Rethinking regulatory approaches for tech giants

To understand value creation and regulation in the digital economy, we need to start with understanding the economics of data.

Treating data as a physical asset and regulating data access and residency only scratches the surface. To regulate data well, we need to understand how data accumulation leads to control.

DATA ECONOMICS 101

But before we get there, let’s start with Data Economics 101 to set some base principles for this discussion. This is a 101 article but a good place to start thinking about data regulation.

The economics of data 101

These simple ideas have profound implications on how data gets used and how it should be regulated:

“Data has become a key input in modern economic production alongside land, capital, labor, and oil… But data is unlike other inputs, including oil, in important ways.

The same data can be used by many people simultaneously without being depleted, which means its accumulation stands a better chance of boosting productivity and long-run growth…

When personal data about someone is exchanged by two companies, the transaction impacts the person’s privacy and can leave them at a strategic disadvantage. If that person isn’t compensated or even made aware of the transaction, the data market can generate undue harm that leaves us worse off.

Finally, it’s expensive to prevent the theft or misuse of data by others… individual firms are unlikely to consider the impact of their decision on the public’s trust in the broader data economy.”

Understanding the economics of data is important.  A lot of data regulation today treats data like a physical asset, regulating access and residency.

DATA INTEROPERABILITY

Instead, one of the key focus areas of data regulation should be interoperability.

Data lock-in is how platforms control their ecosystem. Data portability and interoperability could offer a solution to this. To really provide agency to users, platforms need to implement data interoperability. This would involve API-based interoperability between platforms that allow users to multihome across them.

Here’s a good opinion piece to go deeper into this topic.

Will data interoperability curb monopoly positions for platforms?

Money quote: “These platforms would have to offer interfaces that let users download their own data or transfer it to another service. Platforms would also need to let users authorize third-party apps that could access and interact with data as long as these “custodial third-party agents” register with the Federal Trade Commission and meet basic ethics and security standards.”

This is progressive regulation. In Europe, the Payment Services Directive (PSD2) is driving a new wave of fintech innovation as regulators require banks to open up data. A lot has been written about how today’s large platforms centralize innovation. Data interoperability could help startups win users over, on the merit of their services, rather than having to compete with massive network effects.

EMERGENT UNIONIZATION

Data interoperability can also create new forms of agency where machines can negotiate on behalf of humans. 

In my research with the ILO on the future of work, I proposed a worker API as a solution for gig workers to counter platform domination.

Key idea: “In order to empower workers, platform regulation would give a worker the right of access to data about their own actions… Worker access to data would be enabled through an API, mediating common standards and capabilities shared by all platforms. A worker- centred API would allow a worker full rights over their own data and would allow them to transfer these data. A worker could then use third-party applications to gain a better understanding of patterns in their data vis-à-vis the workings of the platform.”

Source: Data interoperability and emergent unionization

This leads to an interesting idea of Emergent Unionization.

Key idea: “A worker-centred API could also create an entirely new form of unionization for the platform age… With a worker-centred API, workers could port their data to third- party applications, giving them greater agency. Just as industrial-age workers were represented by unions, worker-centred APIs would allow platform workers to be represented by algorithms. This would enable a form of emergent unionization, whereby workers could be matched and coordinated based on certain needs, allowing workers with similar needs to band together and negotiate with the platform.

These algorithms would enable platform workers to organize among themselves, managing much of the organization and coordination algorithmically. They could even help workers to use real-time metrics to bargain with platforms. Just as data access would, for the regulator, enable effective collaboration with the platform on the shaping of regulation, data access would facilitate more organized and coherent bargaining for workers.

Algorithms could also empower individual workers. A worker-centred API would allow workers to use a third-party algorithm to analyse data and plan future participation on the platform so as to maximize their outcomes. Such an algorithm could operate at both the individual and the collective level.”

Feel Free to Share

DUAL REPUTATION SYSTEM

Reputation portability is a particularly important issue in data portability. 

I covered this in detail while working with the ILO. Reputation portability needs to balance two conflicting design constraints: Making reputation portable enough for users to gain agency, while ensuring that some part of user reputation is unique enough to a platform to give it competitive advantage.

One of the solutions I propose in this paper is the idea of a dual reputation system.

“In a platform setting with regulation determined by multiple stakeholders, a dual-level reputation system could prove beneficial. This would give each producer a single base- level reputation, in addition to a platform-level reputation on each platform used. The regulator would be responsible for creating the base-level reputation, which would help all platforms identify bad actors. The regulator would also be responsible for giving producers access to the learning and re-skilling mechanisms needed to improve their reputation. This base-level reputation would be complemented by the platform-level reputation signalling the worker’s capability to provide a differentiated service, tailored to the requirements of a given platform. The dual-level system would ensure that platforms share data about bad actors across the board, and thereby save individual platforms considerable time and money that would otherwise be spent on identifying those bad actors. Moreover, it would also ensure that platforms could continue to exercise some form of control over the worker via their own platform-level reputation mechanisms that seek to differentiate workers’ skills.

Implementing a dual-level reputation could also help to ensure that all reputation systems are designed from a reward perspective. As noted earlier, platforms that mediate the exchange of highly standardized work use reputation systems primarily for punishing poor actors rather than rewarding high-performing workers. With a dual system, the platform-level reputation, which pertains to the worker’s skill, could be used exclusively for rewarding high performers, while the base-level reputation could be used at the outset to determine who gains access to the platform.”

Read the full report here, particularly Section 8.3 on regulation. 

IDENTITY AS CONTROL

Identity is an important control point for platforms. 

Every platform gains dominance by holding onto user identity exclusively and denying its portability. Platforms will fight tooth and nail to hold on to identity exclusivity and implementing true identity portability is going to be non-trivial.

Implementing identity portability 

This paper talks about why data portability as currently provided by platforms (download your data) is merely lip service to the data regulation regime. To really impact a platform’s control point, identity portability is required. This ensures that a user’s identity is shareable across platforms, allowing eventually content and data to move more freely across platforms.

Identity interoperability works best with cross-platform communication.

An example from the past:

Perhaps the clearest example of mandated interconnection was the interoperability between AOL’s Instant Messenger application and other messenger applications that the Federal Communications Commission required in its approval of the AOL–Time Warner merger in 2002. The regulators were concerned that the market had tipped or would soon tip in AOL’s favor, giving it network effects that would make entry impossible. Those assessments were controversial, but AOL was required to make changes so that its messaging application was able to accept messages from and send messages to other providers’ products.

ADVERSARIAL INTEROPERABILITY

Sometimes, where regulators fail, a smart competitor may prove a better bet towards encouraging interoperability.

When platforms get too large and start imposing their standards and restricting user rights, new alternatives that give agency back to users can create a more competitive market and also enable users to wrest advantage away from platforms.

Adversarial interoperability: Democratizing revolt against platforms

This article talks about the idea of adversarial interoperability where a B-player wrests advantage away from the A-player by consistently ensuring interoperability with the A-player while also building its own standard. Apple’s success with Pages, Numbers, and Keynote, despite Microsoft’s multi-decade dominance is a great example.

Money quote: “Since Pages’ launch, document interoperability has stabilized, with multiple parties entering the market, including Google’s cloud-based Docs offerings, and the free/open alternatives from LibreOffice. The convergence on this standard was not undertaken with the blessing of the dominant player: rather, it came about despite Microsoft’s opposition. Docs are not just interoperable, they’re adversarially interoperable: each has its own file format, but each can read Microsoft’s file format.”

Airbnb interoperating with Craigslist was a similar adversarial move. As platforms become more dominant, adversarial interoperability is an alternative to the break-up-big-tech-by-regulation model. Providing users the tools to break up monopolistic control of a platform can prove very effective.