Thank you! That clarifies :) I'm also glad for all the innovations Mullvad has invented/supported/etc in the VPN space -- anonymous account numbers, multi-serve

Thank you! That clarifies :) I'm also glad for all the innovations Mullvad has invented/supported/etc in the VPN space -- anonymous account numbers, multi-server SOCKS proxies, Wireguard over TCP, post-quantum Wireguard, stboot, open APIs, the list goes on. It feels like VPN for apps is very different than a VPN for browsing. While in both cases I want my traffic to be mixed in with a lot of other people's traffic (so service provider dealing with complaints about neighbors is part of the value proposition), browsing use case is tied to IP reputation (so don't want someone to run a Tor exit on the same IP), whereas the app use case is much less IP reputation-sensitive but definitely benefits from port forwarding (e.g. to anonymously run nodes that powers distributed infrastructure like crypto). I'd definitely pay premium, with longer commitments up front for "this server might be useless for browsing but run all your anonymous crypto nodes behind forwarded ports" type of service. Maybe if port forwarding is active only if you have 6+ months of outstanding service commitment (and you forfeit the balance if your port gets used for C&C or whatnot) is enough of a deterrent. Some VPNs are doing some traffic segregation already, e.g. having dedicated servers for P2P, though nothing exactly like this.