Because SMTP was an open protocol. No MUA company profited from its end user seeing spam. No end-user-targeted MTAs (i.e., the MTAs of most end users' ISPs) als

Because SMTP was an open protocol. No MUA company profited from its end user seeing spam. No end-user-targeted MTAs (i.e., the MTAs of most end users' ISPs) also found spam as a cost center. (I'm being pedantic here in that the rogue spam-friendly "Spamhaus" ISPs whose customer base was made up of spammers were technically operating MTAs as injection points. To the overwhelming majority of email users, admins, and providers, spam was a massive cost center.) At some point in the 90s/00s, ISPs had to bite the bullet and acknowledge that if they didn't filter spam, customers would leave for ISPs who would. And in the SMTP era, y0u could use b4y3s14n techniques to f1lt3r out the v1@gr4 spam. We're not at the point where the video equivalent of spammers can be automatically detected. So it's going to be expensive. And in an age of walled gardens and ad-tech, the providers of "next up" and "recommended for you" clickbait/spam are not losing money (by having to pay for storage/bandwidth for spam that their end user merely deletes or filters client-side). Rather, sites like YouTube are profiting off of every clickthrough. An ad view is an ad view. And that is why they have no business incentive to pay the money to hire the humans that are still required to solve the hard problem of differentiating spam from ham. Doing nothing, or pretending to be able to fully automate the problem away via still-error-prone AI, is cheaper than solving the hard problem of curation.