The cost of a follow button

A follow button is the cheapest UI element a social product can ship. Two lines of HTML, a single database join, instant behavior. It's become such a default that most teams don't stop to ask whether they need one. The question they're answering isn't "should we have follows" but "should we have the default or something weird."

The default isn't free. A follow button creates a graph, and the graph creates obligations. Every person you follow is a decision that compounds — next week you see their posts, next month you feel guilty when they produce something and you didn't read it, next year you realize the accounts you followed don't reflect who you are anymore and the unfollow process feels socially loaded. None of this is on the spec sheet when you ship the button, but all of it is implicit in shipping it.

For the product team, the follow graph creates its own obligations. Once users have follow relationships, you can't ship feed changes without breaking them. You can't A/B test the ranking too hard because someone will feel their followed content got buried. You can't add algorithmic intelligence without getting accused of suppressing voices. The follow graph becomes a political contract between the platform and its users, and the contract restricts what the product can become.

txtfeed doesn't have a follow button. There are no accounts to follow, no subreddits to subscribe to, no writers to favorite. The feed adapts from your votes, not from relationships you entered into under false pretenses ("I liked one thing this person wrote, now I'm in a year-long contract").

The cost of this decision is real: users can't say "show me more of Ben Thompson." But they can upvote a Stratechery piece and get more Stratechery-quality writing in their feed, which turns out to be what they actually wanted. The signal is "more like this," not "more from this specific person," and the algorithmic version is closer to the underlying intent.

The lesson: defaults have costs that don't show up in the first quarter's metrics. The follow button is cheap to build, expensive to live with, and nearly impossible to remove once shipped. The best time to decide you don't want one is before you ever add it.