Note that flatpaks are using systemd’s sd_id128_get_machine_app_specific API, which generates an app-specific id that’s derived from the original machine-id.

The app-specific id will be unique per-app so it won’t be shared between different apps / flatpaks. The apps can’t know the original machine-id, but the id they get would still be the same on every restart of the app and in theory does not prevent different sessions within the same app to be fingerprinted as coming from the same user.

Yeah… according to the stats in this post, about 1.9 million people are behind bars.

11 billion a year between 1.9 million prisioners is about 6k per year per prisoner.

So that’s about $500 a month per prisoner. Which is not as big of a number… and according to the article, the prisoners still get paid a fraction of it… it’s just that 80% is used “for room and board, court costs, restitution and other fees, including the construction and maintenance of prisons”.

I always saw I2P as a more modern and distributed onion-routing alternative to Tor.

The thing is that people are used to making use of Tor in different ways than the way they use I2P, but you can also have outproxies (ie. exit nodes/relays) in I2P the same way as in Tor… and you can also host a service inside the Tor network without relying on an exit node, like in I2P. It’s just that people only seem to want to host exit nodes for Tor and not so much for I2P, this led to internal communications in I2P being more common (which is a good thing), whereas in Tor it’s common to use it for anonymous access to the clearnet (which strains the network and causes chokepoints, specially with big downloads or torrent sharing). That’s just a matter of usage, not capability.