We've been in a world for 10-15 years where recommendation systems needed tons of data and tons of compute. That drove things towards a) large user bases and b) largely free-to-use, ad-supported models (it was only very expensive to produce audio and video where subscription based models worked, a la Spotify and Netflix).
The resurgence of long-form text (Substack) definitely seems to rhyme with the earlier days of blogs and forums. It does seem like people still like to write things and put them on the web. What's interesting is whether the technological paradigm has changed enough to where you could get interesting get discovery without getting mired in the slop of social media.
There are some interesting players taking another attempt at helping people explore the web by being aggregators - kagi.com for example. And I hope they make cool things. At the same time, I feel like we've walked that path a few times now and it hasn't made too different of an outcome. I wonder if there are laws in averages that mean it will become the same thing no matter intent. Even if that is wrong, I do worry we are stuck trying to make a faster horse.
The resurgence of long-form text (Substack) definitely seems to rhyme with the earlier days of blogs and forums. It does seem like people still like to write things and put them on the web. What's interesting is whether the technological paradigm has changed enough to where you could get interesting get discovery without getting mired in the slop of social media.