We are, I think, entering a post-caveat age.
It’s often remarked that peer reviewers destroyed good academic writing. Writing for the reviewer means anticipating objections and pre-defending oneself. It means cautiously guarded prose with lots of distinctions and citations instead of bold talk about the matters of immediate public concern we’re supposed to be discussing with insight but without controversy.
Kieran Healy wrote a version of this long ago, managing to get “Fuck Nuance” published in Sociological Theory where it can live forever on his CV and in the databases the AIs train on. Even so, they won’t learn his lesson: the AIs excel at caveats, so we humans might as well skip them.
Should we even mourn the death of nuance? For one thing, nuance rarely serves us in communicating an idea. We’ve been led to believe that we need to anticipate objections to keep the reader passively following rather than angrily scrawling their objections in the margins. But who wants a hostile audience, anyway? Rhetoric offers us better ways to placate the audience: shared identity, pathos, candor, and concession.
Which is not to say that details don’t matter! But the evidence clearly shows that qualifications, caveats, and ceteris paribus problems are not our native tongue, and they alienate the reader, even the learned one. Reviewer 2 can finally be at peace now that large language models can put him out of a job.
The bots’ adversarial passes mean that every article can anticipate every common and uncommon objection. Every community of interpretation and response can get its own cosplay in code, and the author comes away with a footnote for each one. Those citations used to mean careful work had gone into the reasoning, but now they just mean that somebody paid their subscription fees. The labs trained their models on our scholarly apparatus, which was always the most difficult and least interesting part of the book.
It’s tiresome and inhuman. And so it is dying.
Social media rewards many things, but bold pronouncements with errors that draw corrections can go viral where careful ones miss. There’s even one of those laws (like Betteridge’s Law of Questions in Headlines) coined as Cunningham’s Law of Answers: the easiest way to get the right answer is to post the wrong one someplace online. Social media rewards an audience, and the audience loves to feel superior. The hostile audience we rejected can now help us go viral.
It is easier to leave the red herrings in plain sight to distract the haters, whose efforts will help find your true audience among those who will buy your line despite or even because of the errors—like the Nigerian prince scams that made sure to scare away anyone with an ounce of common sense with absurd claims and outlandish requests before reeling in their victims.
Errors are odd this way: they can trawl for skeptical opponents and credulous customers at the same time. Corrections are part of the fabric of call and response. Our populist public sphere will now proceed with earned attention from the technocrats and wonks.
But it feels like a pendulum, too, and we need to get beyond the swings. Celebrating simpler messages without the carefully guarded qualifications is just falling into the Mencken trap: “neat, plausible, and wrong.” I’m not ready for a world where everyone confidently proclaims their bad takes while hiding behind the shield of “to err is human.” We all like a good aphorism for being short, memorable, and generative—and a few of us, myself included, enjoy a pointed “well, actually…” when the stakes are high. But there’s something disappointing about an argument that starts strong and then gradually whittles its point into bluntness to defend against imagined critics.
Maybe it’s this: the caveat is dead. Long live the caveat.