Tesla CEO Elon Musk sent Twitter shares down on Friday when he said he was going to “wait” for the $ 44 billion acquisition of the social network while investigating the proportion of fake and unwanted accounts.
Although Musk later clarified that he remains committed to the deal, he continued to whistle over the issue of bogus accounts. He wrote on Twitter that his team would do its own analysis and expressed doubts about the accuracy of the numbers mentioned by Twitter in its most recent financial statements.
In its first-quarter earnings report this year, Twitter acknowledged that there are a number of “fake or unwanted accounts” on its platform, along with legitimate daily active use or revenue-generating users (mDAUs). The company said: “We have conducted an internal review of a sample of accounts and estimate that the average of false or unwanted accounts in the first quarter of 2022 represented less than 5% of our mDAU during the quarter.”
Twitter also admitted to overestimating user numbers by 1.4 to 1.9 million users in the last 3 years. The company wrote: “In March 2019, we launched a feature that allowed people to link multiple different accounts together for easy switching between accounts,” Twitter revealed. “An error occurred at this time, so the actions performed through the master account resulted in all linked accounts being counted as mDAU.”
While Musk may be justifiably curious, experts on social media, misinformation and statistical analysis say his proposed approach to further analysis is sadly flawed.
See what the CEO of SpaceX and Tesla said he would do to determine how many unwanted, fake and duplicate Twitter accounts there are:
“To find out, my team will make a random sample of 100 @twitter followers. I invite others to repeat the same process and see what they find out.” He clarified his methodology in subsequent tweets, adding: “Choose any account with many followers” and “Ignore the 1000 followers first and then choose every 10th. “I’m open to better ideas.”
Musk also said, without elaborating, that he chose 100 as the sample size number for his study because that is the number that Twitter uses to calculate numbers in its earnings reports.
“Any logical random sampling procedure is good. If many people independently receive similar results for% of fake / spam / duplicate accounts, this would be indicative. I chose 100 as the sample size number because it uses Twitter to calculate <5 % fake / unwanted / duplicate. "
Twitter declined to comment when asked if the description of its methodology was accurate.
Facebook co-founder Dustin Moskovic tackled the issue via his Twitter account, noting that Musk’s approach is not really random, uses a very small sample and leaves room for huge mistakes.
He wrote, “I also feel that ‘I do not trust the Twitter team to help take the sample’ is its kind of red flag”.
BotSentinel founder and CEO Christopher Bouzy told CNBC in an interview that his company analysis shows that 10% to 15% of Twitter accounts are likely to be “non-genuine”, including counterfeiters, spammers, scammers, malicious robots, duplicates and “mono-hate accounts” that usually target and harass individuals, along with others who spread deliberate misinformation.
BotSentinel, which is mainly supported through crowdfunding, independently analyzes and tracks non-genuine Twitter activity using a mix of machine learning software and groups of human reviewers. The company currently monitors more than 2.5 million Twitter accounts, mostly English-speaking users.
“I do not think Twitter realistically ranks ‘fake and unwanted’ accounts,” Bouzy said.
It also warns that the number of non-genuine accounts may appear higher or lower in different corners of Twitter, depending on the topics being discussed. For example, more authentic tweets about politics, cryptocurrencies, climate change and Covid than those that discuss controversial topics such as kittens and origami, BotSentinel found.
Carl T. Bergstrom, a professor at the University of Washington who wrote a book to help people understand the data and avoid making false claims on the Internet, told CNBC that sampling one hundred followers of any Twitter account should not be used as a pretext. “Due diligence” to make a $ 44 billion takeover.
He said a sample size of 100 is an order of magnitude smaller than the norm for social media researchers studying such things. The biggest problem Musk will face with this approach is known as choice bias.
Bergstrom wrote in a message to CNBC, “There is no reason to believe that the followers of the official Twitter account are a representative sample of accounts on the platform. Bots may be less likely to follow this account to avoid detection. “Who knows? But I can not understand that Musk is doing anything other than trolling us with this silly sampling plan.”