The Three-Legged Stool: A Manifesto for a Smaller, Denser Internet

The Initiative for Digital Public Infrastructure's vision for a public sphere has three legs:

• many different platforms with a wide variety of scales and purposes;
• users navigating with a loyal client that aggregates, cross-posts, and curates; 
• all supported by cross-cutting services rooted in interoperable data

"we propose a pluriverse consisting of existing platforms alongside a flourishing ecosystem of Very Small Online Platforms (VSOPs) that serve conversations and communities that are poorly served by today’s digital public sphere" - aka cozyweb? - and "discuss what it takes to develop a new VSOP, using our work on Freq... as a case study."

Then "we sketch out a “loyal client” for navigating the digital public sphere" - essentially software that allows you to manage your conversations across everywhere, which means overcoming "challenges related to privacy, adoption, and usability".

And then "we introduce the “friendly neighborhood algorithm store”,,, marketplace that VSOPs and loyal clients can rely on for curation and Trust & Safety, tools which no single VSOP or loyal client could develop on their own, and which large platforms have developed over decades with significant resources. These include recommender systems, spam detection, anti-abuse tools, and powerful filters for CSAM and terrorist content" - I'm hoping this is where the business model kicks in, as it's exactly what I've been arguing for since my manifesto posts of 2022.

2. Gardening in the Pluriverse

They use the gardening metaphor! I feel seen. "We need a fertile, flexible ecosystem of Very Small Online Platforms which serve different purposes than the existing VLOPs [Very Large Online Platforms]".

• "It is not realistic to expect people to abandon these platforms en masse" - a healthier digital public sphere will come from combining answers like "using regulation to create a better Facebook and a better Twitter" and "decamping for decentralized alternatives in the Fediverse" and the solutions set out here: "neither alone, nor both in conjunction, are the full answer... there are no quick fixes for social media"
• "people are so dissatisfied with the platforms" that the opportunity now exists
• "we are inspired by an example from ecology: the Miyawaki method... Reintroducing native biodiversity" one hectare at a time: "a new home to species that struggled to gain a foothold in ... monoculture ... they grow outwards too, spreading that nurtured biodiversity."
• On civic social media: "We need spaces for discussing local issues which are closer to the experience of a newspaper’s editorial page than a shouting match at a local bar... Smalltown hosts local civic discussions. It is closely moderated and the discussion is scaffolded. Moderators are active"
• Launching a VSOP is hard: it must be as simple as "start a blog on WordPress or start a webstore with Shopify"
• "many VSOPs will likely receive most of their revenue from the communities they serve, similar to how a local newspaper or nonprofit is funded", so community participation in governance is likely essential
• while technical costs are low, revenue is probably essential to cover moderation: models include "subscriptions, advertising, donations, licensing, foundation support, and public funding"

[from the conclusion:] "helping to manage a community social network may become a form of civic education in the way Robert Putnam speculated joining the Elks Lodge or managing a local bowling league once was."

3. A Loyal Client

"a tool that aggregates, organizes, and posts to the various platforms people are a part of" - an old dream dealt "a near fatal blow ... [users] moved primarily to the mobile phone", where the platforms' clients dominate.

But it's nevertheless essential to "moving some power from platforms into the hands of individuals... giving users more control over where and how they participate, and what they see."

They're building one: Gobo: https://gobo.social/, with the Spring 2023 version planned to have read/post across "Twitter, Mastodon, and Reddit... we hope to include platforms like Facebook, Instagram, and LinkedIn". Then it gets interesting:

• "“lenses” for your feed... to give people control over the algorithms that sort their feeds" from one or many platforms. "Lenses can filter out content you want to avoid, can introduce new content you may be interested in, and can rank content according to your preferences." - isn't Bluesky allowing this? it's not mentioned once.
• "a third-party ecosystem of algorithm providers" - exactly what I waswriting about in 2021: "Only by creating a decentralised, open-source ecosystem can we develop cooperatively-owned AI engines to benefit society as a whole."
• This will require "an open standard for developing and integrating third-party algorithms" which users can audit, " inspecting the decisions an algorithm made when constructing their feed".
• Of course, user data capture will be minimised and consent always asked.

The article then explores the "privacy, adoption, usability, and legal challenges", including turning to adversarial interoperability and regulation (citing the EU's DMA) when major platforms do not cooperate.

4. The Friendly Neighborhood Algorithm Store

The algorithms used by the loyal client have to come from somewhere: a marketplace is essential, particularly as "VSOPs typically lack the resources and experience necessary to develop comprehensive tools on their own" for (incomplete list):

• "Spam detection
• CSAM [Child Sexual Abuse Material] detection and reporting
• Mitigating abuse and harassment
• Detecting mis- and disinformation
• Personalized “For You” recommendations or network highlights
• Categorization tools  that feature or filter content by topics such as sports, local, comedy, news, politics, and so on

Local legal environments may also require:

• Copyright fingerprinting and DMCA takedown
• Identification of violent extremist content"

Minimum standards for algorithms:

• "Tuneable... meaningful settings [for using] ... the algorithm in different ways.
• Auditable: you should be able to throw arbitrary content at the algorithm and see how it responds... investigate the data underlying the algorithm...
• Combinable... able to work across different source platforms
• Understandable"

What's the business model?

• "One idea is revenue sharing... platforms pay a commission to third party filters", which will probably require regulation. "A loyal client could use the same model... [but] is unlikely to be able to sustainably fund a large number of providers.
• "third party algorithm providers could charge users directly", like Block Party. "We see real promise in charging users directly."
• "public funding could support ... prosocial algorithms" - I see huge Ministry of Truth issues here.

(aside: Massively reminded of my 2018 post: We need a credibility indicator marketplace to fight disinformation).

Conclusion

This can only be built through "thousands of experiments. We are committed to building open source systems as proofs of concept... [and] to back away from our experiments as others find better ways... organizations committed to the public sphere, including libraries, newspapers, public broadcasters, cultural institutions, local governments and others, have a role to play ... [these] are public goods and should be funded as such [at least partly] from public coffers."