Hacker News Reader

Comment by BiteCode_dev | Hacker News Reader

It's a good disclosure, so I don't want to remove the merit out of it.

But it also shows they have the capacity to target and remove accounts when they need to, but they let bots run rampant.

Last week I quoted HN on twitter, and the article name contained something about a pirated instagram account.

In under a minute, I received a dozen answers to that tweet, phishing attempts by bots pretenting they could help me with what they thougth was me, loosing my insta account.

So twitter can easily find them: setup a honeypot, wait for bots to show up, ban them to oblivion.

Since the service is now playing a game where you need to provide your phone number to create an account (they say they don't, but practically, they do), it's quite pricey to create bots nowaday.

They could destroy thousands of bots in a week.

They don't.

View on HN

It's the same thing the Match Group does with their dating apps. They leave the bots up to drive traffic, and keep real users engaged.

Then they clear out the bots after letting them run for a bit. You see 20 matches, likes, etc. The moment you pull out your credit card, to "see" your matches you'll find most of them are from bots/ deleted accounts.

It's against Twitter's profit motives to clamp down on bots in any meaningful way. Ultimately, these bots do simulate traffic, and when you get your endorphin rush from getting your bot retweets, Wells Fargo or whoever else is advertising on, Twitter gets to monetize you.

Alternate interpretation assuming incompetence instead of malice: they _do_ ban thousands of bots a week, but their security staff isn’t aware of this particular attack vector and so they haven’t banned these particular ones (or the attackers have some way of evading analysis, honeypots, …).

Every casual twitter user jokes about it, so the "they dont know about it" is a claim that would require a lot of evidence - they've taken action against the "elon musk" twitter bots, but the same scam is effected constantly.

After years of incompetence you can start to assume incompetence.

It's simply not a priority for them.

This is the right answer.

It's never that they can't do it, they just don't care enough to do it.

They won't care until liability is attached to it which will make their entire business model of profiting off "engagement" at the expense of others unviable.

The obvious answer why Twitter hasn’t banned bots is to help keep their DAU metrics up

Bingo. Just the other day I saw someone arguing with bob8374922 in a very long thread. When I clicked on bobs profile he had zero followers and account was created this year, no tweets that weren’t replied to others arguing insane “I’m a critical thinker!” Positions.

A very common theme.

I myself had a full conversation with one of the bots that appears if you say "Metamask" before I realized that @meta2mask was not really a helpful support person from the internet, and was actually a scammer trying to get crypto seeds.

So is this a bot or a brand new user who decided to spill his frustration about society online and accepted the default username suggestion that Twitter offered him?

In an efficient market public investors would demand to know the true/accurate DAU metrics with bots removed.

Only if the benefit of actually knowing outweighed the effort of finding out.

Twitter is a mature business. Revenue is driven by ad rates. Ad rates are driven by big buyers who know how effective their buys are — regardless of bots and whatever stats Twitter gives them or us. Therefore, investors have little reason to care about bots.

Devil's advocate;

1) Number of bad actors will always be several orders of magnitude more than number of people twitter can appoint to the topic to. For example: this is the first time I'm hearing instagram-lost-password bots. There should be bots for every topic you haven't even heard about.

2) Increasing the "police" will increase the false-positives. Bots are an inconvenience, but undeserved punishment is a reason to leave the platform. Multiply this by number of connections that user has.

3) You'll inevitably hire one person who will police for fun, trolling, personal gain or religious/political/philosophical agenda for every N number of people you can hire to "police".

4) Bad actors have a lot of time, sometimes capable and rarely even well organized. They create sophisticated tools to mimic real users to "warm up" bot accounts.

5) Bad actors could be, and quite often are, from 3rd world countries. It is expensive for twitter to hire good engineers to fight them, but for bad actors, several dollars per hour could be the best thing they can do with their life at the moment.

Bots contribute to “engagement” numbers. Engagement-driven platforms will only take action when the bad PR from the bots outweighs their “engagement” contributions such as when they post the “wrong” things that go against the narrative they want or when such actions fit nicely within said narrative (and state-sponsored misinformation is all the rage in the current political climate).

Take example of the January 6 insurrection against the US Capitol - the whole thing has been brewing for a year on Facebook out in the open (this wasn’t some sort of advanced terrorist attack organised in the shadows protected by end-to-end encryption) but nothing was done because all this craziness has been generating insane amounts of “engagement”. It’s only when the whole thing blew up that the bad PR from being associated with it outweighed the engagement that actions were finally taken.

Same goes for the Elon Musk scams[0]. They have the same photo as Musk this is not rocket science. On photo upload do a similarity check before allowing new tweets to go live.

[0] There is a vast, unending sea of bots that shill scam cryptocoins or other scams that copy Musk's account and name.

Not just elon, any popular account, movie star, politician, has some scam bots in any thread.

And that's just what is visible. The bot to farm clicks, likes and rt are less so, and probably more numerous.

Twitter allows parody, so this would be hard to enforce without banning legitimate uses.

You could say the same about every other law against actions that are outlawed, and yet things seem to work out well. People can't set up a Ponzi scheme and then claim they're merely running a parody of Bernie Madoff and nobody seems to complain.

The subject is automated enforcement, not law.

Devil's advocate- bots running rampant is merely an annoyance but not necessarily horrible. Targeted disinformation campaigns designed to hurt an individual or group is bad.

Given finite resources- you go after the activity that causes the most harm.

Reply on news.ycombinator.com