A couple years ago, I was asked to help a couple of consultants troubleshoot why the leaders they were working with weren’t working well together. Honestly, the problem isn’t really that important. It was the interaction that followed them briefing me about the problem that’ll give you a chuckle. It’s funny enough that it’s not the first time I’ve told it:

For best effect, I would like to relay what followed as a relatively faithful transcript:

Me: “Ok, so tell me if I have the sense of things: [I echo back roughly what I heard, to make sure I have it right.]”

Them: “Yes, exactly: [they repeat again roughly what they said, leaving me a bit confused as to whether they are correcting me or confirming, but it doesn’t seem too complicated, so I assume the latter.]”

Me: “Ok, so how do you think I can help?”

One of them, boldly: “Well, we need to move the leaders to the next level, and we thought you might have some insights from your psychology background. [Lots of nodding in agreement around the table].”

Me: “Sure. Why don’t you tell me what the levels are?”

The same one of them, more diffidently now: “… Hm… [Looks of mild consternation are exchanged from left to right.] Well, there aren’t really any levels…”

Me, with crinkled brow: “… Right. No worries, what did you mean then?”

[A pause.]

Another one of them, abruptly: “Well, what we really need to do is move the leaders along the journey. [Nods of a more enthusiastic nature erupt around the table. Some smiles too.]”

Me: “Great! [I am smiling too, the enthusiasm is infectious.] Ok, so a journey—we have a starting point, an end point, and some kind of trajectory. Why don’t you describe those for me?”

That same other one, substantially less abruptly: “… Mm… [Looks of a more troubled nature are exchanged from left to right.] Well, I guess… It’s not really like a journey…”

Me: “…”

Them: “…”

Most of the rest of the meeting was spent working out just what the fuck they were talking about.

Consulting is notorious for this kind of thing. The corporate world in general in fact. Phrases that don’t seem to mean anything. Ed Zitron has a beautiful write up on this. He quotes a giddy profile piece:

During the push to grow AI, [ServiceNow CEO Bill] McDermott has insisted his managers improve efficiency across their teams. He is laser-focused on a sales team’s participation rate. “Let’s assume you’re a manager, and you have 12 direct reports,” he said. “Now let’s assume out of those 12, two people did good, which was so good that the manager was 110% of plan. I don’t think that’s good. I tell the manager: ‘What did the other 10 do?’”

Ed goes on to wonder:

You’ll notice that all of this is complete nonsense. What do you mean “efficiency”? What does that quote even mean? 110% of plan? What’re you on about? Did you hit your head on something Bill?

No, Ed. Bill didn’t. See, there are no levels.

¶Karstic traps

If I were to boil down the essential motivation of btrmt., it would be something like:

If you don’t do the thinking, the thinking will be done for you.

And it’s this kind of thing—Bill’s bewildering monologue, or a bunch of smart people talking about levels that don’t exist—that drives a lot of my concern around this fact.

I call this kind of thing karstica. Karst is this peculiar kind of landscape you get when water spends thousands of years dissolving limestone. On the surface, you have rolling green hills, paddocks, and scrubland. It’s famously beautiful—the stuff of national parks and tourism. But it’s deceptive terrain. Underneath, caves and voids and hidden channels writhe in every direction. Some of these voids are roofed by just a couple centimetres of soil and root. You can walk on it. Build on it even. But it’s hollow ground. Dig too deeply, and it’ll give way and swallow you whole.

¶The Naming Problem

Bill and my consultant friends were demonstrating a particular flavour of linguistic karst. I’ve called it the naming problem before. When we confuse ourselves into thinking that simply naming something explains it. Like saying someone has an addictive personality. You sort of feel like a problem has been explained, but it isn’t really—you’ve just deferred it. Why is he drinking? Well, he has an addictive personality. What is that? A thing that makes him drink. What’s the thing though? An addictive personality. Or, saying someone is a natural leader, which mostly seems to mean that they ended up leading something. Why is she in charge? Leadership traits. What are those? Things that put her in charge. In our case, my consultant friends saw a problem and thought, well, our leaders are just at the wrong level. But what are the levels? It’s not clear.

In all these cases, real things are being identified, but they’re just being described not explained.¹ They are attractive descriptions. At the surface. Just like karst. My consultant friends were enthusiastic about the jargon—smiling at each other, and nodding vigorously when they’d come up with the description. But, again like karst, deceptively so—asking questions about the description revealed that they were hollow ground. Words without referents.

¶Malcolm Gladwell shit

There are a couple other flavours of karstica though. I complain a lot about what I call Malcolm Gladwell shit—education as entertainment. You find this in heaps of airport psychology books.

Take for example the Chimp Paradox. He tells us that the brain can be thought of in terms of three categories of processes: the ‘chimp brain’, which is emotional; the ‘human brain’, which is rational; and the ‘computer brain’, which stores our habits and beliefs for the chimp and human to use.

The chimp brain, you see, takes over when we feel threatened, picking messed up habits from the computer and impeding our performance. So we need to train the human brain to maintain control over the chimp brain, so we can pick better habits and perform better. The spends some time painfully simplifying brain science to support his claims. The book is a bestseller.

You read his book and, drowning in references to limbic systems and prefrontal cortices, you feel like you’ve learned something new! But you haven’t, actually. If you think about it for a while, you realise that you already knew that you were a bit too emotional sometimes; a bit too impulsive. The only thing the book has given you is new words and a vague air of academia to justify it to yourself. You can tell that this is the case because the identical idea has been sold many different times, under different names. Kahneman’s Thinking Fast and Slow, Goleman’s idea of amygdala hijack. Polyvagal theory. Etc.

What’s happening here is that we think we’re educating ourselves, but actually we are just entertaining ourselves. Concepts we’re already pretty familiar with are being represented to us in a new and exciting manner, under the guise of self-help or performance literature, and we mistake the experience of recognition for a glint of learning. Almost every Malcolm Gladwell book does this,² that’s why he’s so famous. Or Atomic Habits, or Positive Intelligence, The Power of Now, The Subtle Art of Not Giving a F*ck. Name one, it’s probably this. Rarely do you come away with anything that actually has bearing on how you might live your life differently, appealing though it might be. Here the ground isn’t so much hollow. More like it sits over the same channel as something else does. A channel you’ve already discovered.

¶Linguistic laundry

There’s another flavour I’ll name here, a more insidious one, before I move on to talking about what we can do about it. But it’s worth understanding that these are all, to me, the same thing. Simply that they are weaponised in different ways. The naming problem is a form of karstica that’s used to hide, or protect ourselves from, ignorance. Or, more benignly, it’s just a problem of curiosity. Malcolm Gladwell shit is a misunderstanding—entertainment dressed up as education in a world that values productivity as a moral virtue. Helps that it sells a lot of books. This next one is grosser. More malicious, even if it isn’t conscious. But they’re all examples of the same kind of karstic trap.

This last flavour is one I’ve only just started to notice myself. See, I get a lot of requests to cover content related to gender essentialism these days. It’s very trendy. In particular, it started with calls to review this relationship advice book from the 90’s called Men are from Mars, Women are from Venus. People love this book. I once loved this book. But as it turns out, it’s a crock of shit. A karstic crock of shit.

The author sets the scene almost reasonably. To quote myself:

He says: “we will explore how men’s and women’s values are inherently different. Men mistakenly offer solutions and invalidate feelings whilst women offer unsolicited advice and direction.”

So far, perhaps not controversial. He’d have me believe men … like “outdoor activities like hunting, fishing, and racing cars.” They care more about “objects and things” rather than “people and feelings.” And men don’t talk about things unless they want a solution. “Asking for help when you can do it yourself is perceived as a sign of weakness.”

Which sounds pretty reasonable, at first glance. Men and women do seem like they have different interests. Objects and things vs people and feelings seems to be a good rule of thumb to describe that.

But then, things go off the rails extremely quickly:

The cracks show when he talks about women. See, “instead of being goal-oriented, [women] are relationship-oriented.” And then: “When a woman tries to improve a man, he feels she is trying to fix him. He receives the message that he is broken. She doesn’t realize her caring attempts to help him may humiliate him. She mistakenly thinks she is just helping him to grow.”

See women, unlike men, don’t “value power, competency, efficiency, and achievement”. And so, by asking him to do stuff, they make him feel broken and humiliated.

For context, the terribly offensive things women are liable to ‘improve’ a man with are like “Those dishes are still wet. They’ll dry with spots”, or “You should spend more time with the kids. They miss you.”

Which… seems less like unreasonable criticism of a competent man, and more like reasonable suggestions to an incredibly sensitive man.

What the author is doing here is laundering incompetence into something more virtuous using the language of masculinity. He has many personal anecdotes that describe a fairly fragile, rejection sensitive nature. But, by framing this fragility as stoicism, or hiding it behind a value set oriented around ‘objects and things’ rather than ‘relationships and people’, it makes it seem palatable. He’s a man! Stop trying to mother him, woman! If only he didn’t need mothering.

A little while ago I published an article on incompetent men, where I made this point more concretely. Where I point out that a great deal of the conversation about masculinity is exactly this:

The standard narrative goes something like this:

1. the world is becoming increasingly feminised;
2. masculine traits are thus increasingly devalued, becoming seen as ’toxic’ instead;
3. men are losing out;
4. crisis.

Now, this is probably true to some extent. But also, it’s used to launder incompetence. Several societal changes have happened over the last 70-odd years that mean that men have to learn new skills to be useful men. They can’t rely on bringing home the ‘bread’ anymore, not just because women can do this too, but because living on a single income is nearly impossible these days. They can’t rely on knowing ‘masculine skills’ anymore because women have spent half a century responding to social pressure to adapt to the ‘world of men’ and learn those same masculine skills. And they can’t pretend that being a man has value anymore because social media makes this untenable—men do far too much visible bad shit than good shit to make this story sensible.

Essentially, men don’t have the skills the modern world needs of them. Feminine skills. And social media makes it visible. But rather than spend time working out how to help men develop these skills, we frame this problem as a problem of identity. As I say in that article:

The young men who can’t hold down a job, manage a household, or sustain a relationship aren’t failing at adult competencies that women have been quietly mastering for decades—they’re “searching for their role”.

And in doing so, men are given a reprieve. They have laundered their problem into everyone else’s problem.

A particularly insidious kind of karstica.

Now, this is my particular bugbear, but you’d recognise this in other contexts too. Think about ’hustle culture’ and the ‘rise and grind’. The laundering of exploitative working pressures into productive virtue. Or ‘move fast and break things’, the laundering of recklessness into innovation. Or even the weaponisation of self-care terminology to justify selfish and antisocial behaviour. You think of Jonah Hill’s famous example of ‘boundaries’, like “no surfing with men” or “no friendships with women who are in unstable places”. Controlling behaviour dressed up in therapy speak.

Linguistic laundry. A form of karstica that has you falling into one hole and coming out an entirely different one.

¶The common core, and the stakes at play

These are all flavours of karstica, to me. The naming problem is a problem where the pretty words people say point at nothing at all. Malcolm Gladwell shit is where the pretty words point at something quite anodyne. And linguistic laundry has pretty words distracting you from something troubling. But all are karstic—narratives about the world that are superficially pretty, but hide the sinkholes, voids, and channels beneath.

There are probably more flavours. But the flavours aren’t really the interesting thing. Indeed the flavours might make you think that we should take some more seriously than others. The naming problem is just silly. Malcolm Gladwell shit is typically harmless, maybe even entertaining, but otherwise just irritating. Perhaps we should just focus on the linguistic laundering, then.

But I think that all karstica shares a common core. And that common core means that certain kinds of activities inherit certain failure modes. Failure modes that we’re increasingly susceptible to in the age of AI.

¶The common core

The best description of what seems to me the common thread across the flavours of karstic is described by the sociology of the interesting. As I say elsewhere:

Davis goes on to notice that the most successful theories are those that subvert our weakly held beliefs. The hot-takes on things we don’t care very much about. If our strong beliefs are attacked, then we’re likely to resist the attack. If our existing beliefs are confirmed, we’re likely to do little more than nod and forget. But if the knowledge we don’t care about very much is revised, then it becomes interesting.³

His central claim is that what makes something sticky is not its truth value, but whether it denies certain assumptions of an audience. An interesting theory challenges what people take for granted. In contrast, boring theories are those that simply affirm what we already believe. Importantly, the challenge can’t go too far. If a theory denies too many assumptions, it becomes unbelievable rather than interesting. He then sketches a few basic patterns of what this might look like:

• What seems to be X is actually non-X. The Chimp Paradox and its kin are this. “Oh you’re not irrational, it’s your chimp brain or your amygdala or your System 2”.
• What seems to be a homogeneous phenomenon is actually composed of heterogeneous elements. Think of the 5 Love Languages—relationships look like one thing, but actually they’re five distinct languages.
• What seem to be assorted heterogeneous phenomena are actually part of a single unified whole.⁴ The Body Keeps the Score is a good one for this—chronic pain, addiction, depression, autoimmune disorders. All these things boil down to trauma at the bottom.
• What seems to be individual phenomena are actually organisational/structural. Gladwell loves this. Outliers is all about how success isn’t talent, but actually a product of circumstance and the environment. Or Haidt’s Coddling of the American Mind Anxious Generation: youth mental health isn’t their fault, but ‘safety culture’ and social media.
• What seem to be separate phenomena are actually the same. A good example is the finding that insomnia and chronic indecision are actually the same neurological state, just operating at different times of day.⁵
• What seem to be similar phenomena are actually opposite. Another good example from the same paper, about how discipline and willpower aren’t the same thing—they’re inversely correlated. Disciplined people don’t have willpower, so they needed discipline instead, whereas the wilful never needed discipline in the first place.⁶

Every pop-psychology book uses the patterns to sell content. Even I do! See if you can guess which this article is going for. But, as Davis noted, interesting is under no obligation to be true. And indeed, the last two examples I used, I just made up. If you checked the footnotes hoping to find links to the studies, then you proved my point bout how easy it is to generate interest (though the footnotes have the actual examples I would have used if I didn’t want to play).

The problem with this, is that actually, anything interesting probably fits nicely into these patterns. Blog articles. Podcasts. TikTok reels. Any media we consume, that survives, has embedded a pattern of interesting-ness. And since content doesn’t need to be substantive to propagate, only interesting, then karstica rides rampant.

This is a problem sharpened by the dominant grammar of media which, since at least the Mad Men era of the 60’s, has been narrative-based rather than expository. Philosopher Walter Fischer, in his Human Communication as Narration, pointed out that humans evaluate claims by narrative coherence, and not narrative fidelity. We are, as Gottschall put it more recently storytelling animals. Marketers noticed. Think of Seth Godin’s All Marketers Are Liars (Tell Stories) or Donald Miller’s StoryBrand. Because storytelling bypasses the scrutiny we’d apply to an argument, and suppresses argument. Then, this became the operating manual for media. Neil Postman complained that, in the television era, we are exposed to news from nowhere, about nothing, because we find stories interesting, even if they’re about places and people we’ll never be able to do anything about. In the social media era, this is worse, because algorithms reward the kind of dwell time that stories produce.

Our media is less and less about information, and more and more about interesting patterns of information, embedded in story-shaped content. And our media relies on almost nothing else to garnish our attention from us.

All in all, karstica’s perfect delivery mechanism. The naming problem can live easily inside a character arc. Malcolm Gladwell shit dominates non-fiction—books where the information comes in the chapter heading, and the rest of the chapter is story. Linguistic laundry requires narrative to work, because it’s the swapping of one narrative for another.

If karst is created by the slow movement of water over limestone, linguistic karst is created by the fast movement of interesting stories across our feeds.

¶The stakes

I said earlier that the core motivation of btrmt. is something like:

If you don’t do the thinking, the thinking will be done for you.

And I suspect this is important because humans are animals first. Our brains create patterns of thought, feeling, and action that map the structure of the world around us. Graceful solutions to an extraordinarily complex world, but automatic ones. Importantly, humans are a very special kind of animal. We are the animals with the greatest capacity for nurture. Whose core differentiator is the ability to share ideas. The same reason, perhaps, that stories are so compelling to us.

What that means, though, is that these automatic patterns of thought and behaviour, in humans, become ideologies. Patterns that we adopt from others, rather than patterns we learn from experience. And just like patterns we learn from experience, they are patterns that drive our behaviour.

As such, these patterns can be used against our intention. Our gift for sharing ideas can be exploited, or hijacked. Karstica is often this. The naming problem hides the fact that your boss doesn’t know what’s going on, putting the pressure on your shoulders to work it out. Malcolm Gladwell shit is used to make you buy books and courses. To eat away at the margins of your attention when your attention is a commodity. And linguistic laundering is used to distract you from the actual problems that plague us.

This might be enough to convince you that we should care. But now we’re moving into a new age. The age of AI.

¶Industrialised karstica

I attended a conference recently at Chatham House in London hosted by the British Army. I don’t think I still need to keep it secret, because they posted the full thing on youtube.

The point of the conference was, more-or-less, to tell people like me who worked in Army leadership, to get a fucking move on with the whole AI situation. It’s coming, and we’re not integrating fast enough. The Chief of the General Staff stood up at the end and pointed out that practical AI literacy was now a “requirement of command”, and our leaders would be required to “think at machine speed”.

Now, this, you might think, would please me, as the AI specialist in the faculty at the Royal Military Academy Sandhurst. Think of the funding!

But it actually troubles me a great deal. It’s not a different call to action than we see everywhere else. Perhaps more substantive because we are speaking to the adoption “in weeks and months, not years” of this technology in an institution as sclerotic as militaries tend to be.

But the problem with adopting a rapid approach to AI is fundamentally a problem of karstica.

See, AI is trained on precisely the media I just spent a section diagnosing. Decades of narrative-shaped, “interesting” content. Much of this content is going to be factual, but the models don’t seem, in the main, to learn facts from its training data. It learns the patterns. How to talk about the world. It learns those patterns from a corpus that’s been pre-filtered for the stickiness of ideas over their substance.

Worse, the way these models work is structurally inclined to produce karstica. I’ve written extensively about AI reasoning before, but the essential point is that AI produces “plausible continuations” rather than grounded claims. Humans reason from some kind of gut feeling of rightness and wrongness. We have an intuitive answer, and build out arguments to support it. These arguments are typically lazy, because reasoning is a social phenomenon. We only need our arguments to be strong enough to survive social friction. This is why we can convince ourselves of whatever we want, but our friends can easily convince us that that same thing is a bad idea.

AI reasoning has no such strong gut feeling. Simply whatever is implied by the pattern you activated with your interaction. More importantly, it has a strong bias to support your reasoning. It’s just you reasoning with yourself, with extra steps. And if you come to AI with no intuition of your own to drive the interaction, then you get whatever ‘intuition’ is informed by whatever latent conversational space you pushed it into with no guarantee of its groundedness. It’s essentially the AI equivalent of man-guessing. And, as I’ve argued elsewhere, since AI has fundamentally different purposes to humans—purposes that aren’t around navigating the world well, but rather around producing fluent patterns of language—then this ‘intuition’ it forms, and the arguments for it, aren’t even karstica that exists. It’s fresh karstica, on demand, custom fit to whatever you asked.

So, back to Chatham House. The directive from the Chief of the General Staff was to learn how to “think at machine speed”. The implication here is that we need to learn how to delegate our thinking to a karstica factory. If we’re experts, this is fine. Productive even. An expert with strong intuitions in a domain already treats AI output as a draft to test against what they already know. They’re going to catch the karstica. But in a training institution like Sandhurst, or for people trying to learn anything new, apply themselves to any new domain of thinking, they have nothing to test against. We take the surface for the substance, and build on it. A house on a karstic plain, waiting to be swallowed by the void beneath. Human and AI, hallucinating together.

¶Outro

Since I started this blog, over a decade ago, I’ve been worried about this:

If you don’t do the thinking, the thinking will be done for you.

Largely this concern of mine has been developmental. I write a lot about how useful ideologies are—they take an aspect of this chaotic world and make some of that chaos meaningful. The more you learn, the better off you’ll be.

But increasingly, I find myself worrying about this defensively. Karstica is often an accidental product of the way the human animal shares ideas. Interesting facts in storylike shapes are often helpful, but don’t need to be true. And the more one looks around, the more one sees this weaponised. People using the naming problem to hide questions that were never answered, people using Malcolm Gladwell shit to sell courses, people using linguistic laundering to distract us from troubling truths.

Now, though, a veritable karstic factory is taking centre stage in our lives.

The solution, hopefully, is obvious, if not necessarily easy to deliver on. The solution is to develop a new skillset. The ability to detect karstica.

There are already ways to do this. The academic track eventually moves from content delivery to teaching you how to ask the right questions, and get to the answers critically. A PhD is a case-study in exactly this—how to move past the karstica, and into solid ground. In a world where the value of higher-education is falling rapidly, this capacity—to teach critical thought—might well give it new life.

But not everyone has the time or appetite to study for a PhD, and the chances that universities will streamline that process “in weeks and months, not years” seems exceedingly slim. So it’ll be up to us to keep worrying about the fact that:

If you don’t do the thinking, the thinking will be done for you.