Imo the problem here is a failure of law enforcement on the internet. IP addresses + timestamps can be tracked to a subscriber, but apparently it's so ineffective that, rather than allowing pseudonymity (only knowing your IP address) for all countries that fight digital crime (I imagine child abuse is similarly fought in most places), we instead opt to let the likes of Google and Facebook use tracking and magic algorithms to determine who's allowed to have an account, nay, identity on the internet.
Perhaps we need something that is pseudonymous but tied to an individual rather than a subscriber line, to be depseudononymised only by court order, similar to IP address now except you can actually find who did something (or was complicit at minimum, similar to money mules). We can also make it be different for every recipient, similar to how you can create any number of blockchain addresses without revealing the tie between them. It sounds super dystopian to have an internet passport (private key) explicitly tied to a government identity, but at this point it may improve anonymity rather than detract from it. We could get rid of CAPTCHAs (which are mostly ineffective at this point anyway), Cloudflare MITMing, IP address banning, phone number verification, "log in with Facebook", spam filters (because we'd just block spammers), etc. in favor of being able to prosecute and/or block bad actors.