Self-hosted software is growing up, but still has rough edges
Immich 3.0 dropped and the response is warm but honest. Multiple commenters called it one of the most useful self-hosted tools they run. But the thread also surfaces real concerns: no end-to-end encryption, no read-only source folder support, and the general overhead of running yet another self-hosted service with backups, updates, and hardware failure risks.
Podman v6.0.0 generated similar energy. Commenters love it and think it's technically superior to Docker, but Docker's network effects remain dominant. The new networking tools (pasta, pesto) are appreciated, but the 'why isn't this more popular' question still hangs in the air.
The pattern: open source self-hosted tools are genuinely excellent now, but the gap between 'excellent' and 'widely adopted' is still filled with operational overhead and switching costs. The people who love these tools love them deeply. Everyone else stays on the managed service.
So what?
If you're building a self-hosted product or an open source alternative to a SaaS, Immich is the playbook to study. It has strong word-of-mouth and real daily utility, but it also shows where self-hosted tools lose users: encryption, ops burden, and feature gaps. Closing those gaps is what separates a project people admire from one they actually use.