Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Twitter says Musk’s spam analysis used tool that called his own account a bot

arstechnica.com

Post reply on HN
Print view View original

481–490 of 585 posts

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#481

Post by Vespasian »

spaceman_2020 wrote: post #474

I just find it a little hilarious that Twitter is even attempting to counter the bots claim. You have to spend 5 minutes reading the replies to tweets by prominent politicians to know that its filled with bots

None of my tweets ever got a bot reply.

Now that is most likely because they are all about very local topics or things related to my specific field of work.

They are also not very popular (a handful of retweets) and mostly cater to people I know in real life.

I know at least 10 people who regularly use Twitter through the app (aka being shown ads) but never ever tweeted a single thing.

They would be counted as real users by Twitter but 1000 bots who never viewed an ad wouldn't. That is all that Twitter claims in their SEC fillings.

That there are roughly 200 million active daily users who are being shown ads and that of those 200 million (very roughly) 5% are bots but "who knows it's a guess", says Twitter.

If there were an additional 50 Billion bots who never saw an ad, Twitters statement would still be true.

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#482

Post by Vespasian »

cambaceres wrote: post #479

What happens if Musk loses? Will he be forced to buy the company, or just to pay a huge fine?

Twitter is sueing for specific performance (aka pay up the full price) and argues that Elon's funding by banks is still comitted and must be used to buy Twitter at 54.40 a share.

IANAL but this court already did enforce similar contracts in the past.

Of course the parties can also settle out of court at any time under any terms.

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#484

Post by lostdog »

RandomBK wrote: post #462
convery wrote: post #10

Earlier quoted context omitted.

> Musk has no legal standing to use any of that to back out of the merger. The claim is that Twitters SEC filings had bots at ~5%, and Musk made the investment based on that. From the leaked recordings of staff, everyone knew that the percentage was much higher.

Twitter doesn't say that 5% of all accounts are bots. While I don't have inside knowledge of how Twitter does things, the industry standard is to: 1. Build a classification model to determine whether an account is a bot or not. 2. Measure the precision/recall curve of that model, and pick a threshold where the precision would be >95% (i.e. >95% of what the model labels as "humans" are really humans). (Higher recall l…

Twitter describes how they get the 5% number (from Matt Levine [0]).

They randomly sample 100 accounts per day and have human raters decide if they're bots or not. This is the right way to validate that your fancy ML algorithms are working well.

[0] https://archive.ph/GtYzO

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#485

Post by DannyBee »

Sebb767 wrote: post #36
zinekeller wrote: post #18

Earlier quoted context omitted.

> Twitter committed fraud by lying to the SEC about the mDAU numbers with the intent of inducing Musk to buy Twitter at an inflated price. What is that commonly-used aphorism? "One man's trash is another man's treasure"? Like, even if this is true , Musk didn't do (and subsequently waived) due diligence. > No, really, this is the allegation No further comment on it. > Count 2 is that Twitter committed [Texas fraud st…

> What information? The contract is already in the public, but okay you want to waste court time. The contract stated that Twitter had to provide Musk with internal data as required to close the deal. His teams tokens were apparently rate-limited at some point and it took a while for Twitter to give him access to all data, a.k.a. 'the firehose'. Just for completeness sake, Twitter seemed to still try (the token thing…

"His teams tokens were apparently rate-limited at some point and it took a while for Twitter to give him access to all data, a.k.a. 'the firehose'."

Here's the thing though:

None of that data was required to close the deal. Musk had no diligence rights, he explicitly gave them up, and the contract clearly says so. He doesn't get to do regular diligence through that clause, it's meant to cover the literal things necessary to close the deal (like financing info or whatever).

So Twitter didn't have to provide anything here, and certainly not access to the firehose.

His response is going to be smacked down pretty hard by a chancellor - it's 100% nonsense.

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#486

Post by lostdog »

polishdude20 wrote: post #299

How much of this show and dance is really about covering up something else that Musk did? Is it possible this is done to overshadow other newsworthy stories that would be even more detrimental to his reputation?

There must be a million cheaper ways of drawing attention away. Here, let's come up with a couple:

Musk announces that in order to help repopulate the world he's looking for 10 female "paid volunteers" to come stay in a compound during 2023.

Musk begins a run for Senate in Wyoming.

Or even more simply: Keep talking about purchasing Twitter, but don't actually write up an offer!

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#487

Post by kukx »

glenngillen wrote: post #419
bostonsre wrote: post #321

Earlier quoted context omitted.

O, that seems pretty odd. I can't think of any reason for them to do that beyond trying to get out of the deal or incredible negligence on the musk side in not doing that before hand. Were they able to get that data before hand if they wanted it?

It doesn’t really matter at this point, because he signed a binding agreement. He’s the one that made that move. Specifically why is just speculation now. But, hey, we’re all here speculating anyway! My take is: 1) He’s a history of making impulsive decisions without thinking (or caring) about the consequences. 2) It forced Twitter’s hand because they couldn’t act like it was it a non-genuine offer. 3) He’s so rarely…

You have to make a series of countless good or great decisions to be that successful. His ratio of good decisions to bad ones is probably one of the best in the world. He is not above the law, he has to stand before court as every normal citizen has.

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#488

Post by Vespasian »

chowells wrote: post #57
pcmoney wrote: post #6

Musk seems like a smart guy who got overconfident and out of his area of expertise and made a series of incredibly stupid decisions. Pretty much every legal and financial professional without a vested Musk related interest seems to think similarly. Matt Levine has been lights out on this topic. Everyone knows his bot claims are just FUD to try and weasel out of writing a check he no longer has the guts to cash with c…

Where was there ever evidence Musk is a smart guy? His only skills appear to be self-aggrandizement and being born rich. Musk is proof that our economic system rewards being already rich, not creating value. He did not found Paypal, Tesla, SpaceX, or anything else that's been successful with his name attached. He did not contribute any tech to those efforts. At best he's contributed publicity, but it's only ever with…

Even most people being born richer than him and why happen to have a better relationship with the family wealth don't become billionaires not to speak of the richest person on the planet.

So he definitely has some skills in leading businesses through hard times and making them succeed in the end (I've no idea how or what he does so).

Founding a company doesn't matter as much as making it grow and succeed.

Ray Kroc didn't found McDonald's (he didn't even join early) but he turned it into the massive company it is today, so it doesn't really matters who once initially signed the incorporation form.

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#489

Post by loceng »

Sebb767 wrote: post #393
loceng wrote: post #231

Earlier quoted context omitted.

I imagine the argument is whether the method Twitter used, and was seemingly acceptable by SEC, is ethically, morally, or logically a reasonable measure. E.g. Can they 100% say that there are less than 5% bot accounts, or is that 5% number being pulled out of their ass - and saying "less than 20% of accounts are bots" would hold as much water? That is therefore then a misrepresentation, fraud IMHO - and stockholders…

You're missing that they never promised that there are only 5% bots. They said they estimate five percent based on their internal research, which is completely true as far as we're currently aware. Unless you can prove they intentionally faked their own internal investigation, you'll have a very hard time proving fraud. Also, then there is the fact that Elon apparently even didn't believe this number. So even if he p…

So if you have a home inspector inspect a house before you buy it, and they say "yeah, it looks good" - but it turns out they didn't actually investigate in a way that arguably involves an adequate amount of integrity or in a reasonable way to determine as accurately as possible that everything is sound - then you don't think that home inspector isn't liable? And what if the professional organization who has the responsibility of regulatory oversight of that profession, what if they're been lacking and not being responsible as they are supposed to be, allowing for poor analysis as the standard - are they guilt-free too?

Re: Twitter says Musk’s spam analysis used tool that called his own account a bot

#490

Post by swores »

madeofpalk wrote: post #309
8ytecoder wrote: post #98

Earlier quoted context omitted.

He claimed to buy Twitter, specifically, to clean it up; to get rid of bots. That was his stated goal. "If our twitter bid succeeds, we will defeat the spam bots or die trying!" https://twitter.com/elonmusk/status/1517215066550116354

The weird part is that "twitter has lots of spam bots" is a completely independent claim from "5% of the users who use our website or 1st party apps are bots".

While your comment is technically true standing alone, you've also slightly misunderstood what the 5% figure is about.

It's not "users who use our website or 1st party apps", it's "monetizable daily active users", which may be similar to the group of users you described, but most importantly Twitter can choose not to show any ads (or not to charge money for ads shown) to bot accounts.

So even if Twitter only had 1 single non-bot user, as long as that's the only user they are "monetizing" they would be at 0% of mDAUs being bots.

Therefore, as jcranmer commented elsewhere in this thread, the 5% is actually their estimate of their own false negative rate of detecting bots (i.e. deciding which accounts can be monetised vs not).

Post reply on HN