Thread by @anildash

This is the text of the new bipartisan bill aimed at requiring large-scale (1 million+ users, $50M+ revenues) social platforms to provide views of their streams that aren’t sorted by an opaque algorithm. It’s short, and fairly readable. documentcloud.org/documents/2110…

This piece summarizes the sponsors and the intent fairly well. (Note: it uses “algorithm” here in the vernacular sense, not the technical one, but the bill itself is a bit more considered about that.) axios.com/algorithm-bill…

I think @johnthune and his cosponsors mean well with this but it’s ill-considered in its implementation, mostly for being too obviously inspired by Twitter’s optional chronological view of the timeline. They clearly saw that option & thought “let’s require more of that!”

Now generally, for tech regulation, it’s good when policy starts from something that actually exists. But in this case, it solves the problem too narrowly; this might be useful in social *media* platforms (Twitter, YouTube, TikTok) but not in social *network* platforms (FB, etc.)

There are also a lot of inadvertent consequences from “turning off the algorithm”, only some of which are addressed in the bill. The “algorithm” also helps limit amplification of lots of hateful, abusive & dangerous content — even on platforms where we still see too much of it.

More broadly, this class of solutions to tech tend to inadvertently *amplify* the entrenchment of big players, rather than encouraging healthier platforms to thrive. A better approach might be requiring open feeds & the ability to plug in alternate sorting/ranking algorithms.

This has the advantage of giving users more control over data, helping new players get into the marketplace, and avoids the most likely big company response— letting all kinds of awful content flow, and saying “the law won’t let us hide hate anymore!” (You know they’d do this.)

It’s also worth noting that there’s a reading of the bill in which it could also require something nonsensical like a view of Google search results which isn’t parsed by an algorithm. What would that even mean? Any search result follows some algorithm.