On not writing things down

I previously wrote about how pilot projects become contronyms as soon as they’re written down. And in my last post, I described how non-optimal business strategy tends to persist in many conditions where it is profitable to be vague. These ideas are linked in another way: both are instances of capture, where the way we fix our knowledge destroys what we know, and the parts we can’t ever fix in the first place.

Writing is hard

I often struggle to write big things down — I’ve talked about this before in the context of personal knowledge management. But I also tend to believe that most thinking begins with writing. How can we reconcile these ideas?

The operative word here is “most”, because at its core the writing isn’t actually the thinking. It’s rather that writing forces your thinking to deepen, which in turn sets off the core reinforcing feedback loop by which your writing improves. So both are cause and effect at once, and writing becomes more like a threshing floor for your ideas.

But writing also just seems asymmetrically hard compared to thinking! The ease with which wisps of really great ideas bubble up can make you feel like a terrible communicator when they’re relatively so much harder to pin down, or maybe just bad at deciding what makes an idea good in the first place. It’s sort of like when your incredible thought sounds stupid out loud.Not quite the same — talk is cheap, as they say — but the notion that supposedly great ideas go extinct upon medium shift is transferable. So even if you believe in the reflexive relationship between thinking and writing, there’s a certain rarified appeal to the view that writing is somehow more precise or more rigorous, as only the good ideas will have survived. Hence the “ideas guy”, and to a lesser extent the ivory tower.

Of course we can effectively critique this kind of logocentrist take along deconstructionist lines, but I’d like to suggest a different reason why writing is often so ineffective, and maybe why you shouldn’t exclusively write things down. To do so, we’ll have to sift through the candidate arguments and diagnose their underlying tensions.

Three problems with writing

The first reason why writing might be less effective than thinking is due to market-based reflexivity. Writing a technique down often negates the alpha it had previously allowed you to extract. Arbitrage disappears when everyone knows about it! It’s a bit like that Yogi Berra joke, or how your favourite hidden-gem restaurant or coffee shop or whatever relies on not being known to the masses. This is just the economic analogue of the contronym problem, where a pilot becomes a not-pilot once the name is attached. And that’s not so surprising, because it turns out that many things evaporate when crystallized, or even upon exposure to broader cultural awareness, like prototypes and niches and inside jokes. There are even things that everyone knows but that you can’t admit to knowing!Footnote 1The Secret of Polichinelle (opens in new tab) (archived Dec 2025). It’s not hard to think of things that don’t survive transcription.

The second reason comes from applying the Seeing Like a State legibility argument. For example, your last name is supposed to distinguish you from others, but its existence arguably makes you less of an individual in some sense, as you cease to be “the person who lives down the street and is pretty good at piano” and instead become “Daniel Spracklin”, or perhaps even just a maximally identifying yet technically pseudonymous entry in a database. Insofar as last names exist to be written,Footnote 2Or rather, per Scott, to be written down so the referent could be taxed. it’s a reason to question the motives behind the writing instinct.

A third problem with writing is that, no matter the effort you expend to distill your thoughts, the work product often comes out vague and unprincipled, and especially so in the strategy domain. This accelerates the “suboptimal amount of bad strategy” problem I wrote about last time, and my dissatisfaction with the genre stems from a question of reproducibility: if your strategy is so good, why isn’t everyone reaping the rewards?How could anyone fail competitively when adherence to the CEO-cum-NYT-bestseller-du-jour’s views on leadership styles costs $25? Well, maybe the value was never in the written word. As with tacit knowledge (opens in new tab), some things are harder to communicate than they are to know or to learn. And in fact as Friedrich Hayek suggests in “The Use of Knowledge in Society”, the kind of knowledge we care about here

… never exists in concentrated or integrated form, but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.

Hayek’s argument — that much of the real knowledge in life is quietly bound up in the minds of all the other people who also live down the street — isn’t merely that the central planner just needs to dig harder to unearth information. That perverse result already follows directly from the legibility problem. Rather, the knowledge Hayek writes about is fundamentally impossible to communicate or to “enter into statistics” in the kind of data-drivenAn epithet that deserves its own post. text that dominates strategy writing, or project charters, or what have you.

These challenges to the supremacy of writing are in constant tension. The existence of your arbitrage strategy doesn’t somehow prevent you from revealing it, it just makes disclosure dubious. The legibility argument doesn’t magically stop you either; instead it promises a kind of systemic destruction directly resulting from your efforts to impose textual order. And Hayek’s argument is actually one of sheer impossibility. Clearly they cannot all be simultaneously correct across the entire problem space. But what unites them is their suggestion that the true purpose and meaning of your ideas isn’t fully captured in the way you write them down. Some part of it all must live elsewhere.

The thought-river beneath

Rather than reaching for the corporate strategy literature, which would tie up the loose ends I opened at the start, I’d prefer to turn to cinema, which won’t. But I like film here anyway, as it often hides a key part of its meaning somewhere the camera isn’t. This is exactly how Deleuze conceptualizesFootnote 3See this lecture (opens in new tab) (archived May 2023). the works of Syberberg, Straub–Huillet, and Duras in Cinema 2: The Time-Image.Duras washes the others, although S–H have Too Early/Too Late going for them. Calling attention to the disjunctionProprement cinématographique, Deleuze calls it. between the visual and the audible, he says:

A voice speaks of something, someone is talking about something, at the same time as we are shown something else. That which one is speaking about is actually underneath what we see… La parole s’élève dans l’air… cela dont elle nous parlait s’enfonce sous la terre.

We can thus imagine the communication floating above the image; as sound- and visual-images detach uncrossably, the image sinks beneath the speech. Contra McLuhan, the medium is the message’s superstrate.If you hit the medium, keep digging.

However, we need to take care not to suggest that one mode is then foregrounded at the expense of the other. It’s not as if the speech act dominates, or conversely in our analogy that thinking represents some secret, all-consuming mystical force. Rather, as Deleuze points out, “neither of the two faculties is raised to higher exercise without reaching the limit which separates it from the other, but connects it to the other through separating it”. Uncrossable, yes, but “always crossed because uncrossable”. So too with thinking and writing: writing alone misses the Hayekian tacit, and thinking alone fails the threshing test if the bad ideas aren’t winnowed. The limit point creates the reinforcing feedback loop.

With this in mind, what would it look like for your strategy to sink beneath its own presentation in the same way? Start with the paucity of good strategy writing; most of the literature commits the obviously legible bits to paper and yet doesn’t — or can’t — share what makes the strategy really tick, because it’s impossible to give effect to the precise cultural boundary conditions that instantiate your problem space. Then we see why strategy wrongness so often equilibrates at non-zero levels: the cost of transmitting the unwritable exceeds many strategists’ textual credit limit.This metaphor doesn’t quite get things right, as there’s no cost function in either Hayek’s impossibility theorem or in Deleuze’s cinematic analysis. But from a corporate finance perspective, you’ll always pay more to transmit what you can’t write, and capturing it outright might require unbounded credit. It also explains why pilots function as contronyms: the labels we attach to our work are orthogonal to their meaning.

The same disjunction cuts across the ways we operationalize strategy. The “no surprises at the governance table” phenomenon is a good example; this is when a decision-maker confronted with an underbaked or improperly socialized idea — and critically, one whose unwritten foundation hasn’t yet been disclosed — doesn’t take it well. The formal presentation is supposed to float above a prior consultation, or an off-the-record discussion, or something. When those don’t happen, the meaning is captured entirely by the text — and more so by the surprise it produces. To say precisely what should have happened before the meeting would probably destroy its value. And yet my inability to say it is consequentially irrelevant, because in our example the presentation demonstrably fails to land anyway. So amid the uncommunicability of the problem, there must have been a there there.A more esoteric example is the Outlook Kremlinology that occurs when you ignore the text of an email and foreground the metadata: what time the email’s sent, who is and isn’t in cc, what order their names appear, and whether the boss spells the word “thanks” in full.

More broadly, this anticipates two common strategy failure modes. On one hand we have governance by pure execution, which can and often does end invisibly, with minimal credit advanced to the executors; on the other falls governance by PowerPoint, which fails the sundry writing traps I’ve laid out here. But the PowerPoint problem isn’t just a too-lossy knowledge-compression algorithm; that would merely restate Tufte’s Columbia disaster argument.Footnote 4Edward Tufte, “PowerPoint Does Rocket Science — And Better Techniques For Technical Reports” (opens in new tab) (archived May 2026). Tufte is not wrong that slide decks often leave out important contextual material — indeed, most PowerPoints are much too high-level — and that they can’t and shouldn’t substitute for detailed technical reports. But this is all orthogonal to my point: the real systemic problem arises whenever the collection of texts consolidates unilateral strategic authority. In the cinematic case, speech and image operate according to heautonomous logic: each is a complete code, but the art form insists that they cross amid their disjunction. In strategy, we need the equivalent — a tacit system that counterbalances the text without deriving any authority from it. Without this, we create the conditions for strategy to rot, regardless of how granular of an appendix we attach.

What lies underneath the written text is vast and unknowable, just like the inner lives of everyone you’ll pass on the street tomorrow. It is more difficult to grasp than in the cinematic case; you can’t map its dimensions. But I can tell you it exists in spite of that, and in fact the health of your system relies on that existence. That means this post too is merely the shadow of a realer idea sinking underneath the earth.