Hacker News new | ask | show | jobs
by amerine 969 days ago
Not a x/twitter user much these days (mostly quit a long time ago) but I’m happy when companies put a minor barrier up to stop botting. So meh, good idea I guess?
2 comments

Anyone running a TwitterBot through an LLM or other AI tool is paying far, far more than $1/year.

This $1 'fee' monetizes and benefits bot-accounts more than anyone else.

------

Let's say I setup 10,000 bot accounts for $10,000/year total. Will I really be banned from Twitter? Or will I be treated like the high paying customer that I truly am?

Is $10,000 expensive for a marketing team pushing a new product? Sounds like easy-astroturf to me.

-------

The economic incentives here are all wrong. Astroturf / marketing companies pay $10,000,000 for a singular 30 second Superbowl ad. They absolutely don't give a crap about $10,000 for an easy army of astroturf accounts that will upvotes and generate fake LLM discussion around products.

In my mind this is not about spammers not having the money to create 10k accounts.

but about how many CCs would they put for 10k accounts? If you have 1000 accounts with the same CC it does seem to make it a bit easier to say maybe these 1000 accounta are bots.

but let me tell you what Musk can do to actually let bots run: allow crypto payments :D

Each account will require a unique credit card and phone number. https://twitter.com/xDaily/status/1714513153407807835
One CC.

Only Twitter will know that the 10,000 accounts are bots. It'd be up to them to ban, but why would they ever ban a high-paying customer who is paying literally 10,000x more than the typical user?

the change is to require payment for all new users so they won't be paying more than anyone else other than the old accounts that got grandfathered in. They won't ban them if they pay but they will derank their posts and if they do allow a group of them to sign up with the same CC then they can punish the entire block of them if they start spamming with any of the accounts. It also allows for permabans based on CC or other identifying info making it more difficult to create a new botnet. It's not a perfect solution but it does add friction. There will probably be ongoing changes to make it even harder for bot accounts.
With what moderators? Twitter laid off significant amounts of their moderation team.

No one is going through account creation or double checking this stuff anymore at Twitter.

I think this severely overestimates the sophistication of bots. You don’t need to give coherent, thoughtful replies. You just need keyword detection and link spamming.
I'm not a programmer, but it seems like they could make it non-trivial to automate the payment process. If I have to click 30 times per account, with a captcha, and wait a minute for the payment to clear, 10,000 accounts suddenly starts to become very expensive in terms of time/effort.
You literally buy Filipinos to do this for you for like 5-cents per account. Maybe 1 cent, account creation never was an issue in today's bot / astroturf meta.

Hell, for like 5 cents/account, you probably can get Filipinos to post enough legitimate traffic about innocuous topics to look like a human before being turned into an astroturf bot.

IIRC, the typical Filipino baker makes $5/day. They literally can make more money making accounts for you all day long at 5-cents per account than legitimate jobs in their town.

That's 100 accounts made in a 8 hour period. More than doable.

-------

Like seriously, ever since 'Mechanical Turk' and other websites made cheap labor from 3rd world countries available, this has been a solved problem.

https://www.mturk.com/

It seems like they could also limit the number of accounts per credit card. 10,000 credit card numbers are not easy to come by. And all of this is creating a pathway for law enforcement. I’m skeptical it’s as easy as you say.
Why would Twitter reject $10,000 coming out of one credit card?

The economic incentive of this model is backwards. A person paying $10,000 for 10,000 bot accounts is a bigger source of revenue than a $1 normie.

You think they’re making money on $1 per account? Processing fees will eat most of that. It seems likely to me they’re doing it to improve the user experience, which if that recruits more users will be worth much more to them than the 10,000 bots each paying $1/year.
Bots and spammers are already using stolen credit card to pay for Premium. This isn't going to stop them.
Maybe the top end of the distribution is doing that. That’s an entire criminal operation which no doubt does happen of course. But not all spammers put in the same amount of effort. Cutting down spam is about sealing off the big leaks first.