The user volume to support niche communities is the most obvious thing missing in Lemmy. But I have a darker view of the future. Picture LLM bots forging organic-looking conversations that result in a product recommendation. It looks like a genuine human conversation, but it’s actually an advertisement. Maybe it’s mixed with some human comments, but that may only add to the realism of the fraud.

That kind of ”advertising“ could potentially command a lot of money. And it could probably eventually infect just about any text platform. Maybe Lemmy as well someday?

You could deploy it pretty effectively in sufficiently large niche communities.