[pmiller2]

I've seen accounts with green usernames that seem to have been shadowbanned for less. Typically, it's a new account that says some reasonable things, and maybe one or two fairly unreasonable things. Getting flagged for the unreasonable things can really kill a brand new account with under 10 karma.

[dang]

"Seem to have been shadowbanned for less" than...criticizing Rust? come on you guys. That's not even close. Perhaps you saw an account that was affected by a software filter, especially if it's new. We try to review the affected posts and immunize legit accounts against the software, but we miss some.

> Typically, it's a new account that says some reasonable things, and maybe one or two fairly unreasonable things.

Accounts get banned for the worst things they do, not the best. If one day I go get groceries, walk the dog, and set somebody's house on fire, "the majority of things I did were reasonable" is not much of a defense.

It's hard to speak in the general case though. People routinely overinterpret what they see based on what they assume is happening. If you haven't specifically asked and we haven't specifically answered, the explanation is probably wrong. It's not hard to get an answer if you try.

[pmiller2]

Yes, "seem to," as in, I have no idea why, because nobody tells me and it's not publicized anywhere how shadowbanning works.

> Accounts get banned for the worst things they do, not the best. Suppose one day I go get groceries, walk the dog, and set somebody's house on fire. It's not a defense that the majority of things I did were reasonable.

I'm not talking about "set someone's house on fire" though. I'm talking about a comment that triggers a handful of people to mash the downvote or flag link. Specifically, this account: https://news.ycombinator.com/threads?id=rbecker which, obviously someone else noticed as well: https://news.ycombinator.com/item?id=23585354

That user's subsequent comments have not been great, but, at the time that I encountered them, they had something like 10 karma. Note how I wasn't the only one who couldn't figure out why they were shadowbanned?

[dang]

> it's not publicized anywhere how shadowbanning works

https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

The account you mentioned isn't banned, so I'm not sure why you're bringing it up in this context. It has, however, shown signs of being an ideological battle account—quite distinctive signs in fact—and I've already warned it: https://news.ycombinator.com/item?id=23665090.

New accounts that have come here to use HN in the intended spirit don't typically post exclusively about racial statistics, immigration, China, why lynchings were not as commonplace as people suppose, and similar things for their first 52 comments. Could it be that that is just a random streak and a wave of intellectual curiosity is about to come? Sure. But at some point, if it walks like a duck, we ban it.

[pmiller2]

So, you're saying that both bans and shadowbans are manual interventions? I did not have that impression regarding shadowbans. What I, and the comment I linked to earlier saw was a combination of green username (new account) and every comment after a certain point marked as [dead]. If that's not "shadowbanned," I don't know what it is. This user also seems to attract downvotes to fairly reasonable comments for some unknown reason.

That said, yes, this user has some questionable comments, most of which seem to have come after the point where I initially noticed. I don't see any reason why posting about racial statistics, China, immigration, and lynchings is a big red flag, especially the way 2020's been going so far. It is certainly easier to troll on these subjects than it is about how much Rust sucks, but, in isolation, I'd call this a chartreuse flag, at best. I do agree with you about your assessment of this user's current comment history, however.

[dang]

Sometimes it's manual and sometimes it's software. If it's software, it won't permanently ban the account, and you'll more likely see [dead] comments followed by live ones. On the other hand, a banned account can have its comments restored to live status by vouches from other users, which might be hard to tell apart. On the third hand, software sometimes will permanently ban an account, but that's linked to patterns of spamming. Spamming and trolling are different phenomena. Spam is easier.

> I don't see any reason why posting about [...etc...] is a big red flag [...] It is certainly easier to troll on these subjects than it is about how much Rust sucks, but, in isolation, I'd call this a chartreuse flag, at best. I do agree with you about your assessment of this user's current comment history, however

It isn't a red flag in principle but it's a big red flag in practice because these accounts pattern-match extremely reliably with certain classes of users who routinely end up getting banned. Once you've been doing this for 10 years you notice these things early. I don't mean to sound like this is some sort of special power. It's just gruntwork. Believe me, it's boring.

Occasionally someone comes along who (a) is truly a new user who hasn't been banned before; (b) walks like a troll and quacks like a troll on troll topics; but (c) turns out to be an intellectually curious person whose mind, let's say, functions a bit differently from most people's. They might even be that thing we all dream we are and basically none of us is: an independent thinker. Obviously we want such a user. But the sad truth is that the dreariness of pattern-matching turns out accurate the overwhelming majority of the time, and for every such outlier, there are hundreds of trolls - or dozens of trolls making hundreds of accounts. So we play splat-the-mosquito, the mosquitoes respawn, and the karmic wheel turns.