The analysts Elon Musk hired to count Twitter bots found fewer than he expected, claims...

midian182

Posts: 8,303   +103
Staff member
What just happened? The latest argument in the Twitter vs. Elon Musk fight has seen the social media company claim the two analyst firms Musk hired to count bots on the platform found fewer than he expected. Twitter lawyers also say the CEO's legal team has, until recently, been withholding this data.

The world's richest man has long claimed he abandoned the Twitter deal because it lied about the number of accounts that are fakes/bots. Twitter says it's around 5%; Musk claims it could be higher than 20%.

On May 13, Musk tweeted that the Twitter deal was on hold while he investigated Twitter's 5% claim. A month later, Musk said the acquisition was in "serious jeopardy" over the issue. He hired two analysis companies to look into it: Cyabra, which estimated the number of fakes/bots to be 11%, and CounterAction, which said that bot accounts made up 5.3% of all users. Musk abandoned the deal a couple of days later.

A Twitter lawyer (via Insider) said the findings are "very much in line with Twitter's claims," adding that "None of these analyses remotely supported what Mr. Musk told the Twitter parties and told the world in the termination letter he served up on July 8."

The lawyer also claims that Musk's legal team has been purposely withholding the findings. Twitter's representatives have been trying to see the information since the lawsuit began, but it was only made available yesterday (Tuesday).

"If there are any analyses that exist that actually substantiate what Mr. Musk told Twitter and told the world, they certainly have not been produced in discovery in this case," the lawyer said.

CounterAction's 5.3% figure does line up with Twitter's numbers. And while Cyabra's 11% is lower than Musk's belief that over a fifth of Twitter accounts are fake, it's still more than double what the company claims.

The five-day Twitter/Musk trial begins on October 17. The company wants the acquisition to complete for the original $44 billion price, while Musk wants to walk away without paying the $1 billion termination fee. There were rumors earlier this week that a settlement might have been reached after a deposition meeting involving Musk and Twitter CEO Parag Agrawal was canceled, but a new deposition is scheduled for today.

Permalink to story.

 

Eldritch

Posts: 496   +894
What's the real number?

Well, it will depend on hashtags topic I.e. Politics and celebrity hashtags will have huge concentration of fake accounts.
An hashtag about simple and regular matters of life may not have large concentration of bots.
The trending topics will almost certainly have huge number of fake accounts as it's incredibly difficult to trend without fake accounts.
 

ScottSoapbox

Posts: 423   +781
Are these % users or active users, because that's an important difference.

(Twitter itself uses yet another non-standard user measure that reflects advertising potential rather than activity).
 

Uncle Al

Posts: 9,275   +8,435
The story is ever changing. We heard claims of 20% bots and fake accounts, then Twitter refused to share the data & was accused of purging the bots and fake accounts before showing their data, bla, bla, bla.

At this point I think the judge should dismiss the case and fine each party 1 billion dollars for wasting the courts time.
 

m4a4

Posts: 3,082   +4,119
TechSpot Elite
I mean, this sounds like it was counted after Twitter had a purging spree. Some of the numbers could've been correct at their respective times...

I would 100% expect that Twitter diverted resources since the initial claims to bring them down.
 

tkabou

Posts: 143   +171
There was an interview with either a guy ex-CIA or ex-Mil, and he said with a straight face and conviction the platform, somehow by design is at least 40% bots. Shocking revelation but we’ll have to wait for the October surprise when the case gets decided.
 

DSirius

Posts: 366   +759
TechSpot Elite
Elon Musk should ask those experts to make a bot who will count the bots from Twitter. Thus, all of them will have a more clear picture of how things works on Twitter :)
 

umbala

Posts: 768   +1,474
What's the real number?

Well, it will depend on hashtags topic I.e. Politics and celebrity hashtags will have huge concentration of fake accounts.
An hashtag about simple and regular matters of life may not have large concentration of bots.
The trending topics will almost certainly have huge number of fake accounts as it's incredibly difficult to trend without fake accounts.
Trending and hashtags have nothing to do with it. A fake account is a fake account, no matter how it is used.
 

umbala

Posts: 768   +1,474
When the people doing the tally are against you there is little chance
Right, right, because *everyone* is against you, that's why you can't win elections! So the solution is to replace the people doing the tally with hardcore right wing nutjobs, election deniers, and conspiracy theorists. Conservatives have become absolute masters at the blame game and playing the perpetual victim. This is what America is now.
 

axiomatic13

Posts: 413   +480
Right, right, because *everyone* is against you, that's why you can't win elections! So the solution is to replace the people doing the tally with hardcore right wing nutjobs, election deniers, and conspiracy theorists. Conservatives have become absolute masters at the blame game and playing the perpetual victim. This is what America is now.
Dude, my junk can only get so hard!
 

RudyBob

Posts: 815   +801
Right, right, because *everyone* is against you, that's why you can't win elections! So the solution is to replace the people doing the tally with hardcore right wing nutjobs, election deniers, and conspiracy theorists. Conservatives have become absolute masters at the blame game and playing the perpetual victim. This is what America is now.
Sure, Sure
 

Eldritch

Posts: 496   +894
Trending and hashtags have nothing to do with it. A fake account is a fake account, no matter how it is used.
True, but where you will look for it? The sample size can not be everything so the selection is important. What is the sample will dictate what percentage of fakes accounts are noted.
Let me elaborate with the data cited in this article itself that "Cyabra, which estimated the number of fakes/bots to be 11%".
Now the Cyabra did so by using a hashtag as sample. They used #Depp for analysis from Johnny Depp and Amber trial. Therein they got the figure of 10.8% fake accounts.
 

brucek

Posts: 1,284   +1,900
People get caught up in the precise sounding number part ("5 - 20%") while ignoring the chasm of vagueness that exists between a "real" account vs a "fake/bot" account.

Everyone probably agrees that a human using only a single account for personal entertainment is a "real" user, and a large group of fully automated accounts controlled only by computer are all fakes/bots. But that still leaves lots of in between. Is the bar owner who logs in once each morning only to post which band is playing that night real or fake? Does it change if he has a separate account for personal use? What if he has a third account for the bar's softball team, is that real or fake? What if despite being at work he sometimes ends up reading and replying to tweets on his bar account? What if his personal account is only ever used on an ad-blocked PC and has never received a single ad?

Even within bot armies, there are differences. A large group of bots that was created solely to boost follower count and will never log in again might affect total users but isn't affecting total usage much. Meanwhile a single bot designed to spam a message over and over isn't adding many accounts but is generating a lot of fake traffic.

The % number is next to useless without a specific definition behind it, but measurement is already hard even without a definition and gets harder the more you actually try to be specific.