Hacker News Reader

Comment by EmilStenstrom | Hacker News Reader

The natural next step is to require bot labels for bots. Suddenly machine learning models that detect and bans bots without a label will be feasible. THAT will have a big impact on "healthy conversations" which seems to be their goal.

View on HN

This move seems designed specifically to forestall calls for that requirement. Consider: Twitter are first pointing out that bots can be good, IOW "we're not banning them outright". Then they're introducing this non-mandatory bot label, which makes it seem like they're aware of bots and working on ensuring they can keep doing all the good things they do.

Of course, none of this addresses malicious bots in any way. In effect, this is Twitter showing they're aware of their bot problem, but unwilling or unable to do anything about it, and are therefore doing a misdirection as a PR effort.

To me (who doesn't use Twitter) this seems like a fair solution. If you are running a legitimate bot I don't see what you have to hide. It's only people trying to be intentionally deceptive that have to worry.

You seem to be anonymous here on HN, why don't you use your full name instead if you have nothing to hide and you're not intentionally deceptive?

No, privacy doesn't work like that. Even people who are not trying to deceive others are deserving of privacy as well.

Most of my accounts (except one) on the internet are anonymous, as I don't want people to tie anything I do for fun on the internet to my real-life persona. It doesn't mean that I want to deceive people, only that I want my meatspace life to be something else than my online life.

Your bot could be anonymous, the only thing you'd be forced to reveal is that it was a bot.

just a [bot] marker of some kind next to the tweets that were posted by a bot. That's it. Still anonymous.

I could tell you this comment was posted by a bot, that would reveal nothing about me.

Anonymity is fine, they were saying that bots should be labeled as bots.

> privacy doesn't work like that

Except we aren't discussing private conversations. We are discussing broadcast conversations on twitter.

Without tying your anonymity to something else (whether it’s another anonymous account or a real world identity) it’s impossible to determine whether you want to deceive or not, no?

In any case real identities have been a failure anyway

There's plenty you could have to hide as a user on HN. But what reason does a bot have to hide the fact that it's a bot?

You people and your absolutes. Nobody is saying that every anon is deceitful and untrustworthy- just that you need to make sure your BS detector is working.

I don't see how "I am a bot, beep boop" is a privacy issue.

But a bot isn't a person. And you don't have to reveal who made the bot, just that it's a bot. Can you give me a good use case for a bot that is not some kind of manipulation and would still benifit from humans not knowing it's a bot? Usually you want people to know so they aren't confused why the account acts werid i.e. unhuman.

Twitter makes its money as a marketing engine. Who can say what marketing is deemed intentionally deceptive? I'm not a Twitter user either but it seems to me an obvious question.

Most bots aren't marketing bots, they're sort of... Goofs, or utilities, or projects. Like @MyDudes_, which posts a picture of a frog every Wednesday. Or alttextbot, which if you follow it, DMs you when you've posted a picture without alt text. Or wwiiInRealTime. Or wintbot_neo, which is inexplicable outside of twitter but serves only comedic purpose.

> Most bots aren't marketing bots,

wouldn't you need to know how many marketing bots there are to determine that?

making statements about goodbots/(goodbots+badbots) is difficult when the good bots already make it clear they're bots, and the bad bots are misleading.

> Most bots aren't marketing bots, they're sort of... Goofs, or utilities, or projects.

In my experience those kinds of boutique/indie bots appear to make up a tiny percentage of total bots. If you have data that shows otherwise, I'd appreciate if you could share it.

They're the kind people engage with. There's obviously a lot more astroturfing and disinformation, but most corporate marketing on twitter is done by humans. Steak-Umm, Oreo, Microsoft, Wendy's, even indie game publishers have a human running their accounts.

This is what Discord does.

A bot looks exactly like a normal user, except for a [BOT] label after the name. Works like a charm.

But on Twitter, bots aren’t only bots. For example, I have bot accounts that auto-tweet things, but then I can log in to those accounts and manually tweet.

Because of that I’d think the label would have to be applied to individual tweets and not to the profile.

AFAIK the Fediverse requires this (on mastodon instances).

I really appreciate that some masto/fedi clients have the option to indicate bot accounts. Tusky adds a little robot icon to the profile pic of bot accounts, which helps when deciding whether/how to interact.

This is how I see it too. Twitter is boiling the frog on malicious bot accounts.

Bots that want you to think they're human will end up with "bot verified" badges that undermine the strength of their opinion, or they will be banned.

> The natural next step is to require bot labels for bots.

Great! This would convince me to use Twitter again.

Only on Twitter is 'bot' now a loaded term the masses understand. In every other medium you could already (and many did) let some software post automatic posts (most benign example: "Here's my new blog post: xxxx") or scripts on irc.

Where is the line? Technically scheduled tweets are automatic, and so, by some definition, bots.

Reply on news.ycombinator.com