Judge gives Elon Musk access to bot-counting data from former Twitter executive

midian182

Posts: 9,745   +121
Staff member
What just happened? Up until now, it's looked like Twitter has been getting its own way in the legal proceedings against Elon Musk, but the world's richest man just got a bit of good news, and a bit of bad, after the judge ordered the platform to hand over documents from its former consumer product head relating to methods of analyzing fake accounts. Musk's team had requested documents from 22 individuals; the judge only granted access to one.

The crux of Musk's argument against Twitter, and the reason he backed out of the $44 billion deal, is his claim that the company lies about how many fake/bot/spam accounts populate the site. It says the figure is fewer than 5% while Musk claims it's closer to 20%.

Musk's lawyers previously asked the judge in the case to hand over the names of employees who evaluate the number of bots on Twitter. But it has only given up the names of "records custodians" who aren't as familiar with the data.

This week, Musk's team requested documents from a further 22 Twitter employees who it claims have information on the company's bot-counting methods. Delaware Chancery Court Judge Kathaleen St. J. McCormick denied access to all but one of them; the exception being Twitter's former head of consumer product, Kayvon Beykpour, who was fired in May.

As per Reuters, Musk's lawyers described Beykpour as being "most intimately involved with" the process of analyzing spam account numbers. He joined Twitter in 2015 when it acquired the now-discontinued live video app Periscope. Beykpour worked his way up to a top executive position primarily responsible for expanding Twitter's user base through new product areas before being unexpectedly fired by CEO Parag Agrawal three months ago.

Many experts believe Twitter could come out on top when the full trial against Musk starts on October 17. The judge previously granted its request for an expedited trial to take place this year instead of 2023.

Musk recently said he would complete the acquisition of Twitter at the original $44 billion price if it showed precisely how it counts the number of fake accounts on the platform. He even challenged Agrawal to a debate over the matter.

It appears that the world's richest man might not be as confident he will win the trial as he makes out. Twitter wants to force him into buying the company, and Musk has sold 7.92 million Tesla shares worth $6.9 billion to help fund a forced deal should any of his equity partners fail to come through.

Permalink to story.

 
This is some good popcorn.

The extent of Twitter's bot problem will be exposed either way. For Musk to win, he'd have to show it's been misreported. If he loses and had to go through with the purchase, he can reveal it himself.
 
If it helps dethrown twitter, I'm for it.
I get so tired of so called journalist, reporting that "breaking news...bla bla bla posted on
twitter and it has gone viral". So some politician, hollyweirdo etc says something and their
minions "like" it. So what.
 
I don't see the relevance. The definition of a bot is ultimately subjective. Whereas revenue earned to date is objective, was knowable to Musk long ago, and reflects the final collective judgment of Twitter's advertisers as to how much they are willing to pay for Twitter's total audience, humans and bots and in-between.
 
I don't see the relevance. The definition of a bot is ultimately subjective. Whereas revenue earned to date is objective, was knowable to Musk long ago, and reflects the final collective judgment of Twitter's advertisers as to how much they are willing to pay for Twitter's total audience, humans and bots and in-between.

What...?

If you are buying a company you need to know how many actual real user there are...! Advertiser need to know too... bcz kids who have 3 accounts to boost their presence do not matter to GM, Samsung, Apple, etc...

They are fake people with zero buying power...
 
What...?

If you are buying a company you need to know how many actual real user there are...! Advertiser need to know too... bcz kids who have 3 accounts to boost their presence do not matter to GM, Samsung, Apple, etc...

They are fake people with zero buying power...
No one has ever known the actual exact audience size for any media / advertising company ever in the history of the industry. It has always been an estimate. In fact it used to be a lot worse of an estimate before the modern everything-is-tracked computer era. Still, since everyone involved understands this, and if you work in media you understand the ins and outs a lot better than the average lay person, a final dollar figure is eventually agreed upon between the advertiser and the seller. That dollar figure is objective, documented, reported, and available to a perspective acquirer such as Musk. That's the important metric and it's one he has always had. His claiming he doesn't know enough now is a BS dodge IMO. And even if he gets a new method and new estimate for audience/bot size today, he can just claim he wants another new subjective estimate next week. It'll never end.
 
No one has ever known the actual exact audience size for any media / advertising company ever in the history of the industry. It has always been an estimate. In fact it used to be a lot worse of an estimate before the modern everything-is-tracked computer era. Still, since everyone involved understands this, and if you work in media you understand the ins and outs a lot better than the average lay person, a final dollar figure is eventually agreed upon between the advertiser and the seller. That dollar figure is objective, documented, reported, and available to a perspective acquirer such as Musk. That's the important metric and it's one he has always had. His claiming he doesn't know enough now is a BS dodge IMO. And even if he gets a new method and new estimate for audience/bot size today, he can just claim he wants another new subjective estimate next week. It'll never end.

Sorry, you post is just ignorant.

Yes, you can tell. It the exact reason Twitter doesn't not want to be transparent. Because anyone and everyone knows the bots on Twitter exceed 5%... and is more like 30%.... that is why Musk conservatively chose the 20% figure, to be well within legal means. (Heuristics are a real thing...)

Twitter, also have an obligation to make sure none of the people are underage. (They do not.)

BTW, You can tell the IP address from where the account was created and flag their Country.

Lastly, Companies base their advertising dollars on HOW MANY people in their target audience will see the advert. If Twitter is lying to them, means that Twitter is over-valued & Elon is doing his due diligence.

Imagine if ESPN lied to advertisers, of how many viewers they have or people are watching the sporting events, and these companies were paying millions more.

Twitter is using fake numbers, even the old CEO (Jack) said so. So there is zero reason for you to defend Twitter. We already know the facts, we are just trying to get Twitter to admit them..


IMO, the mere fact you are defending Twitter implies you have multiple accounts and don't want to loose them under Elon..
 
It's clear you believe that there is some magic way that media companies know exactly how many real people saw an advertisement. But this is simply not true. As you say, everyone knows there are bots, as well as ad blockers, fake survey responses, shared accounts, shared devices, vpns, proxies, auto-refreshing pages, and many more sources of doubt. There is no one human who knows an objective answer and even if there was the next human might disagree on what the rules should be for that objective answer.

I am not "defending" Twitter or claiming there aren't buttloads of bots or many other reasons that ads are worth a lot less than a gross audience count might suggest. What I am simply saying is that none of this is new information to anyone who works in advertising for a living, and all of it was already built into the price anyone was willing to pay, especially when you're looking at aggregate numbers like total revenue, which reflects the total industry consensus of what the level of BS to real users really is. In the end, that's the only measurement that is real.

btw ESPN has no idea what their audience size is either, and advertisers wouldn't take their word for it anyway. That's why there are third party audience measurement companies (I.e., Nielsen). Those companies can't provide exact answers either but at least they serve as an accepted currency again with everyone involved understanding the many limitations and inaccuracies.
 
It's clear you believe that there is some magic way that media companies know exactly how many real people saw an advertisement. But this is simply not true. As you say, everyone knows there are bots, as well as ad blockers, fake survey responses, shared accounts, shared devices, vpns, proxies, auto-refreshing pages, and many more sources of doubt. There is no one human who knows an objective answer and even if there was the next human might disagree on what the rules should be for that objective answer.

I am not "defending" Twitter or claiming there aren't buttloads of bots or many other reasons that ads are worth a lot less than a gross audience count might suggest. What I am simply saying is that none of this is new information to anyone who works in advertising for a living, and all of it was already built into the price anyone was willing to pay, especially when you're looking at aggregate numbers like total revenue, which reflects the total industry consensus of what the level of BS to real users really is. In the end, that's the only measurement that is real.

btw ESPN has no idea what their audience size is either, and advertisers wouldn't take their word for it anyway. That's why there are third party audience measurement companies (I.e., Nielsen). Those companies can't provide exact answers either but at least they serve as an accepted currency again with everyone involved understanding the many limitations and inaccuracies.

Twitter is not a media company, it's a social network.
You have to have an individual account, to be on it. Individual...

Before you start to lecture anyone, just go to Joe Bidens page and start looking at his followers. U can scroll down for HOURS... and not find a single legit user... out of millions.

This is a fact and is a running joke at work, when someone actually find a real person. Go try yourself.


An outside research firm already did the numbers... you are acting like you don't know about this, then trying to sound advertisy.. and not the actual Product Musk is buying. Valid users. Elon will take it private and might just charge $50/year to use... and have very little advertising.

Twitter users are overhyped. And the richest man in the world can't get the company he trying to buy, to allow him to see the actual numbers... that Twitter doesn't want to release.

That is the whole point of this lawsuit and we the deal is put on hold...
 
Twitter is not a media company, it's a social network.
You have to have an individual account, to be on it. Individual...
...

An outside research firm already did the numbers... you are acting like you don't know about this, then trying to sound advertisy.. and not the actual Product Musk is buying. Valid users. Elon will take it private and might just charge $50/year to use... and have very little advertising.

Twitter users are overhyped. And the richest man in the world can't get the company he trying to buy, to allow him to see the actual numbers... that Twitter doesn't want to release.

Twitter is an advertising company and a social network. Same with Facebook/Meta. If both companies derive 99% (Meta/FB) and 89% (Twitter) of their income from advertising, then they are advertising companies.

Do subscribers count matters? Yes. Do real subscribers count matters? Yes. At the end of the day, companies still spent $4.5B on Twitter ads in FY2021 regardless of # of bots which is all that matters: advertising revenue which is major part of the equation when you value the company.
 
Twitter is not a media company, it's a social network.
Elon will take it private and might just charge $50/year to use... and have very little advertising.
Well the current business model is media, and that's how it's currently valued by the public markets. But fine, it was of course completely up to Musk to bid whatever he wanted using whatever business model valuation he saw fit, and as sole owner if he wants to change its business model that would be up to him. But if he really is contemplating ditching advertising and switching to a subscription basis, worrying whether it is 20% or 30% bots will be hardly a rounding error as far as estimating how many of the current free users are willing to stay on as expensive paid users. My guess at the number of humans willing to pay $50 / year to be on Twitter is no higher than 10% of the current (non-bot) base, and under 1% would not surprise me at all. I personally value my Twitter account at just more than the few minutes of my time it would take me to find the unsubscribe and delete buttons.

btw I'm pro-Musk as far as him wanting to walk away. I don't feel he should have to buy the company if he doesn't want to, whatever the reason. The deal included a $1B breakup fee, let him just pay it so everyone can move on. Personally I don't believe he was ever that serious, I think he treated this whole thing like a joke and wasted a lot of people's time, but that's not a crime, just a jerk move.
 
No one has ever known the actual exact audience size for any media / advertising company ever in the history of the industry. It has always been an estimate.
On the contrary, print media has always has a precise count of paid subscriptions that is independently verified and audited. Broadcast media figures were somewhat less accurate, but were still independently audited and verified (aka the Nielson Ratings and other systems).

Twitter, in contrast, refuses to allow any such independent measurement, and not only insists that their own figures be used, but also shrouds the data and methodology behind those figures in secrecy and obfuscation.
 
This is some good popcorn.

The extent of Twitter's bot problem will be exposed either way. For Musk to win, he'd have to show it's been misreported. If he loses and had to go through with the purchase, he can reveal it himself.
One has to wonder if Musk does get the company and does find that the bot count is significantly higher, can he go back and sue? If the value of the company is partly based on percentage of bot accounts and the current management has tried to obscure the actual numbers, it seems like there is a basis for fraud. Of course, I'm not lawyer and I do not play on on TV, so my legal advice is priceless, literally.
 
This is some good popcorn.

The extent of Twitter's bot problem will be exposed either way. For Musk to win, he'd have to show it's been misreported. If he loses and had to go through with the purchase, he can reveal it himself.
Misreported assumes that they are intentionally hiding the real numbers. There could be many loopholes that would still force Musk to complete the transaction, like a difference methodology in how bots are counted (when and how they're added to the count, like for example after 10 or 100 posts were made by that account - that number makes a huge difference) or the very definition of what a bot is for twitter.
 
Well the current business model is media, and that's how it's currently valued by the public markets. But fine, it was of course completely up to Musk to bid whatever he wanted using whatever business model valuation he saw fit, and as sole owner if he wants to change its business model that would be up to him. But if he really is contemplating ditching advertising and switching to a subscription basis, worrying whether it is 20% or 30% bots will be hardly a rounding error as far as estimating how many of the current free users are willing to stay on as expensive paid users. My guess at the number of humans willing to pay $50 / year to be on Twitter is no higher than 10% of the current (non-bot) base, and under 1% would not surprise me at all. I personally value my Twitter account at just more than the few minutes of my time it would take me to find the unsubscribe and delete buttons.

btw I'm pro-Musk as far as him wanting to walk away. I don't feel he should have to buy the company if he doesn't want to, whatever the reason. The deal included a $1B breakup fee, let him just pay it so everyone can move on. Personally I don't believe he was ever that serious, I think he treated this whole thing like a joke and wasted a lot of people's time, but that's not a crime, just a jerk move.
You are loosing perspective.

It not based on whether or not he wants to buy the company....it's 100% based on how many real user have accounts with twitter. Desire is not what is holding the deal up... it is Twitter NOT WANTING to release actual data.

Secondly, you are correct, current users may NOT want Twitter to go Private and have to pay... because most of them would have to pay PER ACCOUNT and they have many accounts. But People like me and many others would move to Twitter, if it was private company, because we KNOW that fake people will be weeded-out and tweets will be based on RL people, not pushed agendas from a boiler room.

If musk walks away from the Deal, he won't have to pay a dime.... because he already fulfilled his obligation and it is Twitter who has squelched the deal. They are not being forthright and will not give the prospective owner legit accounting.

And yes Elon is serious.... only naive people think it was a joke, because they can't believe anything else. They do not want to believe that Twitter might go private. They do not want to believe it might cost them... or how can they afford $200/year (or a paid based subscription)... to spread their lies.

Elon gave the an inflated offer... and they are balking bcz they do not want the truth to come out. 20% or more of active users on Twitter are fake accounts... everyone knows this. The research has already been done we are just waiting for Twitter CEO to admit it, so the deal can move on.
 
You are loosing perspective.

It not based on whether or not he wants to buy the company....it's 100% based on how many real user have accounts with twitter. Desire is not what is holding the deal up... it is Twitter NOT WANTING to release actual data.

Secondly, you are correct, current users may NOT want Twitter to go Private and have to pay... because most of them would have to pay PER ACCOUNT and they have many accounts. But People like me and many others would move to Twitter, if it was private company, because we KNOW that fake people will be weeded-out and tweets will be based on RL people, not pushed agendas from a boiler room.

If musk walks away from the Deal, he won't have to pay a dime.... because he already fulfilled his obligation and it is Twitter who has squelched the deal. They are not being forthright and will not give the prospective owner legit accounting.

And yes Elon is serious.... only naive people think it was a joke, because they can't believe anything else. They do not want to believe that Twitter might go private. They do not want to believe it might cost them... or how can they afford $200/year (or a paid based subscription)... to spread their lies.

Elon gave the an inflated offer... and they are balking bcz they do not want the truth to come out. 20% or more of active users on Twitter are fake accounts... everyone knows this. The research has already been done we are just waiting for Twitter CEO to admit it, so the deal can move on.
I think Musk is serious about buying, but I'm not so sure he wants to take it private. That would kind of defeat the purpose, IMHO. Making it private will drive many people, regardless of political leanings, away from the platform.

I also think that this is Musk trying to negotiate the price downward, possibly due to his investors being nervous about actual value of the company. I honestly do not understand why Twitter is so reluctant to share the data. It makes them look guilty, even though they may not be purposefully obscuring the true bot numbers.
 
it is Twitter NOT WANTING to release actual data.
Please describe this actual data that Twitter has and is not releasing.

It was a long time ago when I signed up for my one Twitter account, but I do not recall a checkbox or radio button asking if I was a human or bot. I certainly did not pay a fee nor verify my identity in any way. (I don't recall if there was a captcha back then but it wouldn't prove anything if there was.) Because of all this I do not believe a simple database query exists along the lines of "SELECT count(*) FROM users WHERE bot = true".

Without such a straightforward objective definition, you're back to the subjective inference I've been talking about all along. Are you going to base it on number of logins? Post count vs read count? Time spent per day? (Please tell me you already understand how useless IP address would be for this.) None of those are direct matches for human vs bot and all are easily manipulable the second that Twitter or anyone else announces it is going to base some important metric on them.

If I was a judge, I'd want Musk to explain to me exactly what material information he expected to be told that was not available during the various earlier phases. If he wanted to commission five top notch research firms to each prepare their estimates of humans vs. bots, he could have at any point, including before this all started, and each might produce new information vs. what Twitter itself knows. Insisting this late in the game that there is secret, new data only available to Twitter, that is different from all other available information before now, and that is more objective and more meaningful than the revenue data which was disclosed, does not feel like an easy sell to me but of course he'll have his day in court.

users may NOT want Twitter to go Private and have to pay... because most of them would have to pay PER ACCOUNT and they have many accounts. But People like me and many others would move to Twitter, if it was private company, because we KNOW that fake people will be weeded-out and tweets will be based on RL people, not pushed agendas from a boiler room.
Are you trying to claim that individual humans will pay more than say a corporation or a political campaign to get their message out? Because I don't know a single individual who owns a billboard for personal use, or has bought a major network TV ad, but I know lots of companies, brands, and campaigns that have.

While I think your reasoning is way off, I do agree with your bottom line: a private, for-fee might have much less marketing spam on it, because the value of its audience would be much, much less (since it will be 1-2 orders of magnitude smaller.)

go to Joe Bidens page and start looking at his followers. U can scroll down for HOURS... and not find a single legit user
I missed this line first time I read your post, and I regret it because this whole time I thought I was having a business / valuation / advertising discussion but I realize now your interest in this may be largely political, such as wanting to believe that MAGA is actually the majority view in the US and all the competing views are made up ads from a handful of "elites".

Well we do agree on something: I too expect that Biden's follower count is heavily padded. Where I think you may be kidding yourself is if you do not believe the same goes for every other politician, brand, celebrity, product, etc. on Twitter and on every other social network.
 
Please describe this actual data that Twitter has and is not releasing.
It's less "data" and more about the process they use to determine bots on Twitter.
It was a long time ago when I signed up for my one Twitter account, but I do not recall a checkbox or radio button asking if I was a human or bot. I certainly did not pay a fee nor verify my identity in any way. (I don't recall if there was a captcha back then but it wouldn't prove anything if there was.) Because of all this I do not believe a simple database query exists along the lines of "SELECT count(*) FROM users WHERE bot = true".
No, it's not a simple database query. There are commercial services that can estimate numbers of bots on Twitter. Bot Sentinel, Botometer and BotSlayer are 3 that Google came up with.
Without such a straightforward objective definition, you're back to the subjective inference I've been talking about all along. Are you going to base it on number of logins? Post count vs read count? Time spent per day? (Please tell me you already understand how useless IP address would be for this.) None of those are direct matches for human vs bot and all are easily manipulable the second that Twitter or anyone else announces it is going to base some important metric on them.
I don't have the precise knowledge about how the algorithms count bots, but they do count bot accounts. That is to say they give an estimation as to whether an account is a bot account. It's not perfect and opinions vary on how accurate these apps are.

And, I hate to tell you, but Bolometer actually publishes their info. Here is what they say on their site:

"We believe that transparency is key for building a bot detection tool that's useful for everybody. To maintain the transparency of Botometer, we release all the training datasets in the Bot Repository together with other public bot datasets. Although we are unable to open source Botometer, the technical details of its features, training, machine learning models, and accuracy can be found in our peer-reviewed publications, allowing others to reproduce our work."
If I was a judge, I'd want Musk to explain to me exactly what material information he expected to be told that was not available during the various earlier phases. If he wanted to commission five top notch research firms to each prepare their estimates of humans vs. bots, he could have at any point, including before this all started, and each might produce new information vs. what Twitter itself knows. Insisting this late in the game that there is secret, new data only available to Twitter, that is different from all other available information before now, and that is more objective and more meaningful than the revenue data which was disclosed, does not feel like an easy sell to me but of course he'll have his day in court.
He is questioning Twitters claim of 5% Bots. He very likely has those reports from other Services or analysts and is now trying to reconcile what Twitter is saying versus what his other sources are saying. Why is Twitter refusing to explain how they calculate the 5% number? They should be ready to produce that info, if for no other reason than to force Musk into the deal. But, they aren't producing that info. Why?

Are you trying to claim that individual humans will pay more than say a corporation or a political campaign to get their message out? Because I don't know a single individual who owns a billboard for personal use, or has bought a major network TV ad, but I know lots of companies, brands, and campaigns that have.
You don't know a single politician that has bought TV and Radio ads? Local politicians do this all the time. And some don't spend very much to do so. You're right, that most individuals won't pay more than a corporation to get advertising, but that's not to say there aren't a lot of individuals that do spend money on advertising. Billboards? Ever see a lawyer's billboard? Now you may say that's a "corporation" but the lawyer and the corporation are many times one and the same.
While I think your reasoning is way off, I do agree with your bottom line: a private, for-fee might have much less marketing spam on it, because the value of its audience would be much, much less (since it will be 1-2 orders of magnitude smaller.)
I honestly don't believe anyone wants to take Twitter private in the sense of making it a pay service. They can still be a "privately" owned company and be free to use while attempting to earn a profit.
I missed this line first time I read your post, and I regret it because this whole time I thought I was having a business / valuation / advertising discussion but I realize now your interest in this may be largely political, such as wanting to believe that MAGA is actually the majority view in the US and all the competing views are made up ads from a handful of "elites".

Well we do agree on something: I too expect that Biden's follower count is heavily padded. Where I think you may be kidding yourself is if you do not believe the same goes for every other politician, brand, celebrity, product, etc. on Twitter and on every other social network.
The difference is that Joe Biden's "value" isn't determined by an inflated follower count. His value is determined at the ballot box and as such he had sufficient votes to win the election. Twitter, on the other hand, is partly evaluated based on the percentage of bot accounts there are. How they asses their bot accounts is not only important to Musk, but also anyone that buys their stock or any company that wants to advertise on Twitter.
 
He is questioning Twitters claim of 5% Bots. He very likely has those reports from other Services or analysts and is now trying to reconcile what Twitter is saying versus what his other sources are saying.
This is the crux of my original point. I'd question 5% Bots too. But given that all those reports from other services and analysts were readily available to Musk all along, and more importantly to everyone who has ever advertised or considered advertising on Twitter, how can Musk plausibly claim that his understanding of Twitter's bot situation is materially different now than it was at any prior point in the bid & diligence process?

My point is not that Twitter's definitions are correct or useful. My point is that no material change has occurred. Musk attempting to justify a back-out now as if he somehow knew less about Twitter's bots than say the posters on this forum until recently does not pass the smell test.

You don't know a single politician that has bought TV and Radio ads?
That's what I meant by campaigns which I mentioned later in the sentence. They are large advertisers in their season, but are usually not backed by the budget of a single individual.
 
Back