Mike Caulfield Profile picture
New book: Verified, from UChicago Nov 2023. Moved away to mikecaulfield . bluesky . social
Aug 5, 2023 15 tweets 3 min read
So let me use this as an opportunity to demonstrate why we have to revive cognitive dissonance theory as a lens on this stuff. The question here is "If people really *believe* X, then why don't they Y" 🧵 E.g. if they really believe elites are drinking blood of children in ceremonial caves or whatever then why aren't they out in the streets. Etc.
Jun 10, 2023 6 tweets 2 min read
A post. Basically there's two ever-warring factions in misinfo studies -- the first with naive theories of direct impact, and the second with a belief that since facts are often adopted in the service of belief facts have no impact on belief. mikecaulfield.substack.com/p/reasonablene… I propose a third route, largely inspired by Leo Festinger's work on cognitive dissonance: facts are essential to the *maintenance* of belief, because crucially people would both like to believe things and be thought reasonable.
Jun 9, 2023 8 tweets 2 min read
There's much to like, and some to dislike in this evisceration of "innoculation theory" disguised as a book review, but this aspect is key -- you cannot tell the truth or falsity of a statement byt the construction of a statement. The fundamental problem is ... People express true things in emotion-laden language all the time, and express horrible lies with clinical flatness. Bizarre things are true, and lies can seem like common sense.
Mar 3, 2023 14 tweets 3 min read
I know I talk on and on about argumentation theory lately. But let me tell you why this video is fascinating from the perspective of something called the Toulmin Model. The Toulmin model was introduced in 1958 by Stephen Toulmin, and while initially rejected by the philosophical world of the time eventually helped usher in a new approach to informal logic. thriftbooks.com/w/the-uses-of-…
Mar 3, 2023 5 tweets 1 min read
Good example of how ChatGPT fails much of the time by riffing on common meanings of words instead of specific niche meanings. Image (In informal logic/rhetoric, an instrumental claim is a subclaim used to support an argumentative claim. So for example the argumentative claim might be: We should put a lid on the demography minor. An instrumental, supporting claim might be "the minor is too large")
Feb 15, 2023 12 tweets 3 min read
People seem to think that ChatGPT is a competitor for Google search. I think it's more of a (presently inferior) competitor/supplement to Wikipedia. If you get this part straight, a lot of other stuff will make sense. The truth is that Google is already on a productive path using ML and LLMs -- this stuff has been slowly integrated into the product in ways that are less flashy but ultimately more productive and I think epistemically grounded.
Feb 15, 2023 5 tweets 2 min read
But this actually proves the point there is something uniquely broken in superhero films, because war films have a rich variety that is self-evident, and superhero movies are similar enough that people say they are all the same. The question isn't whether things are in the same genre -- they are. All those things are war films for example. The thing is whether the genre is vibrant and in general if you're going to reach for a vibrant genre you'll reach for war films and not superhero films, like here.
Feb 3, 2023 9 tweets 2 min read
Oh, geez, I just was alerted to this -- Roger Schank, my old boss, innovator in AI, script theory, and educational design, has died. paperman.com/en/funerals/20… He's going to have a complicated legacy, to say the least, because of his defense of indefensible people toward the end of his life. I'm not here to whitewash that.
Feb 3, 2023 10 tweets 2 min read
Paying tribute to Black History Month by treating pictures of different Black entertainers as interchangeable is -- well, a choice, I guess 🤷‍♂️ Here's something on the real Baby Esther, complete with an actual photo that is actually her. It's not hard if you try. en.wikipedia.org/wiki/Baby_Esth…
Feb 1, 2023 6 tweets 2 min read
Hey, I have a book with @samwineburg coming out from University of Chicago Press in October this year. It's called VERIFIED: How to Think Straight, Get Duped Less, and Make Better Decisions About What to Believe Online. It's a how-to guide that compiles the educational research and practice Sam and I have honed over the past decade around making sense of claims and sources online -- and doing so under realistic time and attention constraints.