like it solves the puzzle. Without some way of controlling information, everyone will soon become aware that their knowledge work is being immediately and profoundly appropriated by AI models, and proprietary businesses built on those models, which are much better positioned to reap the value than they are. In order to avoid this collapse in the incentives for knowledge production, we need to think about forms of friction that look different from traditional IP. Part of the answer might be radically reimagined legal IP rights. But honestly, states don’t seem like they’re up to that task these days. Therefore we also need to think about new technical systems that would let individuals and communities track, gate, and titrate out private information with more precision than has been possible in the past.

There are more questions than answers here. But I think they’re the right questions. This is what draws me toward the New Frictionists.