Surviving Capability
Why the field celebrates what the theory condemns. The practitioner’s notes praise the informal taxi swarm, the analog protocol, the constraint-born workaround — exactly the off-map divergence the theory treats as pathology. It looks like the corpus contradicts itself. It does not, and the answer is in three things I lived: two transport systems — one that repaired its map, one that froze — and Uber, which proves the point by changing the technology and nothing else. The field never celebrated the inefficiency. It celebrated the capability that had not yet been lost.
Witness · one
Barcelona, and a commute that put me in the street
I learned this in the flesh, young, as a student commuting into Barcelona from Corbera de Llobregat, out in the Baix Llobregat. On paper there was public transport. In practice, for a peripheral user, the map did not work. Fares were separate by operator and mode, so every change cost again — pay double, pay triple for a single journey across the network. The trips were long and stitched together from multiple transfers, and the connections were unforgiving: a small delay on one leg meant the next was gone. I lost classes. I lost events I had to be at. More than once a missed connection meant there was no way home that night and I slept in the street in Barcelona. That is not a complaint about hardship; it is a precise description of an insolvent map — a network that, for the user at the edge, could not deliver the outcome it was nominally there to deliver.
Witness · two
The reorganization that made the map work
Then the system was reorganized — pulled together under a metropolitan transport authority, the Autoritat del Transport Metropolità (created in 1997), into a single integrated network. In January 2001 its Integrated Fare System (the Sistema Tarifari Integrat) came into force: one ticket valid across the roughly fifty operators of the area, with free transfers between modes and fare zones inside a time window, so a change of line or mode was no longer a second fare. Before that date — the years I was commuting — every change cost again. The operator on my line — Autocorb, the kind of operator that under naïve per-trip accounting looked like it was losing money once transfers stopped being charged separately — was compensated for that externality out of a common fund, on the correct logic that the value created (more riders, a network that behaves as one) is captured system-wide, not trip by trip.
The result was not a subsidy propping up a loser. Ridership rose sharply; the operator ended up better off on volume; and the network began to serve the citizen and the participating companies at the same time — which is the only sensible meaning of «working»: it minimizes inefficiency and informational noise for everyone in it. European metropolitan transport, where it works, works like this — either it never broke, or it was restructured back into coherence. And this is precisely why Barcelona’s was the map exported to Tuxpan (Part II): not because Barcelona is admirable, but because the architecture functions, and a functioning map is a thing you can carry somewhere else.
Witness · three
The taxi, and what I actually celebrated
Now the opposite case, and the one that looks like me praising disorder. In Cuba, and elsewhere in the Americas, I found something I thought was marvellous: you could take a taxi for roughly the price of a bus, point to point, with the flexibility to go where you needed, the driver picking up others heading the same way, routes assembled on the fly with other drivers. As a capability it was real and it was good.
But look at the system around it. The taxi plaza sits idle most of the day — unused capacity, parked. When you do use it, you are paying that driver’s day or week, plus the state, on top. Low utility to the user, low utility to the driver who often barely lives on it, low utility to the system as a whole: few jobs, a price structure that serves almost no one well. Inefficient in every direction — and frozen, because there is no competition forcing a change, and because the decision that would fix it is collective, with too many interests, none of whom win in the end but each of whom defends their momentary slice. If a genuinely good taxi-service capability ever existed in Europe, I never saw it in my life.
So when I celebrated the American case, I was not celebrating the inefficiency. I was celebrating one specific thing:
The capability had not yet been lost. The flexible, multimodal, point-to-point service still existed — surrounded by inefficiency, but alive — where in the over-rationalized systems it had already been clarified out of existence.
That is the whole of it. The practitioner’s eye goes to the survivor.
Witness · four
Uber, and why the technology was never the point
One more case, because it isolates the variable cleanly. Uber never, in anything I saw, had to resolve an inefficiency — it only had to exploit whatever the local structure already allowed. Where a local system would hand out taxi plates without limit, for private advantage, Uber did the same thing at scale: it poured in licences and oversaturated the market consistently, because consistent oversaturation is what its own maximization rewards. There are cities sitting under exactly that — too many cars, every one of them under-utilised, the whole thing as inefficient as the frozen taxi system it replaced, and in some cases the local trade died of it.
And then there are the places where it worked — temporarily. Where there was a real, unmet demand, the entry of capacity helped: more supply against more demand, coordination through an app where before there was only a radio dispatcher, a lower price at the start. In those openings the injection was a genuine improvement. But notice that in neither case did the structure change. In Europe, after all the regulatory wars and the caps on the number of Ubers, the net effect was marginal, because the total system was not — could not be — restructured by the new entrant. As far as I have seen, Uber has nowhere modified a system structurally, because an actor inside the game has neither the need nor the convenience to do so, even where restructuring would be best for it in the long run. It is a maximizer, exactly as the theory says. A new maximizer with new rules injects some differential capacity into a structure that does not move, and so the improvement stays marginal.
Two layers in one system: the dying map, and the surviving capability
The apparent contradiction comes from reading «the system» as one thing. It is two. Every one of these cases contains a map — the formal architecture, the fares, the decision rights, the as-designed flow — and a stock of capability — the real adaptive ability to deliver the outcome. The theory condemns a divergent, insolvent map. The practitioner celebrates capability that is still alive. They are pointing at different layers of the same object, and they do not disagree: both treat the inefficiency as pathology. The only difference is where the eye lands — and the practitioner lands on the survivor because it is the perishable, recoverable asset. «Not yet lost» is a clock, not a virtue. You celebrate survivors the way you celebrate finding survivors, not the way you celebrate the wreck.
The taxi is not a counterexample — it is the theory exactly
A real capability, sitting in a Pareto-inferior equilibrium that no one wants, unable to move to the better state available to it. Why it cannot move is the sharp part, and it names a mechanism worth keeping:
The double-map deadlock. To move from a worse state to a Pareto-better one you need two maps: a map of the real current state, and a map of the target that is better for everyone. An insolvent, over-entangled system has neither — you would have to discover the real state before you could even specify a credible target, and the entanglement (many actors, no decision clarity) is what destroyed both maps in the first place. With no current map and no target map, the collective move cannot be computed; and with too many parties each maximizing a momentary private interest, the system locks. It stays Pareto-suboptimal permanently — not by anyone’s choice, but because the decision that would free it is uncomputable on the maps that exist. [Claim type: proposed construct — to be formalized and tested.]
This is the meeting point of three pieces of the program: the decide-before-you-discover bottleneck (you cannot target what you have not resolved), the collective-decision paralysis of accountability entropy (many claim a say, none own the move), and insolvency as a frozen state rather than a sudden death.
Uber isolates structure from technology — and proves the deadlock
The Uber case is the natural experiment that strips the question to one variable. The technology was maximal, and within a day it was identical — every clone had the same app, GPS and dispatch from the start. The better solution was fully visible; everyone could see it existed. And still the structure did not change. So the binding constraint was never the tool, and never the visibility of a better state. It was that decisions are made on the map, and on the map there is no path to the better state — the count of decision-makers, atomic decisions and dependencies between them leaves the Pareto-better outcome unreachable, however plainly it exists in the real world. That is the double-map deadlock shown rather than argued: knowing the destination changes nothing when the map contains no route to it.
Uber also behaves exactly as the theory’s agent taxonomy predicts. It is a frame-bounded maximizer, not a cross-frame coherent agent: it optimizes inside its own frame and has neither need nor convenience to restructure the system it enters, even where restructuring would serve it long-term. A new maximizer with new rules injects differential capacity; the structure stays put; the gain is marginal. And the sign of that gain flips along the program’s lethal-on-the-sick, safe-on-the-healthy line: into a genuine demand gap (a relatively healthy opening) the injected capacity helps; into an already-pathological structure (licences multiplied for private advantage) the same maximization oversaturates and kills. Same tool, opposite outcomes, decided entirely by the health of the structure it enters. This is the cleanest statement in the whole corpus of why structure matters more than technology.
The two exits — and why Barcelona is the jump and the taxi is the slow death
A frozen Pareto-suboptimum has exactly two ways out:
— Forced out by competition. When something is made to compete with it, the frozen system cannot reorganize in time and is driven from the market. This is the taxi’s fate, and the migration / slow-death pattern: the surviving capability does not get rescued in place — it drains to whoever can capitalize it, or it dies with the system.
— Reorganized by maps brought from outside. Barcelona is this exit. The metropolitan authority supplied what the frozen system could not generate internally: a coherent target map (the integrated network) and the externality accounting that made the Pareto-move both computable and fundable — compensating the operator whose per-trip ledger understated the value it created. That is revalue before you clarify, at metropolitan scale, brought in from outside the part of the system that profited from the murk. Tuxpan is the same move at city scale: the control center supplied the missing current-state map (observability independent of the drivers), which let the network correct itself without destroying the adaptive flow.
What this costs Lesson 6
It tightens an over-claim. The simple, constraint-born solution — the Cuban analog protocol, the paper transfer ticket — is not robust because it is simple. It is surviving capability filling the gap of an insolvent map, and it is celebrated because it survived and works, not because scarcity is a design virtue. The proof is Barcelona: once the formal map was repaired, the analog workaround was not needed. Simplicity under constraint is a symptom of formal insolvency that happens to be worth saving — not a general law that constraint produces better architecture.
Theory: The Cost of Clarity · The Attribution Gap and Capability Loss · The Map and the Flow · the whole program.
Practice: Tuxpan (II) · Chicontepec (III) · Entropy & Activation (IV).
In one line
The work is to get the survivors out
The field celebrates the survivors; the theory condemns the wreck; they are looking at the same system. The practitioner’s instinct is right and the theory’s verdict is right, and they fit together once you see that a divergent map and a living capability can occupy the same body — one dying, one not yet dead. The work that follows from both is the same: find the capability that is still alive, repair the map from outside before competition arrives, and revalue before you clarify — because a frozen Pareto-suboptimum, left alone, does not stay a curiosity. It becomes a closed exit, and the survivor goes down with it.
— Iván Abril Palma · a bridge note · The Integral Management Society