Against the Epistemic Status

I've been considering the idea of assigning an "epistemic status" to each of my blog posts, in the fashion of Scott Alexander. Basically: adding an addendum at the top of each blog post indicating the degree to which I really believe what is said in the blog post. Perhaps I no longer believe what I wrote a year ago — in that case, I might add an epistemic status warning readers that I no longer believe it. That's the idea.

I've decided I'm against epistemic statuses. TLDR: I think at best they are useless, begging the problem they seek to address; and at worst, I think they could very well decrease the total, long-run truth-value obtained within a writing/reading community.

The epistemic status gives a false sense of rigor and humility. One reason is because there's no epistemic status for the epistemic status. An ES is not a confidence interval, derived by some transparent calculation procedure. It is probably more subjective and error-prone than the full blog post. One reason I never post an ES — when I've sometimes had the urge to, especially after weaker posts — is that I always feel so radically unsure of my post-writing impressions that for an ES to actually increase the transparency/reliability of the post, I feel like I'd have to say I'm also utterly unsure of the ES, and so on to infinite regression. Thus, tacking on an ES at the top of the article feels to me primarily like rational self-skepticism/humility-signaling, which doesn't in any way solve the problem. Also, from the reader's perspective, the epistemic status begs the question of how reliable any blog post is, because they still have to decide whether they trust the epistemic status. For new visitors, the epistemic status therefore solves no problem, and merely adds text while bumping the trust/credibility problem up a level.

The practice of adding post-hoc epistemic statuses lends to the entire blog an impression of always being epistemically up to date, but I don't feel I will ever have the time or conscientiousness to really keep all the posts' epistemic statuses up to date with my current judgment. Therefore if I simply overlook some old posts I don't really care about anymore, and readers see there is no epistemic status downgrading them, they might reasonably infer I still fully own those beliefs.

For return visitors and regular readers of a blog, the ES is essentially an appeal to one's own authority, a cashing-in on past trust and cultural capital earned by the author's substantive content.

Ultimately, every claim I make, or inference I imply, nested in every article I write, nested in every collection of articles, has to be given some level of credence by each individual reader. Whether one line is a joke or not, whether one claim is likely to be true or mistaken — these are questions every reader must make for themselves based on whatever information they have about my claims, and the project I'm embarked on, and my reliability as a source. Assigning an ES to each unit I publish would be to lull the reader's vigilance into an unjustifiably comfortable slumber. It might make them feel like I can take care of their meta-rationality for them, when in fact it's an irreducible existential burden for all thinking adults. I don't want my readers to feel like they are cast adrift in the wilderness, but alas they are. So I don't really want to make them feel otherwise.

I think the normal presumptions about the nature of blogging are meta-rationally superior to epistemic statuses. It's just a blog: take everything with a huge grain of salt, but if something is really well demonstrated and supported then believe it, as you see fit. If you see a post from three years ago, of course the author has probably changed their views to some degree. The best response to this is to read more contemporary posts, to judge for yourself what this author really thinks on the whole. If a reader doesn't care to do this, no epistemic status is going to ensure their initial exposure is lodged into their long-term memory correctly. Such a person will either never remember the blog post or, if they are so unwise as to memorize and repeat to their friends something I reported in one blog post three years ago, I suspect they would bulldoze right over even the most cautious epistemic status warnings.

Better is to just put super-wide confidence intervals on everything one writes. Some things I say will be dumb, biased, and/or mistaken. But some things I write will — hopefully — get closer to way bigger truths than I can even appreciate! If you assign epistemic statuses to your blog posts, you really should also say when and where you think you are super correct. Most sane people will not want to place at the top of a blog post "Epistemic status: I feel a 5% chance that the claims below could change the course of world history." But any serious and passionate intellectual gets some taste of this genuine feeling every now and then! Thus, if this epistemic status business does not include such self-aggrandizing caveats, that too might be systematically biasing. I'd rather just give one big caveat about my whole body of writing, that it is merely the inspired guesswork of one person trying their best to be correct. Implicitly, some stuff will be more wrong than it might seem, and some stuff will be even more right than it seems. The only commitment one needs to make is to do one's best, in a way that updates moving forward, rather than attempting to move backward with post-hoc re-evaluations.

I admit that some of my intuition on this question is due to my temperament: I like to work fast, always move forward, never look back. I can do the disciplined work of editing but I'm not exceptionally high in Orderliness; I run mostly on the dopaminergic movements of exploration, inspiration and creation, adding just enough conscientiousness to complete things responsibly. As far as bloggers and "content creators" go, I'm high-variance: I put out a lot of high-quality stuff that I take very seriously, but I also put out a lot of random stuff sometimes bordering on bad comedy. So part of what I wrote above is just rationalizing all of this. But this is also my personal alternative to the epistemic status: self-conscious reflections weaved immanently into any given unit of production.