Browse HN with showdead on and check out some of the shadowbanned posters. You can find many with long histories of "dead" comments that keep posting anyway.
On e.g. Twitter you can find all sorts or people saying random stuff and getting zero likes. They wouldn't necessarily notice either.
And try reading the comments section of a Fox News story. No way most of these people know or probably even care if they're banned, it's more like a tic than actually engaging in a conversation or even real trolling.
Shadowbanning is effective at increasing the SNR against essentially casual vandalism. Against deliberate trolls, the kind of people who are regularly checking from a logged out browser to see if their posts show up, not so much.
My theory is that the bots keep posting because there are enough "legit" bots that pull HN data from the API to put into an assortment of search engines and other statistical tools that ultimately Google and Bing will crawl those links from the secondary sites pulling data from HN. I could be wrong but the only other explanation is that the bots are just incredibly poorly coded. At least I can't think of another reason. I have no idea if the legit bots using HN's API have showdead enabled. If they do that could be part of the problem.
I say this having fought bots in the past on forums and the bot operators would certainly know if their posts were no longer visible to others so they would just keep creating new accounts. I would block IP's, then they would use proxies. I would block proxies, they would use Tor. I would block Tor then they would just create a few accounts a day then let them "warm up" with real content. This led me to putting people into "ranks" and only people that had been interacting with the site could post messages that could be read real time, whereas low rank accounts had to be moderator approved. I had SQL code in a cron job that would delete posts over 30 days old in the newbie rank.
https://news.ycombinator.com/user?id=aaron695
My theory is that such users do know their status but just figure it doesn't matter and that people like me--who absolutely browse the site with showdead and would want all my friends to as well, as I don't want to not hear what someone says merely because they got angry--do read their content, and people like us are their true audience.
it harms the site as well, not only it's users. just look at whats happened to the ghost town formerly known as reddit.
Detecting user's ban evasion is to engage in a costly cat and mouse game between evaders and detectors. Shadow banning, however, is easy to implement and cost effective. Platforms that do it save resources compared to those that don't.
Your comment reads like you don't like that platforms have more power than users and you resent that it's not an even playing field. That's a fair point, but it would be easier to say that if that's how you feel.
Using it on actual people was never recommended, because it was always very obvious to anyone who actually engaged with the site. They'd go back to the same old threads on another computer and find their posts were gone, which was incredibly obvious. This just created more forum controversy, e.g. back when pj of Groklaw started doing it back in the day to suppress differing opinions on that TurboHercules controversy.
The "shadow ban" on Twitter that's garnering so much attention was manual down ranking of content. I'm not even sure if it's productive to call this shadow banning. Some content is given more visibility by the algorithm than others. Nobody is surprised by this. The main distinction in Twitter is that the downranking was manual, done to specific accounts. This realistically can't be detected, not evaded since abandoning a popular Twitter account is not an option. And for famous people, the downranking would be applied to their new account anyway.
If you strip individual people of voice, which is directly tied to their agency it tends to have only a few different outcomes, almost all negative.
Its a stupid idea put forth by stupid people.