The Levels

Between a thought and a thing that stands in the world lie several floors. Each has its own language, its own touchstone and its own personnel. Whoever fails to keep the floors apart explains the failure of undertakings by the qualities of the people involved — by short-sightedness, by inertia, by greed. Whoever does keep them apart sees something else: a wiring that forces certain outcomes even when everyone involved does their job well.

We describe the wiring. Not its occupants.

I. What a level is

Without a criterion, any such enumeration becomes arbitrary. We propose three features that must be met together.

A level has a validity test of its own. It examines the matter against a question that is asked on no other level.

A level has a profession of its own. There are people whose training, language and career consist in administering that test.

A level has a veto. It can halt the matter by itself, and no other level can overrule it.

The third feature is the decisive one. It distinguishes a level from a mere influencing factor. Weather, chance and the spirit of the age act upon undertakings, but they halt nothing. Levels halt.

II. The eight levels

Knowledge tests for truth. Its profession is science, its veto refutation. It is the only level whose holdings persist independently of everything else: libraries are full of secured knowledge from which nothing ever came. That makes it the foundation and at the same time the only floor that can stand on its own.

Fiction tests for coherence. Here the matter exists first — as a design, a drawing, a description of a thing that does not yet exist. A patent specification is the pure case: it describes something absent so precisely that a legal claim adheres to it. Fiction here is not a disparagement but a statement of condition.

Realisation tests for function. Its profession is designers, craftsmen, technicians; its veto is not working, and it is the only veto that is not negotiable. A thing that does not run does not run even when everyone wants it.

Commerce tests for demand. The question is not whether the matter is good but whether someone who did not invent it will pay for it. Its veto is the order that fails to arrive.

Finance tests for return and term. Not whether something yields, but whether it yields enough, fast enough, compared with other uses of the same money. Its veto is the withdrawal of funds, and it takes effect at any stage.

The framework tests for admissibility. Standards, laws, approvals, regulations. Its profession is authorities, committees, testing bodies. Its veto is prohibition — and it is the only one against which success on all other levels avails nothing.

Adoption tests whether what people actually do changes. This level is regularly confused with commerce, although it is separate from it: bought is not used. Its veto is non-use. It falls late, silently and without an addressee — the thing is procured, installed, invoiced and then circumvented.

The public tests for assentability. It manifests itself in public opinion, and this is shaped essentially by the media — its profession is editorial offices and journalists, today joined by the operators of the distribution channels. Its veto has two forms: silence and scandal. The second is feared, the first is more effective, because it remains invisible. Against a no one can argue; against not being noticed one cannot.

This level differs from all the others in one respect, and the difference is fundamental. The other seven test the matter. The public tests its representation. What is at issue is never the thing but the image of it in circulation. Its question is not whether something is true, functional or admissible, but whether one can want it publicly without paying a price.

Seven levels form a stack. The eighth stands athwart it.

III. Three kinds of connection

The levels are not joined by a single kind of relation but by three, which must be kept apart.

Supply runs upward. Knowledge feeds fiction, fiction feeds realisation, realisation feeds commerce. This direction is rich in information: what goes up carries reasoning, evidence, drawings, calculations with it.

The veto runs the opposite way, and here lies the first structural finding. The return path does not have anything like the same bandwidth. A thousand pages go up; one bit comes back: yes or no. A rejection almost never contains the reasoning from which something could be learned. The proposer learns that it will not do — not at which point his translation failed.

In control terms this is a mutilated feedback path. A control loop whose feedback transmits only the sign of the deviation and not its magnitude cannot settle. It can oscillate or come to a halt. Both are to be observed.

Retroaction is the third kind and the only one that really holds the stack together. Realisation generates new knowledge: every thing that runs is an experiment proving what was previously only asserted. Commerce generates new demand and thereby new fiction. And the framework arises in the first place from congealed experience — a standard is knowledge that has changed levels and become binding there.

IV. The translation problem

Each level accepts only its own currency.

Realisation understands tolerances, not intentions. Commerce understands benefit and price, not function. Finance understands return and term, not benefit. The framework understands categories, not individual cases. At every transition a translation must be made, and at every translation whatever the target language has no word for is lost.

From this follows an observation that changes one’s view of institutions: the most effective ones do not sit on a level but on a boundary. The patent lies between fiction and framework. The standard lies between knowledge and framework. The balance sheet lies between realisation and finance.

The prototype is the most instructive of these boundary cases. It lies between fiction and realisation and belongs wholly to neither. For fiction it is already too committed — it has settled on one of many possible executions and thereby given up the others. For realisation it is not yet finished enough — it meets no specification, holds no tolerance reliably, carries no approval. Both levels judge it by the currency of the other, and in both it fails. That is the structural ground of its financing problem: not a shortage of money but a shortage of jurisdiction. There is no level whose test it could pass.

V. The actual finding: the time constants

The levels run at wholly different speeds.

Three of these values can be documented, and they mark out the span.

The framework. German standardisation procedure is governed by DIN 820-4. After publication of a draft standard the public has four months to submit comments; over a further three months the working committee deliberates on them. Seven months for the closing phase alone, before which lies the committee work proper. At the latest every five years each standard is reviewed as a matter of routine — which is also the lower bound on this level’s reaction time.

Finance. Around the year 2000 execution times in exchange trading stood at several seconds; by 2010 they had fallen to milli- and microseconds. The London Stock Exchange states an average latency of 126 microseconds for its trading platform. In 2015 the European securities regulator proposed synchronising trading clocks across Europe to within one nanosecond.

Between the slowest and the fastest level there lie some twelve orders of magnitude. A standard responds in years, a trading algorithm in millionths of a second. Both decide on the same matter.

If a fast control loop is coupled to a slow one and the fast one is given the veto, the slow one can never finish. It is broken off before it has completed its first full period. The termination is not a judgement on the matter — there is no time for a judgement. It is the inevitable consequence of a wiring in which the fastest level decides over the slowest.

This explains, without imputing any intention, why long-cycle undertakings systematically fail to come about. They are not rejected. They are timed out.

VI. A case in which it becomes visible

The fixed Fehmarn Belt link between Germany and Denmark is an example that requires no technical background.

In October 2013 the Danish promoter applied for the opening of the plan approval procedure for the German section of the tunnel. The procedure took seven years. The competent state authority received 3,100 objections; every single one had to be dealt with in hearings. For one section the plan approval decision was issued only in March 2024 — more than ten years after the application.

Construction had originally been envisaged to start in 2014 and the link to open in 2020. Opening is now expected in 2029.

Nine years of delay, and no one rejected the undertaking. There was a treaty between two states, there was the financing, there was technical feasibility. What there was not was a level that could have worked faster without violating its own validity test. For the 3,100 objections are not a mishap of the procedure, they are its purpose: the framework tests for admissibility, and admissibility means that those affected have been heard.

Here it becomes clear what the model does and does not do. It does not say the procedure is bad. It says that a level with a time constant of years sits in a structure whose other levels reckon in months and quarters — and that every undertaking which must pass through this level thereby acquires a property that has nothing to do with its merits.

VII. The same pattern, twice

The public does not stand in the sequence of the other levels. It acts upon the framework. What is admissible follows what is publicly held to be admissible — with delay, but reliably. Laws and standards are the slow precipitate of a fast movement.

The same constellation thus appears a second time in the building. Public opinion reckons in days and weeks, the framework in years to decades. A fast loop commands a slow one — exactly as finance commands realisation.

Two occurrences are no longer coincidence but a pattern. Wherever a fast level commands a slow one, what arises is not the slow outcome in poorer quality. A different outcome arises: excursion instead of settling. The framework gets occasion-driven legislation that answers the last scandal and is corrected at the next. Realisation gets undertakings cut to a quarterly horizon whose construction time exceeds the attention span of those funding them.

In both cases the diagnosis is the same. Not a want of will, but faulty coupling.

VIII. Two asymmetries

The first we have named: knowledge does not need the other levels. It is the foundation of the stack and at the same time independent of it. Every other level presupposes at least one beneath it.

The second is of more recent date and the more consequential. The financial level has detached itself from the stack. It brings forth products that pass through no realisation — claims upon claims, whose value derives from other claims, without anything being manufactured anywhere. It thus reaches directly from fiction to return and skips everything in between.

There are therefore two routes through the building. One leads through eight floors, each with its own test and its own veto. The other leads from the ground floor straight to the roof. Whoever can take the second never has to make anything. Whoever cannot must go through all eight — competing for the same funds with those who take the short cut.

IX. Beginning and end

The stack begins in fiction. The design describes something that does not yet exist and treats it as though it did.

The stack likewise ends in fiction. The financial level trades in claims upon a future that does not yet exist and treats them as though it did.

Both ends are the same operation: making present what is absent. The difference lies only in the object. One fiction is about things, the other about numbers. Between them lie six floors whose task is to test the invented against reality.

And when the two fictions contradict each other, the one about numbers wins.