This could be diplomatically summarized in this way the lively confrontation that has taken place in the past few hours between the CEO of Twitter, Agrawal, and the aspiring buyer, Elon Musk, with the latter having very badly labeled all the explanations received.
Listen to this article
The Twitter acquisition process does not seem to proceed properly on the tracks of serenity and fair play, for now suspended by the candidate buyer Elon Musk who, after having criticized the algorithms of the platform, only to be scolded by the former CEO and co-founder Jack Dorsey , also had to say about the number of fake accounts operating on the blue canary.
According to Elon Musk, it is difficult to believe that Twitter’s estimate is true, according to which only 5% of its users would be formed by bots or fake accounts and, for this reason, from the same came the proposal to his followers to conduct an experiment on the good, operating – as his team did – on a sample of 100 followers, then sharing what was discovered. It goes without saying that the more than detailed response of the current CEO of the social network, Parag Agrawal, immediately arrived.
The latter explained that, with reassuring margins of error, the estimates on fake accounts, in the last 4 quarters, were always less than 5% of mDAU monetizable daily active accounts (formally identified by the platform as the “Twitter users who have logged in or were otherwise authenticated and logged into Twitter on a given day via Twitter.com or Twitter applications capable of showing ads “).
In obtaining this calculation, Agrawal explained that, complete with human reviews performed on user samples, both public data, such as the lack of photos in profiles or the publication of strange posts, are taken into account, but also private data that , ipso facto cannot be shared, such as what the user does when active, the telephone number, his IP address, the browser-client signature, the geolocation, etc.
Similar estimates, again according to Agrawal, cannot be made by ordinary users, both because they are not even aware of how many users are counted every day as active and monetizable, but also because spam campaigns are often very complex, using bots and components. human, with many real accounts compromised to support a spam campaign, accounts that look real on the surface while not, and accounts that are very malicious even if on the surface they may seem quite legitimate.
In the case of Twitter, Agrawal, almost admitting the perfectibility of the system to fight fake accounts employed internally, revealed that the platform continuously updates its systems and rules, to remove all possible spam, without generating friction for real people or inadvertently suspend some of them but, above all (to quote another Musk idea), without having to resolve a Captcha code every time the social network is used.
Obviously Musk did not miss the opportunity for a rejoinder, indeed very concise, to the entire speech made by Agrawal, substantiated in a single icon, the poo, a mound of feces, smiling and with big eyes.