Fiction vs Reality: My Texts with Michael Shellenberger
For several months earlier this year, I’d been chatting with Michael Shellenberger about the Twitter Files. He’d reached out to me in a Twitter DM on December 30th, telling me he wanted to better understand social media content moderation and transparency policy. Via phone, email, and text, we discussed topics ranging from state-sponsored disinformation campaigns, to the basics of content moderation (he was unfamiliar), to online speech. We talked about public trust on Sam Harris’ podcast. I treated him with respect, even in disagreement, regularly asking, “What do you think should be done?”
But then Shellenberger submitted testimony for a March 9 Congressional “Twitter files” hearing, and I learned we’d been having different conversations. He wrote about me extensively, mentioning me 45 times and characterizing me as a “government-funded censor” with hidden ties to the CIA. Things I’d written or said were decontextualized or applied to other topics entirely. Things I had not written or said were wrongly attributed to me or my organization. The full litany would take hours to refute and I’m not going to bother. But to briefly address the characterization: I’ve never worked at a platform or censored anything.
Details about Stanford Internet Observatory’s funding, and refutations to claims about our work in the Twitter Files are here and here; we received an NSF grant after our 2020 election and 2021 covid projects had ended and no government funding went into this work. My purported secret-agent double life was an undergraduate student fellowship at CIA, ending in 2004 — years prior to Twitter’s founding. I’ve had no affiliation since. It is additionally worth noting that Shellenberger believes that all content moderation — including labeling — is “censorship” unless the content is illegal; I disagree. I’ve long advocated against content and account takedowns in favor of the context - the counter-speech - of a label in the majority of relevant situations.
Had it just been the testimony, I would have ignored the smear. But in a second Congressional testimony, and then an appearance on Joe Rogan’s podcast, Shellenberger continues to mislead. He cited fabricated statistics and claims from a crank he recently met, and now promotes. The claims that come from this individual are not only false but absurd — i.e., that SIO somehow flagged and got Twitter to censor 22 million tweets during the 2020 election. Shellenberger’s tales get taller each time — by the time of his appearance on Rogan, the number had ballooned into hundreds of millions, and we had censored Shellenberger, his book, and Rogan himself. These are lies.
There is also innuendo — SUNY Stony Brook, where I went to college, he intones, has ties to the NSA; I don’t know if that is true, or how it is relevant, and I have no such ties. There are, once again, too many sloppy mistakes to address — ie, Shellenberger correctly notes that I did work for the Senate on Russia, then name-checks Adam Schiff, recently a Twitter files villain, who I have never done work for and who is not a Senator. Finally, Shellenberger has begun escalating into personal grievances: calling me a “snob”; speculating about whether I’m a good mom; saying he could tell that I “felt like [I] was an expert on climate”, that I thought I knew more than him despite his years of research (we never discussed climate; I don’t know much about it).
Smears like this, which throw dozens of little insinuations together, are often effective because it takes an order of magnitude more effort to refute bullshit than to put it out, and the audience often doesn’t see, or doesn’t trust, the rebuttal.
So in this post I’m publishing our full WhatsApp chat, and a wide-ranging January interview that I agreed to do with him (long version | abridged by him). It focused on content moderation and the Twitter Files; he pulled things from it in his Congressional testimony, put some of it into novel contexts, but never released it. The messages and interviews provide a view into not only our interaction, but my actual opinions about moderation and free expression.
One of the strangest parts of this situation is that despite being in regular, good-faith contact for two months, Shellenberger — who identifies as a journalist — never asked me about these ‘undisclosed CIA ties’ or ‘millions of censored tweets’ or other accusations he presented as fact. Misrepresenting opponents, turning them into caricatures, claiming they hold opinions they don’t, and ascribing false actions to them is counter to having productive debate about important issues. It does, however, likely have a positive impact on Substack subscriptions.