Tiktok and the Fediverse

Today, the United States House of Representatives voted to require the Chinese company Bytedance to sell its stake in the popular service Tiktok. If the company does not comply, the bill would ban the use of Tiktok in the US. The bill still needs to pass the Senate and get delivered to President Biden’s desk, but there are a lot of really interesting parts of this decision that I want to unpack.

Foreign companies own shares in all kinds of entities in the United States. There are some interesting restrictions on domestic transportation (listen all of y’all: it’s cabotage), financial services, atomic energy, and real estate. All of them have pretty strong justifications in terms of domestic security — that some level of domestic independence would be lost if all our railroads were owned by Chilean investors, for example.

But we don’t have restrictions on selling foreign-owned magazines or newspapers or movies, or serving foreign-owned Web sites, in the US. You can read Pravda or Al-Jazeera or China Daily on an iPad in the comfort of your lovely American home. It’s not universal — copyrighted information illegally shared and terrorist recruitment content are often blocked or the domains are just plain taken over. But for the most part, the US is pretty OK with you reading or watching content from other countries, even if it has a strong editorial slant against the US government’s current policies.

So, why is Tiktok different here? Realistically, most American viewers aren’t watching Chinese-made content on their Tiktok apps. They’re watching short videos made in the US, Canada, the UK, Australia or other, y’know, “western” countries. But the videos are curated by an algorithm provided by Tiktok, which if it is partially-owned by Bytedance, is somehow entangled with China’s interests.

The House bill, then, is an acknowledgment that algorithmic curation of feeds is a powerful feature that can have a major influence on individuals and society. It at least makes the point that allowing a foreign company, under its own government’s influence, to have some level of control of the algorithm, is a potential danger for domestic security.

This raises a few really important questions. First, for everyone outside of America, it raises the question of algorithmic feeds created outside their own countries. Is the Internet a nice, friendly post-national free-trade free-speech zone, or is domestic control of this technology important in France, Guyana and New Zealand, too?

In addition, for Americans, it should probably give us pause in thinking about our domestically-controlled networks. Does Congress think that algorithmic feeds under domestic control are equally powerful? If so, who’s making the decisions on how they’re used, and what connection is there between different factions who want to influence opinions or behaviour through curation?

I think the Fediverse provides some interesting answers here. The Fediverse is a federated network of social networks — a social internet or social web — connected by the ActivityPub social standard. The networks can be owned by all kinds of different entities — governments, private companies, community groups or individuals. Each network has its own local control mechanisms, but users on one network can follow, reply to, and otherwise communicate with people on any other network. Content created on one network gets published out to all the other networks depending on how many followers are there. Mastodon is a common example of software on the Fediverse, but new platforms are joining all the time.

For the international question, countries can consider implementing domestic social networks that federate with ones in other countries. This allows content to be received and sent across borders, but algorithmic feeds to be managed locally. If there is concern about algorithms being manipulated by foreign governments, using fediverse-enabled domestic software prevents the problem.

Within a country, it raises one potential solution for people concerned about the influence of algorithmic feeds, namely, running social network services under your own control, and following users from other networks. The home feed you read can be curated by an algorithm built into the server, or built into the client, or you can even leave it uncurated — just in chronological order. The Fediverse allows pushing the control of algorithmic feeds closer to users, who can make their own decisions about how content is prioritized.

Federation provides the possibility of some interesting changes in the locus of control in social networks. If we are starting to acknowledge how powerful this curational control is, we should start structuring our social network infrastructure to allow experimentation in whose hands are on the levers.

9 thoughts on “Tiktok and the Fediverse

  1. @evanprodromou

    > But we don’t have restrictions on selling foreign-owned magazines or newspapers or movies

    There should be. It should also be limited so that no large company can own or influence journalism and art in a way to make them profit. (e.g Amazon being allowed to own Washington Post is insane)

  2. @evanprodromou This is something I’d really like to tinker with using machine learning, an algorithmic feed “closer to users” based on the firehoses we can already individually & privately get out of a Mastodon streaming API or (more directly) an ActivityPub inbox.

    One of the things that makes TikTok great (or I guess harmfully addictive) is that their algorithm really seems to zero in on interests.

    I wonder if there’s a way to do a “federated” algorithm, where you could have your own personal ML algorithm agent “follow” and collaborate with other folks’ agents (with permission) to get socially augmented recommendations?

    (I mean, there probably is… I’d like to figure out how)

    1. So, I gave a talk about algorithmic feeds at the ActivityPub Conference in 2020:

      https://conf.tube/w/c79457a9-aae5-47dd-8731-617e6b09fe06

      It wasn’t supposed to be about algorithmic feeds, but I bent my assigned topic to my own purposes.

      Essentially, I see a lot of power in algorithms for optimizing where you spend your attention. But I think they should be used to deepen and strengthen your personal relationships, rather than only being about getting you to stay on the site and click ads. Algorithms that favour interaction, connection, strengthening social ties and making new ones seem really relevant.

      Consent required, of course, but who doesn’t need help being a better friend?

  3. I should probably be clear that I haven’t mistaken the original intent of this bill. It’s about 80% China-baiting, another 12% reactionary punishment of young people in America, 8% a money grab by potential “investors” in Tiktok, and really only 2% protectionist Internet policy.

    But I think the question of control of algorithmic feeds is now on the table, and I think it’s as good a time as any to discuss them.

Leave a comment