26Feb24 Roundup: Matt Mullenweg Teaches a Masterclass on Why We Have Trust & Safety Teams

We’re doing things differently today; let’s call this issue a special edition. I’m posting the main story of the Roundup today on its own and then the rest (Good News, Supportlandia content, and Bad Job Bingo) separately later this week.

Why? Well, there was some breaking news last week that I feel warrants special attention.

Matt Mullenweg, founder and CEO of Automattic,¹  made a series of progressively more terrible and baffling choices in response to accusations of transphobia in Tumblr’s moderation policies, which include some of the most egregious lapses in judgment that I have witnessed in my decade of working in Tech and 20-plus years of being a denizen of Al Gore’s internet.

It all started on February 20th when Mullenweg (who is supposed to be on a sabbatical) answered an ask on Tumblr about transmisogyny on the site with specific details of a particular moderation decision, breaking Tumblr’s very reasonable policy of not commenting on individual cases.²

He then went on to argue with Tumblr users both publicly³ and (allegedly) privately in DMs, even going as far as finding the user whose banning sparked the original controversy on Twitter and replying to one of her tweets.⁴

That terrible decision led to a predictable and ill-advised back-and-forth with the original banned Tumblr user, including Mullenweg’s absolutely jaw-dropping choice to tweet a list of the user’s other Tumblr accounts,⁵ which any reasonable Tech professional would recognize as private information.⁶

All of this culminated in a remarkable response from some of Tumblr’s trans staff in which they admirably navigated being put in the incredibly difficult position of clarifying the events leading up to the original user’s suspension while distancing themselves from the communications surrounding it, all while toeing the line of publicly holding their boss accountable without denouncing him.⁷

Now, I will say upfront that I’m not qualified to comment on the broader accusation that Tumblr’s moderation policy is transphobic or transmisogynistic, nor am I comfortable trying to comment on the original ban decision. Having worked in moderation and managed the Trust & Safety team at Khan Academy, I know first-hand how complicated and nuanced these decisions can be, and it’s impossible to make any kind of judgment without access to details to which I don’t (and shouldn’t) have access.

I will say that it’s not at all uncommon for marginalized groups in communities to identify negative trends in moderation before we as staff do because they’re more attuned⁸ to the specific injustices that affect their experiences in a community.

I will also say that moderation policies are designed by people, and people are imperfect. Community standards have to evolve over time in conversation with the community. This is how healthy communities work: members raise concerns about moderation, Trust & Safety teams do their best to find a balance between addressing concerns and protecting the broader safety and well-being of the community, and then they do it all over again the next time the community raises an issue.

It’s a never-ending, imperfect process that will never satisfy everyone. Running community platforms like Tumblr is not for the faint or uncaring of heart, and it’s not something just one person can practically do. We put together teams of experienced people – ideally resembling the community they’re moderating – because we need everyone’s collective expertise to adequately serve the community’s interests.

Which is why I’m dumbfounded by the trouble Mullenweg has gotten himself into and the harm resulting from it, because it was entirely avoidable at pretty much every turn.⁹

So, we’re going to talk about this. Not just because Automattic has a fairly high profile in CX spaces¹⁰ and I think to ignore it would amount to giving Mullenweg a pass, setting a bad precedent regarding how we expect leaders in our space to conduct themselves, but also because his behavior violates norms and standards of community practice his own company has helped set.

Let’s start with the first bad decision that set all of this in motion, which was Mullenweg answering this ask on his Tumblr:

Mullenweg founded Automattic in 2005, nearly twenty years ago. I note this to underline that he’s not new to the internet or to being the face of a company to happy and disgruntled users alike.

Mullenweg has cultivated a brand of being approachable to users, and his answering asks on Tumblr is somewhat routine.¹¹ At this point, I’d expect him to have the judgment to recognize when a question is suitable for The CEO of Automattic™ to answer (because that’s what he is whenever he’s speaking on Tumblr) and when a question needs to be referred internally for further discovery and action (if any).

This ask is very clearly the latter. This is a user who, for better or worse, is venting their frustrations at the person they think is ultimately responsible for it, but here’s the thing: this is part of being CEO of a social media platform.

Users will say things about you or your company that you think are mean, inaccurate, and/or unfair, and that’s just how it goes. The only way to refute those things with any kind of efficacy is through the collective actions of your platform. Trying to refute every accusation personally is not only pointless, but also wildly inappropriate conduct from a CEO (for reasons we’ll get into shortly).

Aside from choosing to answer the ask at all, there’s this even worse unforced error:

If you’re the CEO of a company, you can’t publicly violate your own policies. You simply can’t, for a few reasons:

1. If you, the CEO, are breaking your company’s policies to comment on an individual moderation decision that users are protesting as notably unfair, you’re reinforcing the idea that something about the case is notable enough to warrant the exception. You’re pouring jet fuel on a fire, and everybody’s going to get burned.

2. By speaking publicly about an individual moderation decision in violation of your own policies, you are undermining your Trust & Safety team. You’re indicating that you don’t trust them enough to handle what should have been a routine moderation decision without your intervention. This doesn’t just damage your relationship with your Trust & Safety team, but also their relationship with the community.

Moreover, the way Mullenweg chose to comment on this moderation decision – by giving specific examples of “death threats” from a user – is also inappropriate and ill-advised. I’m quoting Denise Paolucci, who is co-Founder of Dreamwidth Studios, because she has unique insight into this as someone who also runs a community blogging platform:

Then Mullenweg’s decisions go from worse to almost unfathomably terrible: He started to reply in the comments of his post, publicly disclosing private account information in one of them:

He chose to argue with users of his platform in another (now-deleted) ask:¹²

And, in perhaps the most egregious lapse in judgment (not to mention some of the worst behavior I’ve ever seen from a Tech CEO), Mullwenweg found the originally-banned user’s account on Twitter, began arguing with her there, and, in the process, again publicly disclosed private account information:¹³

I’m going to address his decision to publicly disclose private information about a user multiple times on different platforms separately in a moment, because, uh, holy shit.

But first, I need to make something as clear as possible: the content or accuracy of the originally-banned user’s comments (or that of any of the users Mullenweg interacted with) absolutely does not matter. 

Matt Mullenweg is the CEO of Automattic. His company owns Tumblr. It is totally inappropriate – not to mention displays a stunning misunderstanding of power dynamics – for him to interact with users this way, regardless of the circumstances.

Mullenweg has the power to ban users. As he so clearly demonstrated, he can see all the accounts associated with their IP address and he has the power to ban those as well. He has the power to see all of their friends’ accounts and ban those, too. 

He has the power to turn off Tumblr for every person on the planet, basically, so why the hell is he replying to users on his posts, answering asks about moderation decisions, and following his users to other social media platforms?

As for going on Twitter: any person who’s been the CEO of a social media platform for five years and a blogging platform for nearly 20 years should understand the weight and attention their speech draws online. It is unacceptable – and frankly dangerous – to find a user on a completely separate platform, where you have no control over moderation or safety practices, and start arguing with them there. Doing this essentially paints a target on that user’s back.

Then add the fact that he knew the user is a trans woman, with all of the additional dangers of online and physical harassment that comes with that identity, and then did it anyway? It’s negligent at best.

AND ALSO: any Tech CEO should understand on a molecular level what constitutes private account information and what doesn’t and how they are allowed to handle that information, both as a matter of company policy and in terms of legal and regulatory compliance. Mullenweg publicly disclosed private account information not once but twice on two different platforms.¹⁴

I’m not a lawyer, I have no idea what the legal ramifications of any of this could be, so I’m not going to comment on that. But combine a CEO¹⁵ commenting on an individual moderation case in violation of company policy, arguing with users on both Tumblr and a high-profile platform like Twitter, and publicly disclosing private account information multiple times in multiple places, and you have the kind of spicy recipe that should, in my opinion, result in some kind of accountability.

I would not be surprised if we learn in a few weeks that Toni Schneider has quietly transitioned from interim CEO to permanent CEO so that Mullenweg can spend more time with his family or, perhaps more likely, that Mullenweg’s sabbatical has been indefinitely extended.

Finally, before I end this, I want to say this: I know I’ve been hard on Matt Mullenweg, and for good reason. He’s bought the trouble that he’s in. 

But because this is the internet and it’s often easier to drum up mobs than empathy, I think it’s important to emphasize that everyone in this situation is an imperfect human being deserving of consideration, including Mullenweg himself. 

I don’t know why Mullenweg made the choices he’s made or what’s happening in his life that he didn’t have anyone around to tell him no, but he is clearly going through something. It’s not an excuse by any means, but I do hope someone is checking on him.

And, in case this ends up in front of Mullenweg, I’m going to offer some advice: all of this, including your final post on the matter explaining your beliefs and principles, are the justifications of a man who is more concerned that he (and an organization he's attached his identity to) is being perceived as transphobic than he is about how his actions are actually impacting people who are trans.

You can’t use Tumblr and other platforms – intentionally or not – to enable the harassment of a trans woman and then write a screed about how much you support trans people. No one is going to believe a statement like that because it’s demonstrably untrue.

Instead, you could refocus on making amends. You could take down that post and replace it with a sincere apology (that you’ve run by your Comms and Trust & Safety teams at Automattic, because there’s a reason you have them). Then you could voluntarily step back, take your sabbatical, and use it to think about ways to hold yourself accountable and spare Automattic from dealing with the fallout of your behavior. 

You could also use your sabbatical to figure out why you did all of this in the first place and, through that, find some peace. I hope you do.

Update 28Feb24:

404 Media recently reported on leaks from Automattic revealing the company’s plan to sell Tumblr and Wordpress user data to OpenAI and Midjourney for AI training, forcing Tumblr to hastily announce an opt-out feature to its users.

I mention this because it’s likely no coincidence these internal documents were leaked so close to Matt Mullenweg’s behavior last week - it indicates to me ongoing internal strife exacerbated by Mullenweg’s misconduct.

Update 23June24:

Welp, Matt Mullenweg announced his return to Automattic on May 15th, sans any mention of the events covered in this post, leading me to conclude that Mullenweg (and Tech in general) has an accountability problem.

1  Automattic famously purchased Tumblr for just $3 million in 2019 from Yahoo, who themselves bought it for $1 billion in 2013.

2  For any non-CX person who might be reading this, it’s pretty standard policy for Trust & Safety teams not to discuss individual moderation cases publicly. Policies like these protect the privacy and safety of everyone involved and help T&S teams avoid unproductive arguments about moderation decisions, which generally involve a lot of nuance that’s impossible to accurately convey without getting into details.

3  Some of these posts have been deleted, but KnowYourMeme has screenshots of the posts in its user-curated coverage of this saga.

4  I’m not linking to the tweet because it would be irresponsible to direct more attention to a user who is already at extreme risk of harassment, but you’ll find a censored screenshot of Mullenweg’s reply later in this piece.

5  This tweet has now – unsurprisingly – also been deleted, but unfortunately, as KnowYourMeme has proven, the internet is forever.

6  Not to mention any reasonable person, period.

7  It’s almost like Tumblr hired them to do this professionally and should let them do their jobs.

8  Note that I didn’t say sensitive – marginalized communities are not being “overly sensitive” to perceived slights – they’re experiencing real micro- and macroaggressions that add up to cause real pain, real exhaustion, and, as is horrifically common, real psychological and physical harm.

9  As I’m sure my gratuitous use of adverbs has signaled at this point, but goddamn, y’all.

10  Automattic is a frequent sponsor of Support Driven conferences and its Support and Trust & Safety teams are highly respected for their skill in supporting large, technical communities and platforms like Tumblr, Wordpress, WooCommerce, and more.

11  Which, by the way, I also think is absolutely fucking bonkers, but that’s a conversation for another time.

12  He also allegedly direct-messaged (DMed) users on Tumblr, but I’m not including those screenshots here because, unlike public posts, reblogs, and replies, it’s impossible for me to independently verify DMs.

13  Again, I’m censoring any info about the user because this is not about her, this is about Matt Mullenweg’s behavior.

14  I repeat: holy shit.

15  A CEO who, again, isn’t even supposed to be working right now! He’s on sabbatical!