This is what happens when you train neural networks largely…

This is what happens when you train neural networks largely on tone and its stylistic relics. They pick up formal features of arguments (not so much fallacies as tics) that have almost nothing to do with semantic content (focus on connotation over implication).

This is what happens when you let philosophers try to write about real life. This ridiculous, game-playing, feigned innocence. Journals have been full of this for years, this elaborate performance of *doing philosophy* and saying nothing. I cannot adequately express my contempt pic.twitter.com/ciDeWuEkET
— Jack (@jackeselbst) January 14, 2021

This is a secular problem in the discipline. It's got nothing to do with the Analytic/Continental split in the anglophone world. They've both got the same ramifying signal/noise problem, it's just that the styles (tics and connotations) are different in each pedagogical context.

And this is before we start talking about tone policing and topic policing, which are both rife and essentially make the peer review journal system completely unfit for purpose, populated as it is by a random sampling of pedants selecting for syntactic noise over semantic signal.

We've allowed a system of self-reinforcing and ratcheting filters to evolve that effectively *fuzzes* our contribution to the growth of human knowledge (https://t.co/VmW15pGt7J), because it selects for properties only loosely related to those we claim to want. Let that sink in.

This is literally the opposite of what a filter is supposed to do: extract signal from noise, syntactic compression that preserves semantic content. Instead we are awash in syntactic artifacts optimised for minimal criticisable content and maximal pedantic posturing.

When you suspend the assumption that anyone who has been selected by the system *must* understand what they're talking about and/or have something to say, taking the maxim 'fake it till you make it' as a methodological frame, it all looks like competition over formal prowess.

Yet here 'formal' no longer means 'logical', but something closer to 'rhetorical' and 'bureaucratic'. Once you see this, you see that it's exactly the same effect that debating societies and MBAs have had on political and industrial management in the neoliberal era.

Once mighty cognitive institutions (e.g., universities, political parties, and industrial corporations) have gradually devolved ways that are essentially synergetic, as the stupidity in one provides leverage on the growth of stupidity in another. This is the cunning of stupidity.

And so we return to the (social) epistemology of ignorance, but with a couple new points of reference: the anthropologically inspired theories of James C. Scott (cf. https://t.co/kOJPRNzDLj) and @davidgraeber (cf. https://t.co/CMQoBVd44o).

What we find in their work is an account of the socio-epistemic dynamics through which power breeds stupidity and stupidity breeds power, the genesis of stubborn knots around which active ignorance crystallises, generating complex defences that impede the flow of knowledge.

Here's what I mean by synergy between the devolution of cognitive institutions from various domains: the establishment of active ignorance in one place provides points around which alliances form, and through which leverage is exerted to establish it in other places.

The cunning of stupidity climbs the cognitive cliff face by finding purchase wherever it can within a given set of institutions and without. Bureaucratisation and managerialism are secular trends that have spread through the ratcheting the petty personal powers of the ignorant.

The real story of the dissolution and dissipation of the counter-culture of the 60s in the US and Europe is the story of its failure to generate new administrative norms to accompany the ethical, political, and aesthetic ones it had such success with: no persistent counter-power.

Instead we got the slow integration of the acceptable remnants of those norms into a different administrative culture, not simply that which had preceded the 60s (e.g., the party form), but which was born in the 60s, in the form of general purpose management and business schools.

The result was not simply the constitution of a new class of *managers* distinct from industrial workers, whose culture is projected downward through org charts in a way that aligns the interests of *bureaucrats*, but the universalisation of HR as the blueprint for its power.

I think most people on the left appreciate the point that media consolidation created an unprecedentedly homogeneous epistemic landscape in which the formation of political counter-consensus was effectively suppressed. I don't think they see the analogy in the dominance of HR.

I should show them the online forms I've had to fill in to complete academic job applications, whose mind-boggling irrelevance is nothing so much as a trap to discourage those who cannot play the relevant formal game. They build the filters, the obstacles, and the ignorance.

When you arbitrarily empower people who have no idea what it is they're exerting power over, the best intentions in the world will not stop the systemic proliferation of these knots of power/incompetence. The old banality of evil has a new face (https://t.co/XAZxtOFHLP).

I'm neither an anarchist nor a Quaker, but their critiques of these institutional forms, rooted in the autocatalytic growth of power (both centralised/distributed and personal/impersonal), and the serious alternatives they have tried to build are incredibly instructive here.

To return to my story about administrative culture, I think this lets us understand the split between the socially liberal left and the economically progressive left that became formalised in the 90s, and which haunts every attempt to push us out of this autocatalytic nightmare.

Every liberal concession to the counter-culture of the 60s and the liberation movements that have persisted since then has been channeled through the administrative culture of which HR is the nexus. This produces some positive results, but they're always systematically skewed.

For instance, it's great that hiring processes have been evolving in a direction that actively tries to compensate for demographic bias, but when this is merely one aspect of broken filter that otherwise removes the most useful signals, it's a depressingly pyrrhic victory.

Similarly, it's great that the technical language of liberation movements and the deep thinkers who to bring their struggles to self-consciousness disperses into the wider world, but when its principal use is to play out petty micropolitical feuds in formal games, it's maddening.

Managerial overreach, bureaucratic metrics, and quasi-academic jargon are different facets of the same secular process of self-reinforcing power/ignorance. My hope is that academics' perennial hatred of management might form the basis of a teachable moment regarding our own sins.

That's the end of my thoughts on this topic for now, but as usual I will refer you to some more of my extant work on these themes:

1. My recent thread on communicative interaction, dynamic fallacies, and the (synthetic a priori) maxim 'fake it till you make it': https://t.co/2k2QHRYaC7

I won't let you claim the mantle of Socrates until you've earned it by besting me logically, rather than merely rhetorically. So be prepared to find out you're the one with the rubber mask, not me. Don't hate the player, hate the game.
— pete wolfendale (@deontologistics) January 13, 2021

2. A couple recent threads responding to @cstross, discussing the broken education system and the relation between the plague of terrible metrics and general purpose management: https://t.co/SW2keOnZcv

A short thread on education as a model of how bad assumptions about the nature of information distort and eventually break social infrastructure (h/t @cstross). https://t.co/MmJVhuAwCm
— pete wolfendale (@deontologistics) January 5, 2021

3. Another recent thread on the Foucault Wars, which lays out the basics of my theory of jargon as pathological technical vocabulary: https://t.co/ZEu1i39QaC

Solidarity with those who need philosophical vocabulary to bring their ongoing intellectual labours and political struggles to some new level of self-consciousness, might require refusing complicity with those who carelessly sabotage its semantic substance by playing pretend.
— pete wolfendale (@deontologistics) January 4, 2021

4. A discussion with @Chican3ry about the difficulties of reading philosophy and how to navigate the canon, which includes thoughts on Sellars idea that 'the history of philosophy is the language of philosophy' and a brief critique of 'Sokal Overfitting': https://t.co/NLvAfU0aQT

Anyway, that's my advice, and the reasoning behind it. If you ever want more specific advice on either side of the process (synopsis/analysis), feel free to ask. Good luck!
— pete wolfendale (@deontologistics) December 27, 2020

5. A very old, but theoretically fresh post on 'The Systemic Problems of Contemporary Academia' with an eye to philosophy in particular. This contains my first attempt to think through these issues in information-theoretic terms: https://t.co/sUvKjlf85M

6. Some more recent threads that begin to think through some solutions, by narrowing the problem down to the role of public philosophy:

a) Video Essays as a Philosophical Medium: https://t.co/a6epfzXYLS

b) Enlightenment and Opportunism: https://t.co/V8DJcgf9jm

I'm fielding a lot of objections to the claim articulated in this tweet (which you can read down thread), but it might be worth starting a new thread that takes a different tack, and synthesises my perspective on the video essay as a legitimate format in which to do philosophy. https://t.co/hHDD9ndUJo
— pete wolfendale (@deontologistics) December 28, 2020

7. 'The Going Price of Power' (https://t.co/aa2rcbOc40): An outline of the broader economic narrative I'm trying to construct, which ends with a discussion of managerialism and the secular trends I'm complaining about.

Finally, I'm going to reference a couple recent pieces by @OlufemiOTaiwo and @lastpositivist that have been jostling around in my brain as I'm articulating these thoughts:

1. https://t.co/K9z8XUxSIu

2. https://t.co/e76UYs5Zkq

3. https://t.co/ThvjoWStHQ

4. https://t.co/5izUq7ufea

And that's the morning thread! Additional thanks to @deonteleologist for occasioning some of these thoughts.

This is what happens when you train neural networks largely on tone and its stylistic relics. They pick up formal features of arguments (not so much fallacies as tics) that have almost nothing to do with semantic content (focus on connotation over implication).

More from pete wolfendale

More from For later read

You May Also Like