Illocution: Why Regulating Trump’s Speech is So Hard

Watching Trump try to overthrow Biden’s victory on Twitter, such as in tweets like this…

… it occurs to me why the warnings and other content moderation rules constructed by Twitter (and other social media) seem to miss the mark. These rules focus on what is being said, not who is saying it.

For example, labeling the claim about fraud “disputed” — while true — seems beside point here. Yes, there is technically misinformation here — it is not true that “all” states “have found massive voter fraud.” But this assertion is hardly the point of the tweet. This tweet is a threat by the nation’s chief executive to legally punish those who do not use their legal powers to award him victory. That is, the tweet itself an attempt to overturn the election by extra-legal means. The tweet is not signaling Trump’s intention to overturn the election, or promoting a theory that supports it being overturned, it is the attempt to do so.

In philosophy of language there is an idea called the “illocutionary act.” An illocutionary act is a statement that performs something — like an officiant saying “I now pronounce you man and wife” performs the act of marriage. Making a promise is another kind of simple illocutionary act. Critical to the implication, and thus underlying meaning, of an illocutionary act is the position / powers of the speaker. The marriage is official because the officiant says it, not because someone says it or it is said in general. Similarly, Donald Trump’s tweets carry all kinds of illocutionary force because he is the President of the United States.

So President Trump is using Twitter to attempt a coup. Whether you support this attempt or not, it’s clear that Twitter is in an unenviable and untenable position. There is no a pre-existing content moderation policy for a situation like this. The harm is not in the meaning of the statement, but in the implications drawn from the fact that the statement was made by the chief executive.

I thought of this hypothetical example to shed more light. Here’s an event that I think is reasonably likely. On January 20th, during Biden’s inauguration, Trump tweets out something like “I don’t care about this inauguration, I AM STILL IN CHARGE!!” If he sends this tweet before Biden is sworn in, the claims — “I don’t care” and “I AM STILL IN CHARGE!!” — are both true. If there is no follow through from anyone, it’s hard to point to specific harms. If there is follow through, then the claim has some truth.

I’m not exactly sure how Twitter and friends get out of this kind of pickle. One thing that’s clear is that by providing the blue check mark to celebrities, they solved one problem — improving authenticity — but created another — increasing the illocutionary force of tweets. If we didn’t really know if that was Donald Trump sending those tweets, they would have less force.

But the bigger principle is that platforms should abandon the idea that there can be the same rules for everyone. Some statements, by some individuals, have enormous illocutionary force because some statements are made by individuals with tremendous formal power. Platforms should think about ways to minimize the damage that can be caused by illocutionary acts by taking this power into account.

Now I’m trying to think of some fun new labels…

— “This assertion is a power grab”

—”This statement abrogates the author’s sworn duties”

— “This tweet is unbecoming”