Taxonomizing platforms to scale regulation
How to implement “rubber banding” on major platforms
|Jason Barrett Prado||Nov 18, 2019||4|
If you’ve played Mario Kart, you’ve probably noticed its rubber band AI. Basically, when you are in the lead, the computer players speed up and get better random items. When you are losing, the computer players slow down and get worse items. This keeps the game competitive in order to make it more fun. You probably don’t notice it when you’re losing but it might make you mad when you’re winning.
We should be thinking about tech platform regulation through the rubber band metaphor. The further ahead a platform is, the more regulation needs to be applied to it. The intuition is that there’s a spectrum of power in a marketplace from an individual actor to a full monopoly. As an entity approaches monopoly, they have more power and the rubber band needs to pull them more tightly. Otherwise, markets are prone to unstable equilibriums where those already ahead get further ahead. This causes innovation to suffer, among other deleterious, anti-democratic effects.
Let’s think about how social platforms could be regulated. I’ve thrown out some numbers and some ideas; real regulations would need to be precise and fluid.
When a platform has 0-5M users, a network should only be subject to basic privacy regulations: explicit opt-in for data collection and full wipeout support. Data takeout isn’t really important at this scale.
At 5-20M users, this network may be a vital communication link for whole cohorts of people. It’s nowhere near a universal monopoly, but it may monopolize some important aspects of some people’s lives. This kind of network needs robust data takeout tooling. Tools for viewing collected data and getting data out should be robust.
At 20-50M users, the network contains a lot of data and needs to start publishing transparency reports about what data is collected and exactly how it is used. Every data storage mechanism and pipeline needs to be published in the form of code and plain language descriptions. Any outside observer should be able to understand the complete flow of data through the system. Audits should be rare but when they occur and discrepancies are found, the default penalty should be CDP (corporate death penalty).
At 50M-100M users, this network is a monopoly over people’s connectivity to their communities. It is dangerous for a free society to hand such power over to private hands, but if we are going to do so the network needs to be controlled as tightly as possible by democratic means. For starters, there is no reason for such a network to be operated by closed-source code. What are they trying to hide? Who does secret source code serve? Not users, not society. As a compromise, all code repositories within the platform are tailed and released publicly with a six month delay. High intellectual property value code (HIP) and security-related code can be granted exceptions, but they will be rare. I’m excited about time-delayed open source licenses; I think experimentation in this area will make it easier to force platforms to open source.
At 100M+ users, this network is indistinguishable from the state, except it’s inexplicably in private hands. Any datasets used internally must be anonymized and shared with licensed competitors. If that can’t be done for technical reasons, the datasets don’t get to exist. Operations on a platform of this scale are inherently political, and decisions over such operations must be made using the best political means we know of. To me, this means democracy. We don’t yet know how to inclusively govern megaplatforms, but these companies will be required to fund research into governance at scale. As a strawman, consider a governing board made up of owners, elected officials, platform developers/workers, and users.
You can think of analogous rubber bands for search: zero-cost inclusion of competitors for Google’s various integrated services (hotels/flights being recent contentious examples). Maybe Google should be required to open its indexes to licensed competitors. And clearly we’re not served by Google’s excellent source code being kept secret.
Rubber bands should apply to commerce and logistics, too. When food delivery apps were taking off, we could allow light regulations while they found product-market fit. Now, we need to be ramping up minimum wages and benefits for platform workers. We should be escalating sustainability requirements as well: set a timeline for banning ICE cars and non-compostable packaging. Data/code transparency requirements should escalate just like with social networks.
I imagine this regulation proposal could draw criticism from the left: We don’t want to have better-behaved monopolists control our technological institutions, we want to own them ourselves. I definitely see this critique, but I think of these escalating regulations as non-reformist reforms. As a collectivity we don’t yet know how to build and govern megaplatforms. Privately-held platforms with strong democratic controls give us a playground to experiment with while we plan for a better future. I ultimately share the same goal as many on the left: platform expropriation.