The assumption here is that there's lock-in. But there is no lock-in. Architecturally, all of the data is controlled by you and available for any other tool to ingest. This is a crucial difference between ATproto apps and other modern web services.

I mean, it really just doesn’t here. Every part of the Tangled stack is open source, self-hostable, and interoperable. There are already clients that never hit Tangled infra.

If Tangled the company made a decision you didn’t like, you could continue to use the software and keep your social graph, repositories, etc. without them. That’s simply not true with GitLab, it’s a closed ecosystem by instance. Tangled is not that.