Stephen Downes

Knowledge, Learning, Community

Half an Hour, Dec 21, 2022

Image: Heise Magazine


Today Alek Tarkowski launched a 24-part Twitter thread into the digisphere expanding on his views about participatory governance and fediverse. It builds on his article Priorities to Make the Fediverse Sustainable, which I commented on last week, as well as some work he and his collegagues have done with the Open Future Foundation, including The Paradox of Open.

The latter resource is a good place to begin. The paradox of open is essentially this:

"The ideas of open access and free reuse of information goods continue to be some of the most powerful challenges to the exclusive control by corporations and states over information goods.Yet making such resources open also exposes them to the imbalances of power that shape these societies – and in the worst cases serves to strengthen these imbalances."

I agree with that assessment. Indeed, that recognition has shaped some of my most significant positions on open access, for example, my defense of the use of CC-non-commercial licenses, to defend us all against the excesses of the commercial occupation and enclosure of open content.

But let's also note that I find CC licenses of any sort to be a small and mostly band-aid solution at best. They entrench the idea that the application of copyright law is and ought to be the starting point for all instances of what we might call 'content', which is to me an intrusion of a commercial ethos into a non-commercial space. As well, the law is a slim defense against large corporations who have demonstrated the ability to either ignore or shape the law to their own benefit.

Thus, while resisting the idea of what people call tech solutionism, I nonetheless think that issues such as this benefit greatly from a practical application of the dictum 'code not law'. As Tarkowski says, "Coded functionalities provide greater gains for creators and users than legal tools do, while the right to remix has been secured by means other than flexible licensing – for better or worse." Railings do more to protect people than regulations. At the same time, I recognize that the dictum is not absolute. There are limits to what design can do, and even if there weren't, nobody wants to live in a techno-determinist hellscape.

In response to the paradox of open, Tarkowski and colleagues propose "a new understanding of open", with two objectives: "to think about how it can harden its core concepts against abuse and unintended externalities," and "to look beyond the core concern of opening up informational resources and develop a better understanding of how open interfaces with other concerns like data protection and distributive justice." There is merit to these proposals, but as always, it depends on how you define your terms.

As I read it, the core concept they support is something like "shared digital public spaces that can compete with the dominant, corporate walled gardens that harm the internet and its users." Specifically,

"Open should not just mean releasing resources into a digital void; the term should stand for managing the use of resources in a way that maximizes public benefit while avoiding harm, a balance between openness of resources and preservation of privacy (and other fundamental rights)."

On this reading, 'open' is a part of a wider social justice movement, and should be understood within that context, and our understanding and promotion of it subsumed under it. "Ultimately," writes Tarkowski, "we need a shared mission to create digital public spaces."

Do we?

Like Tarkowski, my view of openness is that it is a part of a wider objective. But on my view, social justice isn't the foundation of that objective, it is the outcome of it. On my account, openness is one of four core principles of network semantics, the others being diversity, autonomy, and interactivity. I offer a short presentation of it here. These, along with some structural or non-semantic conditions, are the basis for the sustaining and growth of a responsive and resilient network, in other words, the basis for any functioning at all.

A key tenet of this approach - one that defines a cooperative definition of organization - is that community is defined by association, not commonality. Shared mechanisms for decision-making, not shared decisions. In other words (elided, for the sake of this discussion), society is not based on a shared mission or shared anything, with the exception of the minimal conditions required to support effective networking.

Now these two paragraphs aren't intended to convince the reader on their own. The intent here is to suggest that there is an alternative account to the one that founds society on shared principles of social justice, and to outline what it looks like. And I think such an account is necessary, because I don't think we're every going to reach agreement globally as to what social justice looks like, as I argue extensively in my work on ethics and analytics.

So I think two things follow from this. The first is that Tarkowski's view of society and my own are going to align pretty closely. Despite having different views about how they're justified, we both believe that social justice is important. Neither of use would like to see society descend into a dog-eat-dog world run red with capitalist fervour. We believe society is capable of something better than that. But second, because of the differences about the origin of social justice, we're going to tell quite different stories about how to get there.

So, on to the fediverse. In 'Priorities to Make the Fediverse Sustainable' Tarkowski says that Mastodon faces two major limitations: first, "the commons-based peer production model on which it is built has clear limits to growth"; and second, "lack of participatory governance. The Fediverse is ruled today through code, by a narrow group of its maintainers."

The 'commons-based peer production' (CBPP) model is, as Wikipedia tells us, is a term coined by Yochai Benkler, "in which large numbers of people work cooperatively." This model is common to open source software (OSS) projects, where a code-sharing and version control platform such as GitHub is used to enable people to contribute to a single code-base. As noted, there is not typically hierarchical control over such products and not necessarily a common vision. Though OSS is depicted here as an economic model, it is subject to much wider incentives, and these vary with the contributor, each of whom, as the saying goes, contributes to "scratch their own itch."

Does such a model suffer from clear limits to growth? It depends on the sort of growth the author has in mind. We can distinguish (a bit arbitrarily) between two types of growth: growth in sophistication, and growth by replication. The first refers to the idea that the software becomes more and more sophisticated over time. The idea that this is necessary to accommodate more and more users. For example, the software that supports a social network service of a billion people can't be a simple website and database; it needs complicated load balancers, distributed databases, authentication, content review and moderation, and much, much more. Such software quickly exceeds the ability of CBPP to support it.

Growth by replication, however, is something quite different. Rather than a simple thing becoming more complicated, growth occurs as a simple thing becomes more numerous. Consider, for example, a meme - a digital photograph or video, say, that becomes widely popular, such as Jenna Ortega's dance. The meme (as opposed to the actual video) spreads by being replicated - thousands or millions of people do their own version of the dance and share it however they wish with their own network of friends. The dance doesn't need to become more complicated (indeed, it's a problem if it's too complicated). But it still scales.

There is probably a lot more room to discuss the nuances of scale and replication. But what's important here is that only for some models of software and software use are there limits to scale on CBPP. Simple software that replicates well - such as, say, web servers or RSS specifications - can be maintained by a small number of people with limited resources. That, typically, has been the role filled by OSS, while commercial developers scale with more and more sophisticated applications.

The question of whether we can support the fediverse (and specifically, ActivityPub) to scale through replication rather than sophistication is at this point an open one. ActivityPub is a complex standard that hasn't been tested at scale. There are signs of problems emerging. But the approach here is to fix ActivityPub so that it does scale by replication, rather than by requiring more and more sophisticated software, in my opinion.

It might be asked, does the fediverse even need to grow. I don't think it does, particularly, though I think this debate is a bit of a side-issue. But let's address it. Tarkowski writes, "So why does it need to grow further? Because millions more people need access to healthy, just, sustainable, user-friendly communication tools."

Not only is the audacity of such reasoning breath-taking (I mean, imagine placing that obligation onto the shoulders of a young software programmer just because he happened to write some useful code), it is also unsound. Yes, it is true that Mastodon is an X. And it is true that millions of people need an X. But it doesn't follow that they need Mastodon. The need any of numerous things that could be an X. (By analogy: a banana is food, millions of people need food, but it doesn't follow that millions of people need a banana (good thing, because bananas may be going extinct)).

Tarkowski writes a bit about Mastodon creator Eugen Rochko, his finances, and Mastodon's finances. "It hints at the precarity of the setup," he says. One would think this could be addressed quite simply; pay Rochko (and maybe some colleagues) a meaningful salary to forget about hosting mastodon.social and to develop and maintain Mastodon as a simple, core application supporting the fediverse, and allow to to scale by replication. If it's such a moral imperative, as Tarkowski suggests, perhaps some money could be redirected from some less essential institutions.

But it seems clear, on reading the article, that Tarkowski is looking at scaling by sophistication. He quotes Jeff Jarvis with approval expressing the hope that "individuals, institutions and entrepreneurs will experiment, innovate and offer other, new services and functionalities" while complaining that "there are surprisingly few innovations in the Fediverse, a space that in principle should be generative. For example, it is surprising that there have been no prominent attempts to build decentralized, responsible algorithmic tools that would manage content for and by users."

That's a bit much to ask, I think. I mean, put yourself in the position of Eugen Rochko. You've written a decent application that implements a new W3C standard and you've kept at it with a smallish community for a number of years and have managed to self-fund your way to some sort of income. Then someone comes along and says you have this obligation to society to scale it up and serve the millions of people who need access to healthy, just, sustainable, user-friendly communication tools. You look at him. How does this become my responsibility?" you ask. "Why not," you might say, "just fix Twitter?"

You see, this to my mind is where the ethos of social justice as a shared mission becomes problematic. Because somehow, it allows you to make your mission someone else's mission, and you've done it without even consulting him. No, that's not how it works. That's not how society works.

This is where the argument shifts. Through the reasoning above, Mastodon has become a de facto public service. And all of a sudden, Rochko's small project has become problematic. "As the lead maintainer of the code, Rochko makes all the decisions. In line with a typical open source model, where commitment determines stakes in the governance process, the shape of the service is decided by one person, a self-described 'benevolent dictator.'" This is unsatisfactory because "this model is far from the ideal expected of democratically managed and governed infrastructures." For example, if the world (somehow) demands quote-retoots, the Rochko should deliver quote-retoots, no matter how much he personally disagrees with them.

I can't emphasize how much this line of reasoning conflicts with my ideal of a democratically managed and governed society.Somehow, though some sleight of hand, Rochko is now working for society (as defined by some sort of governance) because society has decided it needs what Rochko has developed. And even though he is offering it to society for free, that's not enough. He must do whatever it is that society deems it needs from the software (which in the current case, seems to be to make it more sophisticated, not more replicable, but I digress).

Why is it not enough to simply give people the code and let them add their own features? "It does not resolve the fundamental inequality inherent in open source systems between those who code, and those who cannot or do not." And it's not just whether one can edit the code, Tarkowski writes. The same problem exists when you tell people they can run their own server. It's not an easy thing to do! The same with the ActivityPub specification itself - many people are not going to be able to understand, let alone contribute to, the W3C project. As Tarkowski writes,

"The crucial divide, just as with the Mastodon code, is between the programming haves and have-nots, the coders and non-coders. The openness of the ecosystem means that it is in principle a lot more democratic, as it creates meaningful possibilities to shape it by contributing code. But this ability is not available to the majority of users, leading to a sort of caste society, built on top of an open source infrastructure. There is no realistic scenario in which all users learn to code – therefore participatory governance approaches, which take control of the code away from the hands of the coders, and into collective decision-making processes, is the only way forward."

All of this is true. There is a divide between people who can code and people who can't. But I still imagine myself in Eugen Rochko's shoes asking "how did this become my problem?" I mean, maybe he agrees with Tarkowski's shared mission or maybe he doesn't, but even if he does, isn't this society's problem, not Mastodon's?

Again, I can count at the drop of a hat dozens of cases where software developers can and should be held to greater account for social justice than Mastodon, a project that by all accounts does a much better job at respecting social justice principles than most. Why, indeed, does any of this become a responsibility for OSS at all? If we want to go after what's broken in our society as a result of the coding haves- and have-nots, why not start with problem-child Twitter and its evil step-sibling Meta? Why is this a problem for the fediverse to fix?

And I think we both know the answer to that question: it's because the others are irretrievably broken, and the fediverse is the best and possibly last hope for software that actually does work for people.

If so, then we'd better not muck it up then.

Let's now, finally, turn to that Twitter thread (which seems like an odd place to conduct a discussion about the future of Mastodon, but again I digress).

The first 15 tweets essentially restate the argument as I've described it here. Tweet number 16 cites my response to most recent article, and number 17 begins Tarkowski's response: "Since there is no realistic scenario where everyone learns to code (thus distributing the capacity to fully participate in the network), we need a pact between the coders and the code-nots: participatory governance."

I could cynically reply that you need a pact; the coders don't need a pact. But as always, it's a little more complex than that.

First of all, most people who are coders belong to an already disempowered class: the non-rich. Most coders in the world are to some degree or another compelled to direct their code in certain directions by the dictates of capitalist markets and empowered employers. They are often unrepresented by any union, often underpaid, and have poor working conditions. This is true even though coders have benefited from generally better education and social standing than many others in society. They are not the poorest of the poor. But to depict them as some sort of power-broker in this discussion is a non-starter.

Second, even if there is inequality of power between coders and non-coders, it does not follow that a pact is necessary. Other solutions may well apply. I could draw from the wealthy person's playbook at say that no pact is needed because we can depend on charity and patronage. Or, alternatively, I could suggest that appropriate coding behaviours could be ensured through regulation and legislation (of course, regulation and legislation may also affect rich people, so we have to be very careful here).

Third, and maybe even more to the point, maybe it's the non-coders who should change. Let's read the article in Noema, recommended in tweet number 18 of the thread:

As users migrate to the fediverse, they often bring the old expectations of scalable social media. We are used to someone else being in charge and taking care of the problems that might arise, and we are used to complaining when they mess up. Users also bring the same racism, sexism and bad behavior to Mastodon that have arisen in other kinds of online spaces. But in the fediverse, we cannot simply rely on a company's trust and safety department to take care of problems for us.

These aren't coding problems; they are social problems we seem to be expecting code (or coders) to solve. If there's a pact needed, maybe it's a need for a pact among users themselves, not between coders and non-coders. Indeed, the more I look at it, the more I begin to see the coder-non-coder distinction as a kind of a red herring.

Indeed, contrary to offering a suggestion for a pact between coders and non-coders, as Tarkowski seems to suggest, the article actually embraces a notion of subsidiarity, that is, the idea "that social and political issues should be dealt with at the most immediate or local level that is consistent with their resolution." This means addressing problems at the instance level, where possible. The article describes one instance where "we use a separate platform, Loomio, for our deliberation and decision-making."

In tweet 19, Tarkowski returns to the idea of redefining 'open'. "We also need to think more in terms communities and relations, not just code and servers. (Jeni Tennison) makes the point that 'Openness is about relationships and building community'. I think it would be nice if people thought about community more. But 'openness' and 'community' are two very different things, and one does not reduce to the other. (Indeed, to be pedantic, they are not even of the same category. 'Open' is an adjective, while 'community' is a noun. You can't have 'an open', but you can have 'a community'. You can, note, have an 'open community', which is mostly what Tennison is talking about. But that's not the same as open source, open pedagogy, or open access. But, crucially, it's not up to open source software developers to create open communities.)

In tweet 20 Tarkowski says, "Being open and decentralized, #Fediverse offers great opportunity to do that (ie., think in terms of communities and relations). There are many examples to borrow from - paradoxically, the interesting ones often happen in #web3. It's worth paying more attention to @gitcoin @metagov_project or @RadxChange." I agree, which is why web3 has been an interest of mine for some time (and has directly led to the idea of thinking about community as being defined, not by sameness, but rather, by agreement on a consensus-building model). This is one of the many things the fediverse offers us the opportunity to do. It's also a great opportunity to share cat pictures. But I'm not asking that the whole fediverse be repurposed to support my desire to share cat pictures.

The three recommendations Tarkowski makes in tweet 21 are from his paper, and I'll produce them in full here:

"Launch a participatory project to define a shared mission for building the digital public space on the basis of the Fediverse. There are multiple methodologies to do so, such as Decidim, deployed in urban settings; or Pol.is, participatory experiments conducted by the Taiwanese government. The Conference on the Future of Europe, a EU wide deliberative process, provided surprisingly good results as well."

Yes, there are many tools that help people make decisions, and I have no particular issue with a group of people getting together, using such a tool, and deciding how they would like to use the fediverse to build a digital public space.

My caution is that such a process should not be taken as defining the digital public space, because people will have many different idea of how to do that. Nor should it be taken that such a process would in any way be binding on the fediverse, or any of the many software projects that make up the fediverse.

Given those conditions, it's hard to see how such a project would produce a useful result, given that the fediverse is already a digital public space.

"Secure greater involvement of public institutions. Dan Hon proposed for organizations to set up their own Mastodon instances and serve as verified, public interest driven, trusted nodes in the network. Public institutions should also bring in resources needed to develop the network – for example invest in public benefit algorithmic solutions for the Fediverse, or a broader range of services (Peer Tube is a good example of a publicly funded Fediverse service)."

This recommendation actually speaks against the stated desires of many people in the fediverse, those would would like the digital public space to be a place for people, not companies and institutions. That said, the time has already passed for that ever to be the case, and a large number of companies and institutions are already creating fediverse accounts and hosting Mastodon instances.

Not only is this recommendation not a good idea, it contains elements that might actually destroy the fediverse by creating special 'trusted nodes'. The real danger here (as has been stated elsewhere) is that these institutions, once having invested in the fediverse, may decide that they want it to contain some 'institution-only' features (like, say, institution-based user authentication, or some such thing), build it as an 'add-on' to the fediverse, and then refuse to federate with instances that don't support it, creating an institution-only version of the fediverse. This, basically, is what Google and other companies did to RSS.

"Build a stronger social and institutional layer. There are many experts in community- and network-building who should be interested in working on such a project. There are also great prior examples of large-scale, commons-based peer production (the BigScience and BigCode projects are the latest examples on my radar)."

This is an extension of the previous recommendation, except that embeds the idea of institutionalization. It takes the worst case scenario and makes it worse.

It feels to me like the sum total of Tarkowski's recommendations are to (1) hand control of the fediverse to the interests of the rich and powerful (that is, those who currently run companies and institutions), and (2) try to negotiate some sort of governance process to protect use from them (or them from us - it's not entirely clear).

In tweet 22 Tarkowski points to the Data Commons Primer as an example of how to build this. The whole question of data is an important one I can't address here. But again, I would point to a deep and important distinction between a commons and a cooperative.

The last two tweets close off the discussion with some hopeful words for the future.

Let's come back to that idea of "if he wants to govern a social network, let him build it," with which he says he deeply disagrees. And let me ask, isn't he doing exactly what I recommend with that social and institutional layer? I mean, it's not like he's asking Eugen Rochko or any of us to build it, is he?

It isn't - and never was - a division between coders and non-coders. When non-coders want something coded, they do what they've always done: they form a company or an organization or a foundation, raise some money, and hire some coders. Maybe they can get some volunteers, but that's not usually how it works. And creating some sort of 'framework' that will govern the fediverse by imposing requirements on people who are essentially doing their own thing with their own time and their own money is a non-starter.

If we don't want a purely capitalist ownership of the fediverse (which is a real danger if we get a special social and institutional layer) then we lobby government (not software developers) to pay for it and manage it. That's kind of what has been done already with the W3C and the committee that designed the ActivityPub standard in the first place. It works, because nobody can own that standard. And the people who work on W3C are mostly paid by companies, contracts or governments.

Mentions

- , Dec 21, 2022
, - Priorities to Make the Fediverse Sustainable, Dec 21, 2022
, - Open Future – Open Future Foundation, Dec 21, 2022
, - The Paradox of Open, Dec 21, 2022
, - , Dec 21, 2022
, - , Dec 21, 2022
, - , Dec 21, 2022
, - , Dec 21, 2022
, - Community , Dec 21, 2022
, - Commons-based peer production - Wikipedia, Dec 21, 2022
, - Wednesday Addams | Dance Scene | Netflix - YouTube, Dec 21, 2022
, - Some thoughts about ActivityPub ยท GitHub, Dec 21, 2022
, - Why bananas as we know them might go extinct (again) | CNN, Dec 21, 2022
, - , Dec 21, 2022
, - , Dec 21, 2022
, - Subsidiarity - Wikipedia, Dec 21, 2022
, - Loomio, Dec 21, 2022
, - , Dec 21, 2022
, - , Dec 21, 2022
, - Data Literacy, Dec 21, 2022
, - Participatory Governance and the Fediverse, Dec 21, 2022


Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Creative Commons License.

Copyright 2022
Last Updated: Dec 21, 2022 7:49 p.m.