Hacker News Reader

Comment by p-e-w | Hacker News Reader

I assume that by "prioritized" he means "ranked higher in the feed".

If so, this actually makes sense and should improve quality on average. Yes, verification is quite cheap and pretty much anyone can get it so it's not a perfect spam filter, but it's a million times better than letting bots and sockpuppets rampage among people who actually have something to say.

I strongly believe that the only effective way to control spam and abuse is to control who gets a voice, rather than trying to control what they say. Content-based methods don't work, and neither does banning "bad" accounts, as long as people can just create 10 more sockpuppets afterwards. This is a small, but welcome, step in the right direction.

View on HN

It is exactly the opposite. The “verification” verifies nothing other than having a credit card and 8 USD to spend. There already have been “verified” accounts with blue checkmarks impersonating the NY government, governor and other people. [1] All this will do is push these impersonating accounts higher in the feed than the actual legit accounts.

It’s an easy and cheap way for fake accounts to gain access to a larger audience.

[1] https://www.nbcnews.com/tech/fake-accounts-pop-twitter-legac...

> The “verification” verifies nothing other than having a credit card and 8 USD to spend.

Indeed. That's why it's effective for reducing spam, since spammers who run thousands of bot accounts aren't going to verify each of them unless their spamming activity generates at least $8 per account. Most spamming activities generate far less revenue, so those become unprofitable, and the overall amount of spam goes down.

Future will show, but I believe you’re wrong about this. It’s still possible to create bots, but now, for a mere 8 USD, I can have it boosted to top, above legit human accounts. This may reduce the amount of cheap spam, but it’s a godsend for any actor that is for example trying to influence public opinion. For a mere 80000 USD I could have a verified bot army with 10000 accounts drowning out a lot of legitimate conversations. Or push advertising. Or I could go all the way and just use stolen credit cards.

> For a mere 80000 USD I could have a verified bot army with 10000 accounts drowning out a lot of legitimate conversations

That’s a different problem. The parent is referring to the cheap high-volume spam/scams that have proliferated every corner of the free internet. These only work because they’re extremely cheap. There’s a reason your email inbox receives 10,000x more spam than your physical mailbox.

If a scam can't even make $8 per account it's pretty much doomed to failure anyway. Think about it.

Paid spam accounts are removed as well, which you left out of your analysis.

Buying an account doesn't give you un-suspendable account.

All spam detection techniques are applied to paid accounts and they are removed as well.

This not only raises the cost but difficulty.

Just having money is not enough. Detecting 10+ payments from the same credit card is trivial.

There aren't infinite amounts of valid credit cards or Apple accounts to be used for verification.

Spam detection at Twitter is horrendous. Search results for many crypto and stock ticker symbols (or cashtags like $SOFI) are deluged with Discord links for shady trading groups endlessly trying to rope in paid subscribers. If you select "top" search results, you'll see much less spam but stale tweets. "Latest" yields tons of spam. They've been tweeting with the same phrases like "don't miss the next move in a few hours" and "top analyst price target" for years, presumably because it's never been a priority to get rid of them.

In a sense it’s legitimizing bots as long as they pay

Was spam ever that much of a problem on Twitter? Stupid humans were always a much bigger detriment in my experience and this move disproportionately boosts stupid humans.

They were if you were a billionaire named Elon Musk whose likeness was constantly being used to promote crypto scams to your gullible followers.

$8 won't stop the really bad actors, just the trolls.

And impersonating someone could easily make a lot more than $8 in a very short time.

Maybe the two things should be different? Blue checkmark for official accounts, but to get boosted you need to pay up?

Paying a tiny sum to get boosted is still a shortcut for grifters. Paying for the effort to do actual verification would be an option, but that’s not what’s happening here.

Anecdotically, accounts whose speech I find valuable have not paid, and accounts with negative-value speech have.

Rationally, it makes some sense, as people willing to pay $8 for being published consider their publication to be worth -$8.

I agree though that verification (which is not what is done here) would be valuable. Not sure why they didn't go for a KYC.

> people willing to pay $8 for being published consider their publication to be worth -$8.

I don't think this is right. I'm not willing to pay $8 for something just because I think it's worth -$8 to you. I don't care how much it's worth to you. Rationally, the value of something to you doesn't factor into my valuation. I pay $8 for something because it's worth at least +$8 to me.

When I give you a barrel of oil in exchange for $80, that oil is priced at $80, and that is often phrased as its value.

When oil prices went negative in 2020, people were giving both the oil and $30, and so the oil was priced at -$30. Sure, the oil was worth $30 to give away, but the oil was not valued at $30; it was valued at -$30.

Many people sell speech (eg. President Obama has given a speech in exchange for $400K). When you give both money and the speech, just like oil, that infers the price of that speech.

> Rationally, it makes some sense, as people willing to pay $8 for being published consider their publication to be worth -$8.

How does this follow? I might spend $8 if I think I have valuable things to say if it means prioritization over thoughts I consider less valuable. It's not like the reach of a publication is determined by its value.

The problem is that the blue check has become a culture war so all the blue checks are aligned with Elon (either politically or in temperament) so the top replies to popular tweets are dominated by that tribe. The old way (whatever it was) wasn’t great but at least you got a mix. Not good for the product imo, it would be nice if he offered a setting to choose your ranking algo for tweet replies.

Maybe things will settle down though and blue checks will become more evenly distributed (or engagement will tank and he’ll roll this back)

> The old way (whatever it was) wasn’t great but at least you got a mix.

I thought Twitter was a complete echo chamber where you only ever heard from people aligned with you, except more extreme.

I never notice bots on Twitter except for those shilling for crypto in Elon’s comment section.

What makes you think bots and sock puppets wont be verified?

The fact that if an entity uses thousands of bots, verification does become expensive. At the very minimum, it's an additional barrier. All else being equal, bots and sockpuppets are less likely to get verified than regular human users. I've seen communities where an individual user created dozens of sockpuppets to troll around. Such sockpuppets won't be verified for obvious reasons. Not to mention, if a credit card is used to verify lots of spam accounts, the card itself can be banned from being used to verify additional accounts, which hampers bad actors who have just a few credit cards lying around.

Bots and sockpuppets have an incentive to be verified -- they're hoping for a return on investment. What benefit is there for a regular person that can compare with that?

It's also not that there will be a single entity using thousands of bots but thousands of individuals all paying because of some real or perceived financial incentive.

What if the strategy of the bots is to signal boost a small handful of accounts? Verifying that small handful of accounts isn't expensive.

Why need thousands of bots? They get banned at a rate of a couple a day at most. That’s a pretty long time to scam.

"should improve quality on average"

Based on replies to that tweet I think this is hard to achieve.

> it's a million times better than letting bots and sockpuppets rampage among people who actually have something to say

You are replying this to the man who just destroyed the only way users had to verify legitimate sources of information on the platform, setting the floodgates of disinformation open.

Please, if you (not you specifically) rely on checkmarks that in some cases have been given as personal favor or to amplify specific voices despite their extremist views, to verify legitimate sources of information, that’s a general educational problem in understanding how to identify legitimate sources in the first place. This should actually be taught in school, because it’s really Fkin important, but isn’t.

The floodgates of disinformation on Twitter were already open and actively abused before the new system. The old checkmark system was beyond flawed and the new one is as well. Just that this version fills the pockets of the man who is desperately trying to destroy the platform he didn’t want to buy in the first place to minimize losses on his way out.

I don’t mind people wanting to support Twitter or Elon. And they should not be ridiculed for doing so.

But I think as a society the deciding factor if your valuable contribution is heard or seen should not be a monthly subscription. Twitter blue is inaccessible for a large part of the population of this planet. Poor people completely loose the ability to make valuable submissions on the platform and participate if they are drowned under people paying 8 bucks to better troll others or push their views that wouldn’t get traction without the checkmark.

The solution is not to change Twitter. It’s getting away from relying on this specific platform as its algorithm manipulates you into a worse, less critically thinking version of yourself.

What exactly stops the bots and sockpuppets from having blue and this getting ranked higher now?

If the value the bot operator gets from the account is less than the cost of buying Twitter Blue, then they'll make a loss.

Reply on news.ycombinator.com