A Tale of Two Outcomes
Earlier Amanda Askell tweeted,
The mild elation and superiority people feel when they express heterodox views is epistemic poison.— Amanda Askell (@AmandaAskell) May 5, 2021
It resonated with my recent experiences on twitter, so I retweeted it. Shortly after, Michael Nielsen rebutted (through my retweet),
Heterodox views are essential to creative progress. If people feel good about expressing them, that's good.— Michael Nielsen (@michael_nielsen) May 5, 2021
I broadly respect Michael, so disagreement (via a retweet/reply proxy) is uncomfortable. It also can be the very best of twitter! It signals that either the problem is being posed poorly or someone is wrong or both. Here, I think it’s the former. For lack of precision, we’re groping at different aspects of “heterodoxy” and failing to see the lack of conflict between them. Since I seem to keep doing this and since it also seems to keep coming up in my circles,Most recently, Jim O’Shaughnessy did a relevant InterIntellect about it with Bronwyn Williams. I wanted to try to map the terrain slightly better.
Heterodoxy as Caustic Social Negation
‘Heterodox’ and ‘orthodox’ are frustratingly subjective labels. To some degree, they always were. But, they are more so now. The abolition of time and space as communicative constraints allowed the complications to multiply. Endless niches now flourish, each of which is in a minority relative to the whole.
In most ways (to me), the resulting Cambrian explosion of belief systems is good. The world need not be so Procrustean. One size does not fit all! However, by algorithmic- or social-mediation, social mediums do not readily afford unbiased samples of human expression. Consequently, trying to infer group-contingent consensuses is error-prone at best.When I searched Google to find out what the plural of ‘consensus’ is, the knowledge box extracted “The plural of ‘consensus’ is just ‘consensus’. It’s Latin” from wiki talk, which is hilarious in the context of this post.
The same sampling dynamics end up creating constant collisions between groups. This can induce a premium on heterodoxy that has no relationship to anything but social (un-)reality. The pattern should be a familiar one,
- You assert some assumed heterodoxy and those around you agree. It feels good. Vindication! And the negative reactions elicited? Confirmation that you are in the heroic minority. E pur si muove!
- With chronic exposure to this process, assertions stop being expressions of what your niche believes, values, or experiences and become probes aimed generating more of (1).
A cheap and reliable way of constructing such probes is to just throw out antagonistic negations until one hits.Alex Danco defined trolling as “posting where the reaction is the content.” I’m not sure if he coined the phrase, but damn is it good. Link to podcast But, we either lose track of our motivations or were never even aware of them. It becomes easy to cast ourselves as a member of the valiant heterodoxy persecuted by the orthodox mob. But the mob isn’t all that discriminating.
To me, this is the “epistemic poison.” Heterodoxy as a means of decomposing social space but mistaken for a method of truth discovery. The specific assertions are barely tested because they don’t actually matter in this process. But that doesn’t mean they aren’t integrated into our beliefs through the social reinforcement. And this is corrosive because credit assignment is fucking hard. The accumulated mistakes could be costly, but there are plenty of rationalizations for why we remain obviously right.
Heterodoxy as The Promethean Fire of Progress
You can’t simply go look up what’s correct because such an authority does not exist. We haven’t figured everything out yet. We won’t ever do so. The world is too big, too complex, and too dynamic for that to be anything but a dystopian possibility. To assume that any deviation from the orthodoxy must be wrong is a commitment to stasis.
In the romantic retelling, heterodoxy is exactly that which rejects this stunted condition. It is what keeps asking “but why?” questions and coming up with answers. Most of the new answers are wrong; many of them are seductive nonetheless; and, a handful of them open up new worlds. Unfortunately – or fortunately, if you find discovery thrilling! – its hard to tell which case you are in for any particular heterodoxy.
If anything, history has the effect of obscuring the difficulty. Both time and historians filter out most of the details. When we “look back,” we see wonderful and cognitively-friendly margins between brilliance and foolishness (or evil). We’re presented with simplified models of what happened. They’re “wrong,” but generally useful – unless you want to try to estimate how often the boundary is fuzzy. Fuzzy doesn’t replicate easily, which means it doesn’t survive.
For these reasons, cultivating a tolerance if not a taste for heterodoxy is, as Michael said, essential to the creative process. If there are penalties to heterodoxy imposed even when eventually proven “productive” or “correct,”I’m scare quoting here because of a fundamental agreement with what my friend Chaos Prime has said eloquently. it’s not unreasonable to conclude people may learn to stick to the low-conflict orthodoxy. And that, too, can be epistemic poison. Orthodoxy becomes something other than an informative prior – it becomes truth “beyond contestation.”
These Statements Aren't In Conflict
There is no contradiction in believing that heterodoxy is essential to the creative process and that heterodoxy as a fashionable and mechanistic posture is corrosive. Losses imposed by inhibited discovery and losses imposed due to errant social inference are different. But they’re still errors. Attending to both concerns is perfectly consistent. Assuming a binary isn’t.I’m not saying Michael is making this mistake. I am saying that I think I and others frequently do though.
So, why does it feel otherwise so often on twitter?
Because on a medium that gives you 280 characters to say what you want to say, readers are going to need some side information in evaluation. Thus, orthodoxy/heterodoxy is an almost archetypal collision space.
I can use my own ExTrEmElY OnLiNe experiences as a concrete illustration. Here, I think Austen is correct in that “Tech needs more crazy, not less.”
Periodic reminder that any "consensus" you perceive and approximate on twitter is probably an artifact of the interaction patterns that this platform selects for, and that probability rises with follower counts. https://t.co/Biex1tTIOT— (wannabe) Ƀreaker of (the Bad) Loops (@generativist) November 29, 2020
Except, by quote tweeting it and emphasizing the costs of bad negation, I’m still reacting to the orthodoxy/heterodoxy frame, which essentially reproduces it! He didn’t actually say anything specific about what the “consensus” rejects. He was just assigning a premium to the Promethean portion. Any inference I draw beyond that is subject to the same confused process of social inference I’m warning against.And the quote tweet was clearly low-level hostile.
I don’t think I would say the same thing again.
Orthodox/heterodox is not a very useful frame.
It’s a trap.