WordPress founder Matt Mullenweg says the New York Post’s 2020 story on Hunter Biden forced parent company Automattic to weigh difficult moderation questions for the WordPress VIP hosting service — although the company concluded it shouldn’t take any action on the story.
In October 2020, the Post ran a report based on material it said came from an abandoned Biden laptop. Facebook and Twitter temporarily restricted links to the story as questions about the provenance of the laptop’s content swirled: Facebook limited its spread under potential misinformation rules, while Twitter placed a more extreme ban on links to it, citing rules against posting hacked materials. (It had previously enforced those rules against leaked police department data.) But at the time, there was little discussion of one key platform: the New York Post’s content management system, WordPress VIP.
“Twitter decided to remove links to the story to the New York Post. Guess who hosts the New York Post? We do,” Mullenweg told The Verge’s Nilay Patel on Decoder. The Post is one of the best-known users of WordPress VIP, a web hosting platform geared toward major brand websites and media outlets. WordPress VIP is based on the open-source WordPress software, but it’s a commercial service run by Automattic and (like the more general-purpose WordPress.com) clients are bound by Automattic’s terms of service.
So as Twitter and Facebook were hashing out whether to let the story spread, Automattic was having its own internal discussions about whether the Post had violated WordPress VIP rules. “It maybe skirts some of the rules we have against non-consensual hacked material, but it also fits in with these other rules, including being a major tier journalism organization and a public interest,” Mullenweg says the team concluded. “We made a decision there to not touch it.”
“The interpretation of the policies is really where I think the art and science of it is.”
Mullenweg says it’s not surprising the topic came up. “There is always a discussion and there are reports. People contact us saying, ‘Take this down,’ or ‘This is violating your policy.’” He calls the policies, however, a “starting point” for real moderation. “The interpretation of the policies is really where I think the art and science of it is. We will also make mistakes. We’ve accidentally taken down blogs, either by some script that went wrong or by a human who clicked the wrong button or made a mistake interpreting our policies. It’s all about how you fix it,” he says. Twitter, for its part, did conclude its rules on hacked material (regardless of whether the Biden laptop counted as “hacked”) were wrong.
Mullenweg is skeptical of claims — often made by Republican politicians and pundits — that political speech is getting routinely suppressed. “Maybe we all need to say that this is actually working right now, or perhaps we should question the framing in the first place — that there’s something fundamentally broken or wrong here,” he says. “The current system will make mistakes. It’s not perfect but gets to correctness pretty quickly, usually within a matter of hours or days, not weeks or months.”
It’s an interesting reminder, though, of how expansive the boundaries of content moderation are. Automattic has spent the last year publicly reevaluating the rules of Tumblr, which it bought in 2019. It’s less intuitive to imagine similar rules applied to a news outlet that predates the internet or a web platform that’s far from a conventional social network. But online, everything is content — so it’s not surprising that WordPress VIP is grappling with the same questions facing Facebook and Twitter.