Tesla CEO musk said on Friday that he would "shelve" the $44 billion acquisition of Twitter and study the proportion of false and spam accounts on the platform. Although musk later clarified that he was still committed to the deal, he continued to attack the issue of false accounts. He wrote on twitter that his team would conduct its own analysis and questioned the accuracy of the figures reported by twitter in its recent financial documents.
"Twitter said I leaked the sample size of the robot and violated the confidentiality agreement," said Elon Musk, CEO of Tesla.
In its first quarter earnings report, twitter admitted that there were many "fake or spam accounts" on its platform. The company reported that "We conducted an internal review of account samples and estimated that the average value of false or spam accounts in the first quarter of 2022 was less than 5% of our mdau in the current quarter."
Twitter also admitted that it had exaggerated the number of users between 1.4 million and 1.9 million over the past three years. "In March 2019, we launched a feature that allows people to link multiple separate accounts together to easily switch between accounts," the company said. Twitter disclosed that "an error occurred at that time, so the action taken through the master account resulted in all related accounts being counted as mdau."
Although musk may have reason to be curious, experts in social media, false information and statistical analysis said that the feasibility of the analysis method he proposed was seriously insufficient.
Without providing evidence, musk randomly selected 100 accounts as the sample size of his study and thought that this would calculate the percentage of false / spam / duplicate accounts.
When asked whether his description of his method was accurate, twitter declined to comment.
Facebook co-founder Dustin Moskowitz weighed this issue through his own twitter account, pointing out that Musk's method is not actually random, uses too few samples, and leaves room for a large number of errors.
Christopher bouzy, founder and CEO of botsentinel, said in an interview that his company's analysis showed that 10% to 15% of accounts on twitter may be "untrue", including counterfeiting, spammers, fraudsters, malicious robots, duplication and "single purpose hate accounts", which usually target and harass individuals and other people who deliberately spread false information.
Botsentinel is mainly supported by crowdfunding. It uses a combination of machine learning software and manual audit team to independently analyze and identify untrue activities on twitter. The company today monitors more than 2.5 million twitter accounts, mainly English users.
Carl T. Bergstrom, a professor at the University of Washington, said that the sample size of 100 was several orders of magnitude smaller than the standard for social media researchers to study such things. The biggest problem musk faces when using this method is called selection bias. He can't understand what musk can do except deceive us with this stupid sampling scheme.