The NETMundial 10th anniversary event was held April 23 and 24 in Sao Paulo, Brazil. As we wrote earlier, “Netmundial was a transformative moment in global internet governance. Inspired by the Snowden revelations, and amplified by the U.S. government’s announcement that it would relinquish its control of ICANN and the DNS root,” it represented the triumph of a globalized Internet governance that ratified the power of nonstate actors to govern themselves.

The organizers of the NETMundial revival described their goal as: “to further multistakeholderism to all relevant global governance and decisional arenas.” In other words, it made multistakeholder (MS) ideology the basis of its agenda and broadened the scope of MS to potentially any digital issue.

A Badly Done Process

The organization of this event lacked the spontaneity and openness of its predecessor. Brazil CGI appointed itself as the official steward of multistakeholder (MS) ideology and then selectively invited people they liked (and ignored people they didn’t) to its executive and drafting committees. A systematic exclusion of Internet governance people from North America was in place. The Chair and co-chairs of the High-Level Executive Committee (HLEC) had four Latin Americans and two Europeans but no one from North America, Asia, or Africa. No one from North America was invited to represent academia or the technical community. Instead of the open participation of 2014, a central committee acted as gatekeepers – one had to submit an “expression of interest” and be approved (by whom?) as a participant. Even some of the people involved in the HLEC did not like this process.

“Consultations” consisted of filling out a very lengthy, badly designed survey instrument that asked repetitive questions and required answers that did not always reflect the views or agendas of the respondents. Worse, anyone who suffered through taking the survey now know that it played no apparent role in the outcomes, as there was no compilation of the results and no reference to it in the final document.

The Multistakeholder Ideal is Over

For anyone playing close attention, the NM+10 outcome document makes it clear with finality how empty and played out the concept of “multistakeholder governance” is as a guide to the future of the digital world. In the past, multistakeholder was a counterpart to multilateral. Upholding multistakeholder governance meant supporting the organically evolved internet institutions – that is, governance by non-state actors organized around private sector nonprofits with a transnational governance scope. Support for MS also meant resisting efforts by national governments or intergovernmental institutions to take over or subordinate these institutions to the multilateral system.

Now, one of the supporters of NM+10 informs us that we have been

“mired in a false dichotomy of multilateral versus multistakeholder approaches. The São Paulo Guidelines clear up this misunderstanding by illustrating how multilateral processes can incorporate multistakeholder engagement, fostering cross-fertilisation rather than competition between these governance models.”

So NM+10 has eroded, if not erased, a key distinction between modes of governance.

Insofar as there is anything distinctive or meaningful about MS governance, NM+10 asserts a fantasy egalitarianism, a world in which there are no conflicts of interest, in which any actor in the digital economy participates in all governance processes on an equal footing, there are no power disparities between the participants, there are no linguistic or cultural barriers to communication: there is only collaboration and consensus. It unrealistically posits a world in which here are unlimited resources available to foster and equalize participation, and there is no identifiable locus of authority which can be captured and abused.

Sorry folks: this is not the world we live in.

The heralds of MS also seem to have no conception of the proper balance of market governance and collective, political governance. Prices and markets are a governance mechanism – and as communist economies learned the hard way, an indispensable one if there is to be prosperity, innovation, competition, and more widespread access to technology. Sometimes, however, political intervention or structuring of market processes is beneficial. The most pressing governance question we face in the digital economy, in fact, is what should be governed by collective (political) processes and when should organizations and individuals be allowed to make choices for themselves in the exchange economy? The authors of the NM+10 statement seem to be entirely unaware of this choice. They have no conception of market governance and no conception of its proper role.

Based on this experience, we can clearly realize that WSIS is over, the IGF is what it is, ICANN and the I* institutions are well-established, and we need to move on. Appeals to multistakeholderism don’t move the needle. We need to be equipped with more sophisticated notions of technology governance.

A primer on governance institutions

The São Paulo statement says:

“…a decade after NETmundial, it is high time to address the lingering unresolved issue: how to help all actors to contribute to a multistakeholder process to create the networked global governance architecture.”

For the broader digital economy, here are the real “unresolved issues:”

  • Governance of WHAT? (what specific governance problem requires coordinated collective action?)
  • Governance HOW? (what gives the institution leverage to shape the behavior of technical systems and the people who use and run them?)
  • Governance by WHOM? (what representational structures?)

As an example of the WHAT, HOW and WHOM, consider the case of APNIC or RIPE, two of the premier Internet governance institutions. Governance of WHAT in this case means Internet protocol addresses. IP addresses require coordinated global governance to maintain the uniqueness of address assignments. APNIC and RIPE do not, however, attempt to government what organizations do with their IP address blocks; they limit governance to the minimum necessary to fulfill that unique identifier function. A significant part of the governance function is performed by the market: the market for ISP services, for IP address transfers, for data centers, etc.

As for the HOW of governance, these two RIRs gain leverage by running a registry of address assignments that serves as a kind of title for appropriate ownership or stewardship of an address block. If you’re not in the registry, other ISPs will not route to you. As fo WHOM, these two RIRs are membership organizations. Sure, anyone can participate in their deliberations, but voting rights require membership and membership is restricted to organizational entities who have been awarded IP addresses.

These organizations are “multistakeholder” only in the sense that they involve governance by non-state actors.

Now consider other, broader issues in the digital economy, like “AI governance” or governance of online content. Despite all the text that has been written about AI governance, it is hard to define WHAT these people want to “govern.” Is it the LLMs, the semiconductors, the specific applications, the data, the networks connecting the data to the models, the training process, the transparency of the algorithms, etc.?  To truly “govern” AI one must have leverage over all these things, and currently no single governance authority does. Likewise, HOW would one obtain the leverage to govern all these applications worldwide? Would one have to get a license from a global MS institution to write code for a LLM? Would a central authority review all software applications? Who would be in a position to control the many data sources generated by online activity? And as for WHOM, what kind of representative structure would be optimal for this governance task. Would “anyone” be qualified to review and rule on the suitability of AI algorithms? Would national governments?

It is apparent that the MS advocates do not understand basic things about institution-building and policy making processes. Governance institutions are a product of social conflict. Contentious social actors negotiate rules for engagement around specific policy problems. Each actor tries to get rules that favor their interests. Their motivation for participation is precisely that they have something to gain or lose from the outcome. Institutions are structured through conflict and contention over benefits. How stakeholders are represented is critical. We do not say anything meaningful about representational issues by saying “everybody should be represented everywhere all at once.” That’s just a cute way of avoiding the issue.

NETMundial 2014 was a wonderful moment in Internet governance. NETMundial 2024 made it clear that that moment is over. May it rest in peace.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.