The Tricky Business of Elon Musk Getting Twitter Fire-Hose Access - 8 minutes read
+++lead-in-text
Elon Musk’s never-ending attempt to [take over has taken yet another weird turn as the social media platform appears to have acceded to the entrepreneur’s request to gain access to a “fire hose of” internal data held by the company.
+++
For weeks, Musk has pressed Twitter to provide data that would allow the South African entrepreneur to test whether a significant share of the platform’s users are fake bot accounts—something he believes would cheapen the price he’d be willing to pay for the company. Musk contends that bot accounts make up more than 5 percent of Twitter’s user base—something even [Musk’s critics believe is wants the company to disprove that.
Twitter has reported lower numbers of inauthentic accounts in its [financial and according to [*The Washington it is willing to give Musk access to every tweet posted daily, alongside granular user information, in order to allow him to look for inauthentic behavior. (Informally, this data is called the "fire hose.” Twitter declined WIRED's request to confirm or deny the *Post* report.) Twitter’s apparent willingness to grant Musk access to the datastream comes days after the suitor’s lawyers sent [a to the company saying it was “actively resisting and thwarting [Musk’s] information rights,” and threatening to pull out of the deal.
The reported shift to grant Musk access to the data is significant, and it raises two key questions: One, will Musk get what he wants from the data he’s been given? And two: What does him gaining access mean for everyday users’ privacy and security?
For Axel Bruns, professor at Queensland University of Technology, the move is Twitter calling Musk’s bluff. “By giving him access to the fire hose, Twitter can presumably say, ‘Prove your claims about the abundance of bots, then,’” he says. Bruns believes that Musk and whoever he employs to track down bots would have a difficult time. But even for someone with the requisite skills to handle that level of data, it’s unlikely to be the right method to answer the question. It’s uncertain whether access to the fire hose of 500 million tweets posted to the social media platform every day will actually help Musk answer the key question he claims is holding up his purchase of Twitter: The proportion of users who are bots. “It seems a bit performative,” says Paddy Leerssen, a researcher in information law at the University of Amsterdam. “My sense is that this data isn’t the data you need to figure out who’s a bot or not.”
Being able to pinpoint what makes a bot a bot has been a hotly debated subject in the field of academia, one that experts have devoted much of their working lives to—which is why they’re skeptical that access to all the tweets posted to Twitter will answer the bot question definitively enough to convince Musk to go ahead with the purchase. “My impression is that people tend to overestimate how easy it is to detect bots,” says Leerssen. “A tool like this [the fire hose] isn’t going to enable you to do that, unless you combine it with all sorts of other research methods. I don’t think that’s something that in a timeline like this, Elon Musk is going to have time for.” The man who could answer how that data would help him identify bots, Musk himself, did not respond to an emailed request for comment.
Giving Musk access to the fire hose of tweets is a relatively innocuous move, says Christopher Bouzy, the founder of Bot Sentinel, a service that tracks inauthentic behavior on Twitter. “It doesn’t expose users’ private data,” he says. “It’s just a stream of tweets.” From that stream, Musk could analyze the data to see whether accounts spammed the same message, or whether a small number of accounts were responsible for the majority of tweets on the platform—both of which would be potential warning signals for bot behavior. Asked whether we should be concerned about Musk gaining access to the fire-hose data, Bouzy said no. “It’s just a massive number of tweets,” he says. And it’s also an unmanageable number of tweets for pretty much everyone outside Twitter: Bruns points out that the US Library of Congress once had fire hose access in an attempt to archive every tweet ever posted and [gave up on the interest in the fire hose data is ironic, given that he [reportedly an offer to look at Twitter’s data room—a collection of information and documents that are collated by companies when touting their businesses to potential buyers—back when his initial takeover bid was launched in April. Twitter spokesperson Jasmine Basi declined to respond to questions, including whether Musk previously asked for access to the data room. Basi also refused to answer direct questions about how many people outside of Twitter, other than Musk, have access to the fire hose data, and whether Musk would have to sign a nondisclosure or usage agreement in order to access it. That gives some cause for concern. “While I understand what Twitter is doing here, it is nonetheless also highly unusual,” says Bruns, who equates it to “giving away the crown jewels”.
The crown jewels are, however, up for sale: Around two dozen companies already have access to the fire hose of data to which Twitter has granted Musk access. Twitter’s Basi declined to name those companies, but their handling of the data has so far seemingly not caused any known issues among those dozen firms. Previously, Twitter has given broader fire hose access, with issues: The company recognized it was [leaving money on the by giving third-party data resellers access, while [spy [previously gained access to user [through Dataminr, a company that had bought fire hose access. and researchers at the [Massachusetts Institute of have previously had access to the same data Musk is now being granted access to. “The sharing of sensitive information is a key part of acquisition procedures,” says Leerssen, who prior to his career in academia worked on data protection issues around data rooms for due diligence. “When you go through the process of acquiring a company, you need to look under the hood,” he says.
Yet the big unknown is Musk himself—someone who has already shown during the course of his attempted acquisition that he’s willing to ignore legal agreements. Many also see his ostensible concern about the number of bot accounts as a pretext in order to back out of the deal, despite the terms of the agreement he came to with Twitter precluding that from happening without massive, punitive fines.
“It basically becomes more about trying to psychoanalyze Elon Musk than about the Twitter data, almost,” says Midas Nouwens, assistant professor in digital rights at Aarhus University. Nouwens has long had concerns about the scale of information available to employees at Twitter—a single tech company in a single jurisdiction in the world. “I find it a bit difficult to separate the way that social media platforms like Twitter operate, and then the rogue element of Elon Musk in there,” he says. “It very quickly becomes trying to figure out what Elon Musk would do, and his behavior is quite erratic sometimes.”
Nouwens is one of those who has access to Twitter data through an API available to researchers that allows him to query 10 million tweets a month. Gaining access to the datastream, he says, was easier than expected given the volume of data he had access to. He had to write a description of the project he intended to use the data for, alongside providing proof he was an active academic. “Whether it was Elon Musk or some other actor, I would always have concerns,” Nouwens says. “Now for him, the added element is his past behavior, sure, but also his commercial interests.” The fear is that even if Musk does pull out of the deal to buy Twitter, the information he’s been granted access to by getting free visibility over Twitter’s fire hose of tweets could be used by him or his companies in the future. Bruns says the fire-hose data set could generate new insights into who uses Twitter and why, how usage patterns change over the long term, and what problematic behaviors users engage in—alongside generating detailed profiles of users' interests and networks. “Unless there's anything in the reporting that I've missed about how long Musk gets access to the fire hose for, I'm assuming Twitter is gambling on him and his team giving up on it fairly quickly,” says Bruns. “Especially if he had the fire hose for months or longer, this really becomes a user privacy and ethics concern too.”
Regardless of whether Musk gains access to the data or not, it’s unlikely to help him overcome the main hurdle he has for his purchase of Twitter. “I don’t know if there’s an answer to the question he is posing to Twitter,” says Nouwens. “Perhaps Twitter knows more than external researchers do at the moment or can do. But it’s a really tricky question.”
Source: Wired
Powered by NewsAPI.org