Note that this is the dictionary definition of shadowbanning, which is the primary thing the free speech advocates on Twitter were hoping Elon would stop Twitter from doing.
Here it seems like the poster and anyone else can directly search for the poster's account or the "filtered" post and still see it. But I guess the tags won't work. Kind of a middle ground. Perhaps they will explicitly remove/block the tags so the poster knows this is happening, which at least isn't tricking them.
To clarify, Musk said nothing whatsoever about tags. Another read of Musk's tweet is that your timeline won't show tweets with a negative sentiment analysis score.
Given recent events, I'm reluctant to assume Musk's vague goals will be implemented in the narrowest and most delicate way possible.
Unless they're pretending to be him (or advertisers) for a joke on Twitter, in which case they seem to very quickly get banned.
(As for the joke angle, twitter has decided it's too risky to allow humour that relies on brazen impersonation. Parody accounts that are clearly indicated as such are still allowed.)
And frankly, if this was a concern, maybe eliminating the way to tell whose account is who they say they are was a terrible business move.
Dude's got the most fragile ego as this whole episode has demonstrated.
Most of the folks getting banned for actually pretending to be someone else were either obviously a joke because they were totally silly, or were only able to look remotely real because of the idiotic $8 "you're real if you pay us" policy.
Shadowbanning applies to accounts.
Also, nowhere in that thread does it say that the user won't be told the Tweet was demonetized, which is the dictionary definition of shadowbanning.
I am not trying to provoke a flame war here either actually if it comes off that way. I think the business of social apps is trendy and you have to know your audience.
But that's only true when you delete posts and block users. It makes mistakes so costly (and with machines being so easily manipulated) that we need to run too many decisions past people because our decisions here are making huge impact at a personal level for those involved.
But if you're willing to accept "hidden at default-browse-level" wrongly appearing every now and then it becomes totally solvable.
We were trying to scale fact-checkers and playground-nannies but the trick is to get rid of those roles and use users to flag what users like and/or don't like directly. You can do a staggered release to previously critical people, and known-objective people, once a few posts have been flagged.