We Know More Than We Can Upload
The body did not receive the memo: on kung fu, ancient machines, and the knowledge that refuses shortcuts.
/* There is a scene in The Matrix that never left me.
Neo sits in a chair. A cable is plugged into the back of his skull. A few seconds pass. Then he opens his eyes and says, “I know kung fu.”
I have been thinking about that scene for years. Not only because it is cool, although it is very cool, but because it asks one of the most seductive questions modern culture has ever produced: if knowledge could be uploaded directly into us, what would we choose?
I used to play this game with myself. Five uploads only. No cheating. Over the years, the list changed. Programming. Languages. Martial arts. Medicine — with TCM and neuroscience included, which is obviously not cheating. Engineering. Some choices disappeared, others returned.
But the more I thought about it, the more the question became less about what I would upload, and more about whether the upload would count as knowing at all.
If I uploaded the complete theory of martial arts, would I know how to fight?
Maybe I would understand the forms. Maybe I would know the names of the movements, the mechanics of force, balance, leverage, breath. I might even be able to explain them beautifully.
But my body would not have fallen. It would not have corrected itself a thousand times. It would not know the distance between theory and impact.
That gap is where this essay begins. */
The Dream of Skipping the Becoming
The fantasy of instant knowledge is older than artificial intelligence, older than the internet, and probably older than books. Every age has imagined some way to bypass the slow, inconvenient process of learning.
The dream is always the same: to receive knowledge without passing through the labour of becoming someone who can hold it.
At first glance, this sounds efficient. Why spend years learning a language, practising an instrument, training a body, studying a discipline, or failing through apprenticeship, if the knowledge could simply arrive?
But the question hides a problem.
Not all knowledge behaves like information.
Some knowledge can be written down, stored, copied, searched, compressed, translated, and transmitted. Other knowledge lives in timing, pressure, proportion, judgement, touch, rhythm, attention, and correction. It is not exactly secret, but it is not fully portable either.
You can describe it. You can point towards it. You can even teach around it.
But you cannot simply upload it.
A person can read every manual on swimming and still drown in deep water. A surgeon can study anatomy for years and still need a hand that knows how much pressure is too much. A designer can memorise every principle of composition and still not know when a layout is almost right but not quite alive.
There is knowledge that enters through language.
And then there is knowledge that has to pass through the body.
The Bruises Were Not Included
This is why martial arts are such a perfect example. They expose the fantasy immediately.
Theory matters. Names matter. History matters. Anatomy, strategy, discipline, philosophy, breath, stance, distance, force — all of it matters. But none of it is the same as being in motion with another body in front of you.
A body learns differently from a mind reading a sentence.
It learns through repetition, resistance, pressure, fatigue, bruises, corrections, and the very specific humiliation of discovering that knowing what should happen does not mean being able to make it happen.
The mind may understand balance.
The body has to lose it.
The mind may understand timing.
The body has to arrive too early, too late, too stiff, too slow, too confident, too distracted, too theatrical, too afraid.
Eventually, after enough correction, something changes. The movement stops being an instruction and becomes a response. The body no longer translates: step here, turn there, shift weight, guard, breathe. It simply does.
This is procedural memory. Muscle memory, if we use the casual phrase, although muscles themselves are not exactly writing memoirs. It is the nervous system, the senses, the joints, the small calibrations below conscious explanation.
The body stores skill as readiness.
Not as a paragraph.
That is why expertise so often looks quiet from the outside. The expert does not seem to be doing more. Often, they seem to be doing less. Fewer movements. Fewer corrections. Less visible effort. More precision.
The beginner performs the rule.
The expert has metabolised it.
No Magic Wand for Mastery
Michael Polanyi gave this problem one of its cleanest formulations:
we know more than we can tell.
He called it tacit knowledge.
Tacit knowledge is not mystical. It is not vague intuition dressed in expensive clothes. It is the part of knowing that cannot be fully reduced to explicit instructions.
A person can recognise a face without being able to list the exact measurements that make it recognisable. A musician can hear that something is off before they can explain why. A craftsperson can feel when the material is resisting. A designer can look at a page and know that the spacing is wrong, even before measuring it.
This kind of knowledge is not anti-rational. It is not below thinking.
It is thinking that has become distributed across perception, movement, memory, and judgement.
This is why craft traditions have often preserved something that modern systems keep rediscovering with surprise.
In Japanese, the word shokunin is often translated as ‘artisan’, but the translation is too small. It does not only describe a person who makes things well. It carries a sense of devotion to the craft, responsibility to the work, accumulated discipline, and a standard that lives inside the maker rather than in a checklist.
That is difficult to upload because it is not merely a technique. It is a relationship to technique.
A shokunin does not only know the steps. They have spent years becoming the kind of person who can feel when the steps are not enough.
The modern world tends to undervalue this because explicit knowledge scales better. It can be documented, indexed, automated, processed, formatted, and sold. It fits into databases. It behaves politely in presentations. It can be turned into frameworks, courses, prompts, guidelines, manuals, dashboards, and onboarding documents.
Tacit knowledge is less cooperative.
It lives in people, in practices, in traditions, in the slow transfer between master and apprentice, in mistakes corrected by someone who has already made them, in context that no checklist fully captures.
This does not mean explicit knowledge is shallow. It means explicit knowledge is only one layer.
The map is useful.
But walking still uses the feet.
Knowing the Rules Does Not Win the Game
This is also why rules never fully explain their own application.
A rule can tell you what to do, but not always when, how much, in what proportion, under what pressure, with what exception, or at what cost.
‘Make it simple’ is a rule.
So is ‘make it memorable’.
So is ‘make it clear’.
But the real work begins when these rules conflict, which they almost always do. Too simple becomes empty. Too memorable becomes loud. Too clear becomes dull. Too refined becomes sterile. Too expressive becomes overwhelming.
The skill is not in knowing the rule.
The skill is in knowing what the rule is doing in this specific situation.
This is where judgement enters. Judgement is what happens when knowledge meets reality and reality refuses to behave like an example.
It is why two people can have access to the same information and produce completely different outcomes. One follows the instruction. The other understands the condition behind the instruction.
The difference may look small.
It rarely is.
A novice asks, ‘What is the correct method?’
An expert asks, ‘Correct for what?’
The Cup, the Machine, and the Missing Hands
Some objects make this gap visible because they survive the people who knew how to make them.
The Lycurgus Cup is one of those objects.
At first glance, it is already impressive: a Roman glass cup from the fourth century, intricate, delicate, technically astonishing. But the real trick is in how it behaves with light. Lit from the front, it appears green. Lit from behind, it glows red.
Today, we can explain this through the presence of tiny particles of gold and silver in the glass. In modern terms, the effect belongs to what we would now call nanotechnology. We can analyse the effect. We can reproduce similar optical behaviour. We can look at the cup and say, with some satisfaction, that we understand what is happening.
But do we understand the knowing that made it possible?
That is a different, uncomfortable question.
How did they know? How many experiments did it take? How many failures? Was the effect discovered by accident and then refined? Was it a rare workshop secret? Was it part of a broader glassmaking tradition that we can only see through a few surviving traces? Did the makers understand the mechanism as we understand it now, or did they understand the behaviour of the material through practice, recipe, fire, colour, timing, and loss?
To reproduce an effect with modern tools is not the same as recovering the world of practice that produced it. It does not tell us how the makers discovered the method, how much was intentional, how it was taught, how often it failed, how many workshops understood it, how much was craft habit, how much was experiment, and how much belonged to a lost chain of hands.
The object remains.
The living knowledge around it does not.
The Antikythera mechanism raises the same problem on an even grander scale. A complex ancient Greek astronomical device, built with gears and mechanical sophistication that still feels almost implausible, it could model celestial cycles with extraordinary ambition.
We have fragments. We have scans. We have reconstructions. We have theories. We can admire it, analyse it, model it, and build versions of it.
But the mechanism itself points to a world larger than itself.
It suggests workshops, tools, mathematical traditions, observational habits, mechanical imagination, and a continuity of skill that did not begin and end with one object. Something like that does not appear from nowhere. It comes from a culture of making.
And yet, for centuries, that lineage was gone from view.
This is the strange sadness of certain artefacts. They prove that something was known. They do not automatically restore the knowing.
A thing can survive the civilisation of practice that made it.
The Tribute Band Problem
This is not an argument against modern reproduction. Reproduction is valuable. Reconstruction is valuable. Scientific analysis is valuable. Without that work, many of these objects would remain mute.
Sometimes the copy is cleaner, stronger, safer, more precise, or technically superior to the original. Sometimes we do not simply recreate something; we improve and upgrade it in ways the original makers could not have imagined.
But we should be careful about the kind of victory we are claiming.
When we reproduce an ancient effect, we answer one question:
Can we make this happen again?
But the lost lineage asks another:
How did they know how to make it then?
Those are not the same question.
One is a technical achievement.
The other is a historical, embodied, cultural mystery.
There is something humbling about this. We like to imagine history as a straight staircase of progress, with us standing comfortably near the top. The past becomes a sequence of charming limitations that modernity has corrected.
Then an object appears from two thousand years ago and politely ruins the story.
It does not mean the ancients were aliens, although one should never close a door too dramatically. It does not mean modern people are foolish, or that every lost technique was great. It means knowledge does not move through history as neatly as we pretend.
Some things are discovered, lost, rediscovered, misunderstood, preserved by accident, destroyed by conquest, hidden in workshops, killed by industrial change, replaced by cheaper methods, or simply abandoned when the people who knew them died without passing them on.
Knowledge is not only stored in books.
It is stored in hands, tools, habits, economies, rituals, materials, apprenticeships, and reasons to keep doing something the difficult way.
When those conditions disappear, the knowledge may not vanish immediately. But it begins to detach from life.
Eventually, it becomes an artefact.
Then a mystery.
Then a research paper.
The Upload Came Back as a Prompt
Artificial intelligence has brought the upload fantasy closer than ever, at least psychologically.
No cable in the skull, unfortunately, but the emotional promise is familiar: instant access, instant synthesis, instant output. Ask and receive. Translate, summarise, imitate, generate, optimise, explain.
It is extraordinary.
It is also very easy to mistake access for mastery.
AI can compress the visible surface of a field with impressive speed. It can explain concepts, compare theories, generate drafts, produce variations, simulate tone, outline processes, and make an unfamiliar domain feel temporarily less foreign.
This is useful. Sometimes more than useful.
But it does not remove the old problem. It intensifies it.
Because when information becomes easier to access, the remaining value shifts towards everything that is harder to access: judgement, taste, timing, responsibility, context, embodied skill, lived correction, and the ability to know when the output is wrong in a way that sounds right.
The danger is not that AI gives us knowledge.
The danger is that it gives us the feeling of knowledge before we have earned the structure to hold it.
A person with no grounding in a field may receive a fluent answer and mistake fluency for reliability. A person with some grounding can use the same answer as a starting point, a provocation, a draft, a map, a sparring partner.
The tool is the same.
The receiver is not.
This is why expertise does not disappear just because explanation becomes abundant. If anything, expertise becomes more important, because someone still has to judge which explanations survive contact with reality.
The upload can provide theory.
It cannot provide consequences.
You Still Have to Walk the Talk
Maybe the real mistake is imagining knowledge as a substance that can be transferred intact from one container to another.
Some knowledge does behave that way. Dates, formulas, names, facts, procedures, definitions. These can travel quickly, and thank goodness they can. Civilisation depends on not having to rediscover everything personally.
But deeper knowing is less like receiving a file and more like becoming compatible with one.
You have to develop the body that can perform it, the eye that can notice it, the patience that can refine it, the judgement that can question it, and the humility to be corrected by it.
This is true in martial arts.
It is true in design.
It is true in science, craft, writing, parenting, medicine, music, leadership, cooking, building, and probably every field where reality has the nerve to be more complicated than the instruction.
We can upload information.
We can accelerate learning.
We can preserve more than ever before.
We can analyse ancient glass, reconstruct ancient machines, simulate expertise, generate explanations, and stand on top of unimaginable archives.
But the oldest problem remains.
Knowing is not only what enters the mind.
It is what changes the person.
And still, if anyone ever offers me the upload, I am not going to pretend I would refuse.
Of course I would try it. At the very least, it would be an excellent base layer.
Whether I would beat Morpheus in sparring is another question.
For that, apparently, my body would still have to show up.
© Alexandra Mark
© Alexandra Mark