The time machine is a lined notebook

Posted on: 23 April 2026

During my last move I came across the notes from a 2007 meeting at the BFI on the Southbank, a room overlooking the Thames, with a group of venture capitalists we were trying to persuade to invest in Microcinema. The project was ambitious for the time: digital cinema distribution via satellite, an infrastructure that used Eutelsat to deliver films to cinemas without the reels and the couriers. It looks obvious now, it didn't then, and the VCs looked at you as if you were explaining radio to Marconi fifteen years too late, or fifteen years too early, depending on their mood.

I have poor handwriting, the sort that deteriorates year on year, and in that meeting I had started drawing. Small sketches on an A4 grid pad, one centimetre squares, the geometry as an illusion of precision, the comfort of a frame already there to contain thought-sketches before they slid sideways. Little triangles for the players in the system, arrows for the flows, a stylised satellite that looked more like a starfish, cinemas as small aligned squares. A sort of proto-slide to pin concepts down, a private language that served no operational purpose. It served me, to understand the situation better in retrospect, when after the meeting I would have to reconstruct what had been said and what had not been said.

I found and reopened that pad eighteen years later, expecting, as one does on these occasions, to discover something new in the notes, an underlined phrase that had seemed unimportant then and now resonates, a warning I had written to myself and then ignored. That did not happen. I saw nothing in the sketches that I had not already seen in 2007. I saw something different, and more interesting: I walked back into the room.

Not metaphorically, I would almost say in a dreamlike way. The meeting returned as a film, complete: the arrangement around the table, the grey light coming off the river, the faces of those speaking and those staying silent, the tone of voice with which one particular partner had dismissed a technical objection of mine, the physical sensation of understanding halfway through the discussion that we were losing the deal and that it was a matter of language rather than substance. My mental state from then was accessible with a clarity that my mental state from six months ago no longer has. Eighteen years against six months, and the notebook won.

If one takes this seriously, it says something specific about how memory works when it is anchored to physical traces produced with cognitive delay. In recent years I have progressively stopped using digital note-taking systems. Not out of ideology, not even out of suspicion, but more out of empirical observation: I was accumulating increasingly well-ordered archives of material I never reread. Keyword search returned the exact note I was looking for, but the exact note was usually useless, because what I needed was not the content of the note but the context in which I had written it. That context evaporates in digital systems. The notebook preserves it, because it contains traces the digital system does not record: the handwriting degraded by a tired day, the ink of a different pen grabbed in a hurry, a coffee stain, a sketch made while someone was talking about something that didn't interest me.

Memory researchers use the distinction between semantic memory and episodic memory: the first preserves facts, the second preserves complete experiences. Now, fast-feedback technologies, designed for informational efficiency, are optimised for semantic memory and punitive for episodic memory. What you retrieve is the data, not the moment the data reached you. Data without the moment is castrated information, because judgement, the real kind, the sort you use when the decisions matter, forms from the combination of data and context. Incidentally, it is the same mechanism that explains why some people get useful work out of LLMs while others get either banal answers or saccharine agreement.

This is worth spelling out, otherwise the argument sounds like nostalgia for paper, and that is a lazy reading. The problem is not technology. The problem is the feedback speed that certain technologies impose and others do not. A well-built digital archive, an abandoned card index, a lined notebook filled slowly, a bibliographic management system used for twenty years: all of these can function as slow-feedback architectures, if whoever uses them accepts the delay between capture and rereading. The trouble starts when the platform pushes towards instant retrieval, towards keyword search, towards the absence of friction between thought and its registration. In that condition episodic memory no longer forms, because nothing forces the thought to settle into a specific physical form tied to a specific moment.

When in 2007 I drew that awkward satellite on the pad, the gesture tied three things together: the concept I was trying to represent, my poor handwriting, and the mood of that meeting. Eighteen years later, seeing the sketch activates all three simultaneously and the memory is complete. Had I typed the same notes on a laptop in the room, I would have saved the concept and lost the other two. I would have more information and less memory.

The question I can hear coming from the sophisticated reader is whether we are not romanticising inefficiency. A fair question, and one I raise myself whenever I hear the word "slow" attached to anything. The honest answer is that we are romanticising nothing. We are describing a precise cognitive mechanism: certain decisions, strategic ones especially, improve when whoever is taking them has access to many past mental states to compare with the present one. Not to much past information, to many past mental states. Mental states are preserved only if they have been anchored, at the moment of their formation, to specific bodily traces: a place, a gesture, an object, a handwriting, a bad sketch.

Whoever operates on fast feedback accumulates information and loses mental states. Whoever operates on slow feedback accepts the loss of information but preserves mental states. The second, at equal accumulated experience, produces better strategic judgement than the first, because they can compare themselves-now with themselves-then. Were this thesis false, we would observe that professionals in fast-feedback domains, algorithmic trading, social media management, continuous A/B optimisation, produce superior strategic judgement to those in slow-feedback domains, diplomacy, old-school private banking, literary criticism, restoration. The empirical evidence of the past thirty years runs systematically the other way, and those who work in those fields know it, even if they rarely articulate it.

There is a further, subtler objection. One could argue that the ones we still see working well on slow feedback are the good ones who made it through for other reasons, and that the method is therefore irrelevant. A serious objection, but testable: observe what happens when a slow-feedback professional is forced to switch to fast feedback. The deterioration of judgement is visible, documentable, and almost always described by whoever suffers it as the loss of something they cannot quite name. That something is access to their own past mental states.

Back to the notebook, because that is where abstract reasoning meets practice. In recent months I have started keeping one again, without elaborate schemes, without Bullet Journal systems, without stationery fetishism. A simple A5 lined notebook, a fountain pen, and the implicit rule that only what seems worth the gesture of writing by hand ends up in there. No longer the grid pad of then, because in the meantime I have understood that the geometric cage was a promise of precision the thought-sketches didn't actually need. The lines are more than enough, and they are quite happy to be ignored when the occasion calls for it. I write badly, I keep drawing sketches, and every so often, months apart, I reopen it. I am not looking for information, I am looking for mental states, mine, past ones, that I want to compare with the current one. The comparison produces, almost always, a cleaner judgement than the one I would have got by relying on working memory alone.

The sense of having become wiser does not depend on having unplugged, which would be magic, and magic does not interest me. It depends on having rebuilt, without knowing it at first, the physical conditions in which judgement can mature: sufficient silence, tolerated waiting, repeated observation of the same object at different moments, the possibility of changing one's mind without having to announce it publicly. The notebook with the bad sketches is one of the instruments of this reconstruction. Not the only one, not even the most important. But one I have verified empirically, with eighteen years of delay, and one I have gone back to using.

Proust understood all this without needing neuroscientists, and he tied it to a madeleine dipped in tea. He understood that certain objects, certain flavours, certain gestures contain, compressed, whole past mental states, and that their reactivation produces a different quality of thought, deeper, more capable of judging what matters. Today the madeleine, for those fortunate enough to have built themselves one, is often a lined notebook with sketches inside that would make no sense to anyone else. Worth keeping. Worth, above all, continuing to fill, in the knowledge that every bad sketch drawn now is building, without asking permission, the archive of one's own future mental states.

If you know, you know. If you don't, you're probably looking for a better app.