Tesla CEO Elon Musk sent shockwaves through Twitter on Friday when he said he was going to “hold” his $ 44 billion acquisition of the social network while researching the ratio of fake and spam accounts on the platform.
Although Musk later made it clear that he was committed to the deal, he continued to hammer out the issue of fake accounts. He wrote on Twitter that his team would conduct their own analysis and that Twitter had expressed doubts about the accuracy of the numbers reported in its most recent financial filing.
In its first-quarter earnings report this year, Twitter acknowledged that its platform contains a number of “false or spam accounts” as well as legitimate monetary daily active users or users (mDAUs). The company reported, “We conducted an internal review of the account sample and estimated that the first quarter of 2022 represented less than 5% of our mDAU in the average quarter of false or spam accounts.”
Twitter has admitted to increasing the number of users from 1.4 million to 1.9 million in the last 3 years. The company wrote, “In March 2019, we introduced a feature that allows people to link multiple individual accounts together so that it can be easily switched between accounts,” the company wrote. “An error occurred at that time, as the steps taken by the initial account resulted in all linked accounts being counted as mDAU.”
While Mask may be justifiably curious, experts on social media, confusion and statistical analysis say he lacks the proposed methodology for further analysis.
Here, the CEO of SpaceX and Tesla says he will determine how many spam, fake and duplicate accounts exist on Twitter:
“To find out, my team will make a random sample of 100 followers of @ twitter. I invite others to repeat the same process and see what they have discovered.” He clarified his approach in subsequent tweets, adding: “Choose any account with many followers” and “Ignore the first 1000 followers, then pick every 10th. I’m open to better ideas.”
Musk added that without providing evidence, he chose 100 as the size of the sample size for his research because that number is used by Twitter to calculate the number in their earnings report.
“Any understandable random sampling process is fine. If many people independently get the same results for% of fake / spam / duplicate accounts, it will tell. I chose 100 as the sample size number because Twitter uses this to count <5%. Fake / Spam / Duplicate. "
Twitter declined to comment on whether its description of the method was accurate.
Dustin Moskowitz, co-founder of Facebook, highlighted the problem through his own Twitter account, noting that the mask method is not actually random, using a very small sample and leaving room for widespread error.
“I also think ‘Twitter doesn’t trust the team to help draw samples’ is a red flag of its own kind,” he wrote.
Christopher Bauji, founder and CEO of Botsentinel, said in an interview with CNBC that his company’s analysis indicates that 10% to 15% of Twitter accounts are probably “unauthorized”, including fake, spammers, scammers, bad bots, duplicates and “singles”. – Intentional malicious accounts “that typically target and harass individuals, and with others who intentionally spread confusion.
BotSentinel, which is primarily supported by crowdfunding, independently analyzes and identifies unauthorized activity on Twitter using a combination of machine learning software and teams of human reviewers. The company today monitors more than 2.5 million Twitter accounts, primarily English-speaking.
“I don’t think Twitter is actually categorizing ‘fake and spam’ accounts,” Bauji said.
He further warned that the number of unverified accounts may appear more or less in different corners of Twitter depending on the issues discussed. Botsentinel, for example, found more unproven accounts about politics, cryptocurrency, climate change and covid than tweeting non-controversial topics like kittens and origami.
Carl T. Bergstrom, a professor at the University of Washington who has written a book to help people understand information and avoid false claims online, told CNBC that a sample of 100 followers on a single Twitter account should not be treated as “due diligence.” For a $ 44 billion acquisition.
He said a sample size of 100 is a small order of magnitude which is ideal for social media researchers studying such things. The biggest problem with this method of masking is known as selection bias.
Bergstrom wrote in a message to CNBC, “There is no reason to believe that followers of the official Twitter account are a representative sample of the platform’s accounts. Bots are less likely to follow this account to avoid identification. ? But I don’t understand that Musk is doing anything but trolling us with this stupid sample scheme. “