Do you really want ISPs to be in the business of deciding what should and should not be on the internet though? That sort of thing typically doesn't work too well.

Deciding what is or isn't a "scam" is really a job for the independent judiciary. But getting a ruling is difficult and time-consuming, and also largely pointless because a new website can be created almost instantly, and there are many foreign ISPs where you can't get a ruling at all.

I don't really disagree with your basic premise that "disconnecting the bad actors' ISPs from the Internet" is the ideal solution, but this is far more difficult than your comment implies – almost impossible with how the internet currently works.

We are oh so close to a balkanised internet.

Thankfully, the powers that be outside of CN, RU and a few others haven't really gone to town on buggering up basic connectivity, yet.

I think that the parent comment was referring to the days when ISPs were local or network services were provided by a school or a some other entity which had a vested interest in keeping their network clear of bad actors.

The fix is to give users the tools to lean more heavily on transitive trust and less heavily on the platform.

Or we build those tools ourselves and let the platforms be dumb pipes which we selectively slurp from.

The days of trusted-by-default banned-selectively are ending. It's dark forest time.

This ignores that botnets, malware, etc exist.

Well, ISPs can hold their customers accountable as well. Get told you're running malware, you got 24h to get it fixed, or you get disconnected.