Wet nodes cannot share a substrate. Two minds do not sit on the same silicon. They sit in different bodies, with different histories, and the only thing that crosses between them is a signal.
That signal is what I mean by a wet protocol. Natural language. A look. A tone. An image. A poem. The means by which one wet node conveys something to another wet node without ever being able to transmit the whole state.
And the defining property of a wet protocol is that it is ambiguous. Not by accident. By necessity.
Ambiguity is the mechanism, not the defect
A dry protocol tries to transmit the full state. It compresses, checksums, retransmits until the signal is clean. Ambiguity is noise, and the whole apparatus exists to eliminate it.
A wet protocol cannot transmit the full state. The full state includes felt consequence — the body's memory of what it cost to learn a thing. That is not transmissible. It lives in the node and nowhere else. So the wet protocol sends a partial signal, and relies on the receiver to complete it, using their own interior as the completion.
This is the point that is almost always missed. The ambiguity in a wet protocol is not a bug in need of correction. It is an invitation to the other node to do the reconstructive work that only a wet node can do. When I say I am tired, you do not receive a specification of my fatigue. You receive a word and a tone, and you complete them with your own fatigue, your own history of being tired, your own felt knowledge of what tiredness does to a person. The signal is ambiguous. The reconstruction is specific. And the specificity comes from you, not from me.
Which is why the same sentence lands differently in two people. Which is why a poem can carry more meaning than a memo. Which is why a wet node cannot be copied — you can copy the words, but you cannot copy the receiver's interior, and the protocol only supplies half the message.
Ambiguity is the mechanism by which two sealed interiors can touch without either one dissolving into the other.
Rules bound the ambiguity
If wet protocols are ambiguous, then any group of wet nodes trying to coordinate has a problem. The signal is never fully specified. The receiver has to reconstruct, and reconstruction can go wrong. Left alone, ambiguity drifts into incoherence, and the network loses the ability to act together.
Rules, oaths, and laws are the stabilisers. They do not remove the ambiguity — they cannot, because the ambiguity is load-bearing. They bound it. They say: reconstruct within these limits, and the collective will hold.
An oath is different from a law. A law constrains from outside, with penalties. An oath constrains from inside, and the penalty is identity loss. A medical oath is not enforced by a court. It is enforced by the fact that a doctor who breaks it stops being the kind of thing a doctor is.
Which maps onto the same distinction that runs through Care. A law is extrinsic constraint. An oath is intrinsic. Which means an oath is wet, and a law is hybrid — it only works if enough wet nodes have internalised it.
Professions make rules because their protocols are especially ambiguous and their stakes are especially high. Medicine. Law. Teaching. Do no harm is ambiguous. Act in the patient's best interest is ambiguous. Beyond reasonable doubt is ambiguous. The rules exist precisely because the protocol cannot resolve the ambiguity on its own, and the stakes are too high to leave it to individual reconstruction.
But rules can conflict
Here the architecture acquires its first real crack.
A wet rule is stored in a node not as a rule but as a disposition, a habit, an identity, a felt obligation. You cannot inspect it the way you inspect a rule table. You cannot see that two rules are in conflict until they fire.
And wet rules can be in conflict. Not because any one of them is wrong, but because they were written at different layers of the network, by different generations of the profession, for different situations, and no one ever ordered them against each other.
A training rule says: never perform a procedure you have not been signed off for.
An oath says: act in the patient's interest.
Both are ambiguity-bounding. Both are doing their job. In almost every situation, they agree. But not in all of them. And when they disagree, there is nothing in the node that says which one wins.
Dry networks have fallbacks
When two rules fire against each other in a dry system, there is a defined resolution path. Priority ordering. Exception handling. A default that fires when no rule matches. A halt state that is recognised as a halt rather than experienced as paralysis. Conflict is a named condition with a defined exit.
The system does not freeze. It resolves, escalates, or stops in a way that is legible to the layer above.
This is not an intelligence. It is a design property. The fallback exists because someone built it in. Because a dry system that deadlocks is a dry system that has failed, and the engineering culture treats deadlock as a bug to be fixed.
Wet networks have no equivalent. A wet node holding two conflicting rules has no fallback defined for that state. Its training did not cover rule conflict. The oath did not cover it. The protocols did not cover it. When two wet rules fire against each other under time pressure, the node is on its own.
And the node does the only thing it can. It freezes.
Freezing is not a character flaw
From outside, a frozen node looks like a coward. From inside, it is an unhandled exception. The wet node has encountered a state its rules do not resolve, and it has no mechanism to escalate, no clock to wait on, and no clean halt. What it has is a body, flooded with stress hormones, holding two contradictory obligations and no precedence between them.
Wet nodes cannot halt cleanly. A dry system that hits an unresolvable conflict can stop and report. A wet node in a crisis cannot stop. The situation demands action, the body floods, and the lock manifests as paralysis rather than as a clean halt. The halt state is indistinguishable from failure.
And because the wet node experiences the conflict as dread rather than as a rule table, it cannot debug the conflict. It cannot see the two rules. It can only feel the paralysis they produce.
Which means the network does not learn from the freeze. The node is blamed, the incident is filed, and the same conflict recurs the next time, in the next wet node, for the next person it happens to. Because the network has no mechanism for turning a freeze into a rule change.
This is where the dry contribution lives
The dry system's role in a hybrid network is not to decide. It cannot decide — it has no stake in the outcome and no interiority to complete the signal with. What it can do is something the wet network structurally cannot do for itself.
It can hold the rules explicitly. It can detect when two of them fire against each other. It can flag the conflict before the node freezes. It can route the node to a wet authority who can resolve the conflict at the moment of crisis, and record what that authority decided.
None of that is a decision. All of it is a fallback — the missing state that the wet network has never had.
The dry system does not become wise. It does not become wet. It supplies a specific structural property that wet networks lack: the ability to notice a rule conflict from outside the node that is experiencing it, and to route around the freeze before it happens.
That is the hybrid claim, stated in one line. It is not that dry systems make wet networks wiser. It is that dry systems make wet networks legible to themselves at the points where the wet network's own invisibility is fatal.
And then there is the slower problem
Wet protocol improvement is inherently slow. Not because wet nodes are slow, but because the improvement has to be paid for in irreversible experience. A protocol changes when enough wet nodes have lived through enough cases to know that the old version fails and the new version works. That is not a computation. It is generational.
The slowness is the safety mechanism. A wet protocol that changed fast would be a protocol that had not been tested across enough bodies, enough contexts, enough interiority. The slowness is what makes the improvement count as improvement at all.
Which means the dry contribution to wet protocol improvement is not speed. It is memory. The dry system can hold the accumulated precedents — millions of pinned resolutions — in a form a wet node can retrieve at the point of decision. It can flag where a protocol is failing more often than it should. It can preserve the reasoning behind a protocol, so that when the protocol is transmitted to the next generation, the why goes with the what.
The slow accumulation stays slow. The dry system makes it legible. That is the correct division. Anything faster is the wrong kind of fast.
What comes next
The two pieces that follow are the argument in its concrete form.
Precedence is an oath, a training rule, and a doctor who froze — the missing fallback made specific in a room with a dying patient and no one else to call.
Conformance and Care is the same failure recurring over decades in a wet network that has no mechanism for learning from it — thirty-two years of watching the visible proxy eat the invisible good.
Both are examples of what wet protocols do when they have no fallback and no memory. Both are arguments for the hybrid, made in the only register that finally matters: what happened to a specific person at a specific time.