We need better tools for online speech

Previous to my current job, I worked on various types of recommender systems for 4.5 years as an entrepreneur (mainly The Sample and Yakread). However, as I thought constantly about this area, I realized that recommendation is only one slice of the problem domain I'm interested in.

I came up with a term that I believe captures that domain: "tools for online speech" (TFOS). Inspired by the tools for thought community, which seeks to create computer systems that extend people's thinking and creative capacities, I think there are still lots of opportunities to create software that improves the way ideas travel over the internet.

Some examples of tools for online speech are recommender systems, search engines, social media, blogging and newsletter platforms, discussion and community software, email and RSS clients, bookmarking tools, chat apps, .... Admittedly that's a pretty broad category, but I do think it's distinct from any already common labels ("communication software" comes to mind, but would that include recommender systems or bookmarking tools?). I also think it's worth creating a new label so that we can try to form a community of people interested in improving these tools.

Why it matters

It's helpful to think in evolutionary terms. The information environment, which includes technology, social structures, and other human factors, defines a fitness function for information. The environment determines which information spreads and which doesn't.

Tools for online speech are important because each one has an effect on the information environment. One of the ways we can improve the environment is by thoughtfully designing, building, and distributing better tools. We can work towards an environment that favors well-supported arguments over memes and misinformation; that helps good ideas find an audience even if the author doesn't already have a large following; that gives influence to voices from marginalized communities and helps to combat discrimination; that helps people to learn instead of just being entertained; and so on.

We probably can't solve all our problems just by writing code. But we can strive to understand how existing social structures, human motivations and incentives, and plain-old entropy affect the information environment, and we can look for ways that better tools could help us balance things out.

Decisions in all areas—whether at an individual level like parenting or career choices, or at a global level like climate change or the decline of democracy—are affected by the information at hand. There are enormous potential benefits to improving our information environment, even incrementally. If you're interested in this domain, it might be one of the most valuable ways to spend your time.

Example: unbundling social media

To make the idea of tools for online speech a little more concrete, I'd like to share some of my personal opinions on the space. I'm not intending to include any of these opinions in the definition of TFOS, so it's fine if you disagree with my arguments or are interested in other kinds of tools. Hopefully this just provides a starting point for discussion.

As a first step, I think we should strengthen alternatives to the social media platforms. I want to make it easier for anyone (not just product managers at Facebook or Twitter) to experiment with different ways of doing things. My rough mental model is that there are four big needs—publishing, consumption, aggregation, and discussion—and there should be separate, interchangeable services for each. Then if someone has an idea for e.g. a better publishing app, they can build it and have it plug into the other existing services for consumption, aggregation, and discussion. No need to bootstrap an entire network from scratch.

This model already exists in a nascent form. For example: you can publish a newsletter with Ghost, Beehiiv, or Substack; I can subscribe to it (consumption) with Gmail, Matter, or Readwise; we can discuss your posts on Slack, Discourse, or Reddit; and we can discover more writers (aggregation) with The Sample (wink) or Refind.

It sure is fortunate that thanks to email, we have a protocol for publishing and subscribing that already has massive adoption! Email makes this scheme much more plausible. As such, I'm particularly interested in tools that are compatible with email-centric workflows.

So the task ahead is to further develop these kinds of tools, get them into the hands of more people, and make the user experience seamless. We don't have to "beat" the platforms—but the larger this unbundled ecosystem becomes, the easier it will be for people to experiment with new kinds of tools and create sustainable businesses on top of them. The experimentation that happens in the unbundled ecosystem might even provide insights that platforms can incorporate.

Building a community

When I originally wrote this article, I was working full-time on tools for online speech and trying to grow a Discord server for anyone else interested in the kinds of things I was building. I failed to make a living out of it, so I got a job and now I have very little time to spend on TFOS. I moved this website from https://tfos.co to https://obryant.dev.

I do still work on Yakread, sort-of, slowly. I'll be able to work on it more once my kids are a bit older. At some point I'd like to start donating regularly to Open Web Advocacy and Servo, which are on the layer below TFOS.

In the mean time, I always enjoy meeting other people interested in TFOS. Send me an email, especially if you have a scheme of some sort—I love schemes.