On Thursday, 9 July 2026, the Batavia Biosciences building in Leiden was evacuated over a smell of smoke — a ventilation motor. Eleven days later, Het Financieele Dagblad published the analysis that exposes the pattern: Batavia, in Leiden since 2009 and acquired in 2021 for €200 million by South Korea's CJ CheilJedang, is shutting down worldwide operations. All 150 employees made redundant. The new 12,000-square-metre building, which cost over €100 million, stands empty. A bee hotel out front. Silence inside.

This is not an accident. This is the pattern. Galapagos, once worth €15 billion on the stock market, collapsed and retreated to Belgium under another name. Halix, the AstraZeneca vaccine producer for which Boris Johnson once considered a military intervention, now sits in total stillness: an empty car park, a €10 million loss, idle machinery. Pharming cut 20% of its jobs. ProQR under pressure. 5 of the 6 major Dutch biotech firms are in sale mode or in flight — while the Americans keep building: Bristol Myers Squibb, Johnson & Johnson, and Eli Lilly with a €3 billion plant in Katwijk.

This is the third article in a series of three. First the confession. Then the outlook. Now the learning function we never had — and the proof that without it, the crash is not years away. It has already begun.

Foreword — Third in the series: the clock is ticking

Three articles, one architecture. Honesty Makes Government Possible brought the confession: the PBW formula showed that €419 billion of the Dutch GDP is not productive, concealed by accounting fictions. Deciding Without a View brought the outlook: laws from the 1960s and 1990s freeze the system without an expiry date. This third article adds the learning function — the mechanism that makes the first two foundations actually work.

And the clock is ticking faster than the first two articles could show. Where the openvizier projection from early 2026 forecast a NEPK crash through the 3% threshold between 2029 and 2033, the July 2026 recalculation on primary sources shows a NEPK of 2.95% — already below that threshold. 5 of the 6 major Dutch biotech companies are in sale mode. The crash date is not 2032. It is 2027-2028.

Part 0 — From GDP to NEPK: why GDP no longer tells us anything

The Dutch GDP keeps rising. Officially, things are going well. But GDP does not measure what the Netherlands structurally earns. It measures total spending, regardless of whether the profit stays here or flows to Indianapolis, Seoul or San Francisco. The measure that does capture this — the NEPK, the Net External Productive Core — stood at 4.2% of GDP at the start of 2026, according to openvizier.org. That was already the lowest of seven advanced economies studied. Singapore stands at 17%. Ireland at 11.9%. Germany at 7.8%.

Six months later, the Netherlands has already sunk deeper. Recalculation on the most recent primary sources — OECD Trade in Value Added 2023 (Export-VA 36.1%), World Bank industry value-added 2024 (α from 17.86% of GDP), OECD Revenue Statistics 2024 (τ = 38.5%) and CBS 85821NED 2023 (foreign control 25.6% of business value-added, φ = 0.744) — gives a NEPK for July 2026 of 2.95%. This is not inconsistent with the earlier projection. It is exactly what openvizier predicted, but with an acceleration of five to seven years. The Netherlands is already below the 3% threshold, beyond which fiscal sustainability is warned to become untenable.

What the figures don't yet show — 5 of the 6 in sale mode

Why is the crash outrunning the projection? Because CBS, OECD and the World Bank report with a delay of twelve to twenty-four months. The official figures for 2023 and 2024 capture deals from 2022 and 2023. What is moving through the economy today shows up in no statistic at all. And what is moving is this: five out of every six entrepreneurs who built something of substance are no longer investing. They are polishing the books for sale. That is doubly destructive for the NEPK: first α falls, because no new investment is made; then φ falls once the deal closes, because the buyers are American, Korean, German, Belgian.

Both movements stay invisible to statistics: the investment freeze is reported nowhere, and the sale appears in official figures only once CBS or the Bundesbank registers the change of ownership — months or years later. By the time the figures catch up with reality, the Netherlands will be deeper than the July calculation already shows. This is exactly why "waiting for the figures" is a political mistake: by the time the figures are ready, the situation is irreversible.

To understand the crash, the number must be dissected. The Dutch GDP for 2024 amounts to €1,128 billion. Within it sit three items that measure no production at all: €97 billion imputed rent (an accounting fiction in which homeowners pay themselves rent), €40 billion FISIM (fictional banking services statistically added to GDP), and €282 billion in the benefits paradox (government spending with no productive counterpart). Subtract these three items and €709 billion of real, production-linked prosperity remains — the PBW, 62.9% of official GDP.

From this PBW, overhead and compliance burden must then be deducted: α = 0.39 for the Netherlands, against 0.58 for Singapore. What remains is the productive core: €277 billion. Of that, 47% is foreign-owned (φ = 0.53) — profits flowing to foreign shareholders. What's left is €51 billion. Four point five percent. That is the Net External Productive Core of the Netherlands: what the country structurally earns from the world, after deducting accounting concealment, bureaucratic overhead and profit repatriation. This is not a pessimistic reading of favourable figures. This is the mathematics of openvizier.org, based on OECD Trade in Value Added, World Bank industry value-added and CBS National Accounts.

Part I — The four levers of the NEPK

The NEPK formula is mercilessly simple:

NEPK = E_tv × α × (1 − τ) × φ

Four levers. E_tv is export value-added as a percentage of GDP. α (alpha) is the productive core — the share of the economy that, after overhead and compliance, actually adds value. τ (tau) is the collective tax and levy burden. φ (phi) is the Dutch ownership share. On all four levers, the Netherlands is losing.

α — the productive core that's shrinking

Our α is 0.39. Of every euro of value added, 39 cents is actually productive; the rest disappears into overhead, compliance, consultants and intermediary services. Singapore: 0.58. Portugal: 0.48. The loss is literally in the FD quote from CEO Geert Jan Groeneveld: the LUMC professor with "really great ideas" who doesn't even start because he gets only 5% of the shares. His potential contribution — new medicines, exporting companies, highly skilled jobs — disappears into α. Never measured, but lost all the same.

τ — the tax burden that hollows out

Our τ is 0.43. Nearly half of every productively earned euro goes to collective levies. Singapore: 0.17. UAE: 0.14. Ireland: 0.19. Portugal: 0.33 — and Portugal also has a welfare state. The difference is that the Netherlands lets collective levies grow without a matching return in productive capacity. Every euro that goes to τ instead of α accelerates the crash.

φ — the ownership draining away

Our φ is 0.53 — 47% of Dutch productive capacity is foreign-owned. In Portugal: 27%. In Germany: 28%. Singapore has φ 0.33 — even lower than ours — but uses that to keep tax burdens low, without a social contract that needs financing from domestic productive profit. The Netherlands has that contract, and is losing the ownership needed to finance it at the same time.

Every time a Dutch biotech company is sold to a foreign owner, φ falls. Batavia to Korea in 2021. Galapagos back to Belgium in 2024. Argenx to Belgium years ago. And every American factory built on Dutch soil — Bristol Myers Squibb, Johnson & Johnson, Eli Lilly — raises E_tv and α, but not φ. Statistically, the economy appears to grow. The NEPK keeps falling.

E_tv — export that can mislead

Our E_tv is roughly 38%. Ireland: 55%. Singapore: 50%. That looks favourable, but Singapore has φ 0.33 — much of that export yields Singapore itself little. High export is only productively meaningful if you keep it in your own hands. Ireland combines high E_tv with low τ and low φ, and keeps some profit domestically through clever tax design — net NEPK 11.9%. The Netherlands does none of that.

Where the Netherlands stands — the ranking

Country	NEPK	α	τ	φ	E_tv
Netherlands	4.5%	0.39	0.43	0.53	38%
Germany	7.8%	0.46	0.40	0.72	39%
Sweden	9.1%	0.49	0.42	0.71	43%
Switzerland	13.4%	0.54	0.27	0.74	58%
Singapore	17.0%	0.58	0.17	0.33	50%*

*Singapore's E_tv is skewed by transshipment trade (179% raw export value-added); the 17% NEPK follows from a corrected calculation.

The Netherlands is below the 3.5% threshold which openvizier.org warns must be held above before 2029, or the implosion becomes irreversible. Under unchanged policy, the NEPK falls through the critical 3% threshold between 2029 and 2033. Add the AI revolution — which will largely automate away the knowledge-intensive services in which the Netherlands is relatively strong — and we fall to 2% before 2035. At that level, the country can no longer finance its own social contract.

This is not an accident. This is the pattern.

Part II — Three cases of NEPK destruction

Abstract figures call for concrete cases. Three files in which the four levers collapse simultaneously, and in which the Netherlands removes its own core without noticing. Each case shows the same pattern: a political decision made without the knowledge and skill required, with NEPK losses that are never structurally recovered.

Case 1 — Leiden biotech: the cluster in flight

The FD article of 20 July 2026 paints the picture every NEPK figure confirms. Leiden Bio Science Park was the largest life-sciences cluster in the Netherlands, top-5 in Europe. Today it is a cluster in flight.

Company	What	Where to	Effect on φ
Galapagos	Once worth €15bn; collapsed	Belgium (other name)	φ down — capital and patents gone
Batavia Biosciences	€200m acquisition 2021; shuts down globally; 150 redundant	South Korea → zero	φ down twice — sold, then liquidated
Halix	AstraZeneca vaccine producer; €10m loss; stalled	Repurposing considered	φ at risk — capacity unused
Pharming	20% of jobs cut at headquarters	Share price under pressure	φ shrinking — core activity hollowed out
ProQR	Under pressure; weak share price	Takeover possible	φ at risk
Argenx (historical)	Originally Dutch; now worth €30bn+	Ghent (BE) — years ago	φ lost — the biggest one there was

Meanwhile, building does continue — but not by Dutch owners. Bristol Myers Squibb built a brand-new facility for cancer cell therapies. Johnson & Johnson invested in a new building with white facade stones. Eli Lilly is building in Katwijk for €3 billion. Every investment raises E_tv and α on Dutch soil. Every investment carries φ = 0, because the profit goes to Indianapolis and New York.

Why this happens: the 5% rule and the lack of growth capital

Two reasons stand out. First: the 5% shareholding rule. A LUMC professor who wants to commercialise their own discovery is typically allowed, under the Dutch Knowledge Exploitation Regulation, to own a maximum of 5-10% of the shares. The patent is registered to the knowledge institution; the university places its shares in its Holding BV; investors receive shares in exchange for capital; the researcher keeps the rest. Under these conditions, he "just doesn't start," in Groeneveld's words. In Belgium — to which Argenx relocated and where it is now worth over €30 billion — the scientist-entrepreneur keeps more. In Denmark (Novo Nordisk, Genmab). In Germany (BioNTech). Everywhere better than here.

Second: the lack of Dutch growth capital. Leyden Labs, one of the few genuinely thriving Leiden biotech firms, raised €40 million in June 2026 — but with capital from the Gates Foundation, Singapore's ClavystBio, the EIC and America's Fidelity. Invest-NL contributed €10 million. The rest is foreign. When Leyden Labs succeeds, the proceeds go to the largest shareholders — back abroad. φ down, even in success.

The hidden solution — what Batavia could have become

Batavia wanted to produce vaccines and biological medicines as a European counterweight to China. When that failed commercially in the bulk vaccine market, the company was declared redundant. But the infrastructure — 12,000 m² of clean space, bioreactors, cleanrooms, GMP licences, technically trained staff — has not disappeared. It stands empty, waiting for a use no one has named because no one sees the connection.

In the World Belt corridor file — covered in Deciding Without a View — it is shown how tropical corridors with Juncao grass and Moringa trees can produce millions of tonnes of biomass annually. Through multi-membrane hydrolysate technology, this biomass is converted into processable building blocks: proteins, carbohydrates, cellulose derivatives, plant sterols. This technology fits exactly onto the bioreactor infrastructure Batavia built. Bioreactor separation processes for viral vaccines are technically and legally the same category as membrane separation processes for plant hydrolysates. GMP cleanrooms work for both. The staff who could make vaccines can make hydrolysates.

That is the zero-base question in action: if we started Batavia over today, with all of today's knowledge — what should it have become? Not an Asian subsidiary losing in the bulk vaccine market, but a Dutch-anchored node in the World Belt hydrolysate chain. Raw material from the Mauritanian GMV or the Andes-Amazon corridor. Processing in Leiden. Downstream distribution to European pharma and food. Dutch ownership (φ up). High margin (α up). Export (E_tv up). Low tax burden if smartly designed fiscally (τ contained). All four NEPK levers moving in the right direction.

But no one has built that bridge. Not in the House of Representatives, where the Agriculture and Economic Affairs committees don't talk to each other. Not in the Provincial Council. Not at Invest-NL. Not in the Ministry of Economic Affairs. It exists at openvizier.org, where the files sit side by side — but that is a one-man newspaper from Palma.

This is the Pinocchio scene in miniature. We tell ourselves: "the market is simply bad" (literal headline from 9 July). We say: "market correction after the pandemic" (HAL Allergy). We say: "we were overtaken by reality" (Marty Stork, Batavia's HR director). Every time, the nose grows. The reality is that we had a solution nobody connects.

Case 2 — The plastics waste law: one rule, one million tonnes of CO₂

The Netherlands has had a good law since the 1990s: waste may not be landfilled. Excellent policy — it prevents wild waste dumps, protects soil and groundwater. The only question is: what counts as waste, and what no longer does? In the 1990s, the Netherlands made one crucial change. Biomass was taken out of the strict waste regime. Organic kitchen and garden waste got its own mandatory collection from 1 January 1994, fell under the BOOM decree, and has since been treated as raw material for compost and biogas. Since then, 1.5 million tonnes of organic waste plus 1.5 million tonnes of green waste per year go to high-grade processing instead of the incinerator — one of the most successful pieces of circular policy the Netherlands has ever had.

Plastic still sits in the old regime. It falls under "household residual waste" and "industrial residual waste" in the Landfill Ban Decree. Landfilling is banned. Recycling is fine but insufficiently covers the volume. What remains must be incinerated in waste-to-energy plants — that produces 2.64-2.86 kg of fossil CO₂ per kg of plastic. In the Netherlands that's 1.32 megatonnes of CO₂ per year. Across the whole EU: 71.5 megatonnes per year, or 2.4% of total EU emissions — for carbon that could have been safely stored underground.

The EU Carbon Removals Certification Framework (CRCF, 2024) recognises permanent carbon removal as a policy goal. Underground storage of stable plastic — in old salt mines, in the Hambach pit, in certified geological repositories — is by definition permanent carbon removal. Yet it is not allowed, because plastic still sits in the old waste regime.

One rule change. Just as biomass was moved in 1994 from "burn as residual waste" to "composting and biogas," plastic must be moved from "burn as residual waste" to "certified underground storage under CRCF." Fiscal reversal for the Netherlands: from €334 million levy + €99 million ETS charge = €433 million annual reversal. At EU scale: €5.4 billion/year. NEPK effect: positive — a new bio-based plastics industry (Bio-PE, Bio-PP, PEF) with Dutch ownership (φ up), high margin (α up), European export (E_tv up).

Why doesn't it happen? Because no committee member in the House of Representatives holds specialist knowledge in chemistry, membrane technology, or waste law. There is someone with a political background, an environment spokesperson, a parliamentary group leader. Incompetence guards the status quo. The nose grows, the CO₂ stays up, the NEPK falls.

Case 3 — Eli Lilly Katwijk: incompetence rejecting investment

On 9 July 2026, the Provincial Council of South Holland voted 47 to 5 in favour of €31 million in traffic measures around a new Eli Lilly plant in Katwijk, on condition that the plant would not become a Seveso establishment. The motion: "We instruct the Provincial Executive to do everything possible to prevent such a company from establishing itself here." The trigger: concern over hazardous substances, traffic pressure, drinking water shortages, effects on the power grid, and the Valkenhorst housing development.

All the concerns are real. The question is whether the average council member has the knowledge to assess a Seveso classification. A Seveso establishment is a chemical production site with defined quantities of hazardous substances above an EU threshold — that depends on precise substance inventories, not on "we're building a big factory." Provincial Executive member Meindert Stolk noted: "Not a single Seveso company sits anywhere in the whole Leiden Bio Science Park. Nobody is waiting for hazardous situations." The Council still voted 47-5 against.

This is the FD diagnosis in political form. An investment of €3 billion, 500 highly skilled jobs, expansion of an internationally recognised science park — hangs on a political vote whose majority cannot assess the technical content. Sinan Özkaya (PRO): "As a Council, we should have been able to find something about the consequences for the power grid, drinking water and the environment." Fair enough. But who on the Council actually can assess that? And if no one can, why are we making the decision?

And even if Eli Lilly comes: 100% American ownership. NEPK effect: α and E_tv up, φ unchanged. The Netherlands supplies land, water, power, traffic infrastructure (€31 million for Katwijk alone), and gets production in return whose profit leaves for Indianapolis. If the plant ultimately doesn't happen due to political disorder, we also lose the α growth. Two scenarios, both inferior to the variant in which a Dutch biotech company had made the same investment with the same land and the same permit.

Part III — Zero-base: the learning function we don't have

Three cases, the same pattern three times over: the knowledge to solve the problem exists, but political decision-making fails to incorporate it. The answer does not lie in more analysis, more committees, more policy papers. It lies in a structural mechanism that lets legislation learn. That mechanism is called zero-base with 1/5 convergence (the learning rule) per year per file.

Everything that learns reaches its optimum within years — except Dutch legislation

Everything that learns reaches its optimum within years — except Dutch legislation.

Every system that produces something genuinely good learns. Neural networks have done so since 1986. Children do it from birth. China does it through its Five-Year Plans. Craftspeople through master-apprentice traditions. The Netherlands and Europe do not — the average fundamental law has a revision cycle of 15 to 30 years. An order of magnitude slower than any learning system we know.

The method: if we started today from scratch

Zero-base works like this. Each year, for every law or policy file, we ask one question: "If we started completely from scratch today, with all current knowledge and technology, what would the law look like?" That is fundamentally different from "how do we adjust the existing law?" Adjusting means stacking on a foundation you no longer question. Zero-base questions the foundation itself.

Then compare that imagined zero-base with the existing law. There is a gap — sometimes small, sometimes enormous. For the plastics waste law, the gap is enormous (changing one rule solves 1.32 megatonnes of CO₂). For the 5% shareholding rule, the gap is enormous (Belgium proves it). For the Seveso classification procedure, the gap is enormous (technical support for council members is missing).

The 1/5 rule — why exactly one fifth

And then comes the 1/5 rule. The government bridges not the whole gap in the coming year, but exactly one fifth. Why exactly a fifth? Because that is the learning rule that shows up everywhere something truly learns.

In neural networks this is called gradient descent. With too large a step, the system oscillates and explodes. With too small a step, you never reach the goal in reasonable time. The optimal step lies empirically between 1/5 and 1/10 per iteration. Adam, the most widely used AI optimizer, typically uses learning rates between 0.1 and 0.2 — precisely 1/10 to 1/5.

Knowledge comes from making mistakes and recovering

Knowledge comes from making mistakes and recovering.

The AI side is the mathematical side. The human side matters at least as much. Every human who learns something — a child learning to walk, a craftsperson learning to build — learns through mistakes of manageable size. Exactly a 1/5 error is the golden rule: small enough to recover from without catastrophe, large enough to actually learn something.

Why not smaller? Because a smaller error provides no contrast — you don't notice what you did wrong. Why not larger? Because a larger error breaks the learner itself. If a pension system loses half its coverage in a legal revision, pension funds have no buffer left to bounce back. Zero-base with 1/5 convergence is therefore not an ideological choice — it is a learning rule that comes from mathematics and from people themselves.

Every treatment case has its own annual cycle

Every treatment case has its own annual cycle.

Zero-base does not work per country as a whole — that would again be a centralist committee. Each file gets its own annual zero-base cycle, with its own dashboard, its own gap calculation and its own 1/5 step. Exactly as each neuron in a network has its own weights and follows its own gradient.

For the three cases in Part II, this means concretely:

• Leiden biotech cluster — year 1 (2027): the 5% shareholding rule for academic spin-offs is revised; researchers get 25% shares as a starting point (1/5 of the way toward the Belgian condition). Invest-NL gets a biotech mandate with a Dutch-ownership clause (φ-lock). The Batavia plant is designated a hydrolysate conversion platform with World Belt supply.
• Plastics waste law — year 1 (2027): 20% of non-recycled plastic is classified as "safely stored" if it meets CRCF criteria, with a partial discount on the levy and half a carbon credit. In 2028: 40%, in 2031: 100%.
• Seveso classification — year 1 (2027): Provincial Councils get a mandatory technical expert panel for Seveso permits. Council members receive binding advice before the vote. In 2028, expansion to environmental impact assessments; in 2029 to drinking water effects; in 2030 to a full technical advisory regime.

Part IV — High-knowledge teams and a constitutional qualification requirement

The usual answer to new policy is: "we'll set up a committee." That answer is exactly the problem. Committees delay, dilute and mute. They are designed to spread responsibility and avoid risk. A learning system needs the opposite: high-knowledge people at the helm, with matching compensation, who dare take risks because they know what they are doing.

What is a high-knowledge team

Per file, one small team of three to five people, each with proven expertise. For the biotech file: one biotech entrepreneur with exit experience, one VC investor with a life-sciences portfolio, one professor with a spin-off history, one lawyer with CAO-NU/Knowledge Exploitation Regulation expertise. For the plastics file: one chemist with CCS experience, one economist with ETS models, one practitioner from the recycling sector, one lawyer versed in EU directives. Each team is responsible for one file, performs one zero-base recalculation each year, publicly publishes the gap calculation and the 1/5 step, and works with a dashboard that is online 24/7.

Why industry-level pay

A biotech entrepreneur with exit experience earns €300,000 to €500,000 per year plus equity in industry. A VC partner in life sciences, comparably. A professor with a spin-off history often combines a university salary with advisory income of €150,000-€250,000. If the state wants to attract these people for structural work, it must offer market-rate salaries. Otherwise it gets second-rate people making second-rate policy.

The Netherlands pays its ministers €183,000 per year. Singapore pays its ministers €1.3 to €1.9 million per year, tied to the top-6 sectors. The result is visible: Singapore has NEPK 17%, the Netherlands 4.5%. The New York Federal Reserve Bank pays its president \$479,000 per year, 2.5 times a federal minister's salary. The reason is well known: monetary policy demands high knowledge, and high knowledge costs money.

Proposal: base salary of €250,000 per team member per year, plus a performance bonus of €100,000 if the file measurably converges according to the dashboard. Four team members per file, twenty critical files: total €28 million per year. By comparison: the fiscal reversal on the plastics file alone yields the Netherlands €433 million per year — the payback time for the entire system is less than one month on a single file.

Constitutional qualification requirement for committees

And further, a constitutional article that touches political selection itself. The House of Representatives selects members on political background, not subject-matter knowledge. Whoever sits on a parliamentary committee deciding on healthcare need not be a medical professional. Whoever decides on energy need not be an engineer. Whoever decides on defence need not have served. This is by now demonstrably the source of NEPK loss — decisions are made without the knowledge to assess them.

Proposed constitutional article: Every standing parliamentary committee shall include at least two members with demonstrable subject-matter qualification in the file's domain: a medical professional on the Health Committee, an entrepreneur or engineer with ten-plus years of business experience on the Economic Affairs Committee, an agricultural entrepreneur on the Agriculture Committee, a doctorally qualified scientist on the Education and Science Committee. These members hold voting rights like all others; their qualification is a condition for committee membership.

This is undemocratic, I hear you say. The answer: it is precisely democratic — it strengthens the quality of democratic decision-making by guaranteeing that every committee contains at least two votes able to grasp the technical substance. Other members may make political judgements, provided those judgements demonstrably account for what the subject-matter members contribute. Whoever votes without substance breaks the protocol.

Pinocchio in person — Groningen, concrete and the prime minister

On 10 July 2026, the prime minister said after the cabinet meeting:

"One of this cabinet's most important missions is to make the Netherlands the strongest economy in Europe, with the best investment climate."

On the same day, his cabinet reaffirmed that the Groningen gas field remains permanently closed. Against all opposing votes, against Shell's arbitration claim, against the fact that in 2024 not a single exploratory well was drilled in the Netherlands — the first time in eighty years. And even as he says this, concrete is being poured into the 300 boreholes. Seventy holes already sealed as of early 2026; all 300 by 2034.

This is Pinocchio in person. The prime minister who promises the strongest economy in Europe while simultaneously having the foundation of 60 years of Dutch prosperity — €363 billion in natural gas revenues — sealed shut. Not because closing the field was wrong; closing it was correct. But because the alternative has already been worked out and is being ignored.

Honesty makes governing possible lays out the circular deal in detail: at a TTF price of €37/MWh, the remaining 400-470 billion m³ of gas would yield €172 billion gross. Of that, €120 billion for affected households (differentiated by damage zone, €400k average). €15-20 billion for the reclamation of Almere Pampus / IJmeerpolder / possibly the Markerwaard — 40 to 60 thousand new homes. Groningen residents would receive five years of first-right-of-purchase with a sunset clause. €50 billion for the state, 60-80 thousand construction jobs for ten to fifteen years, a €25-35 billion GDP effect, €14 billion in tax returns, and earthquake damage properly compensated instead of through an IMG compensation body that spends 78 cents of administrative cost per euro of compensation paid out.

That is the deal the prime minister does not mention. He talks of "the strongest economy in Europe" while simultaneously having 15-18 years of Dutch gas consumption sealed underground. He talks of "the best investment climate" while letting NAM pay out €3 billion in dividends to Shell and ExxonMobil while Groningen families wait for compensation. He promises growth through "talent and innovation" while letting the skilled subsurface-development sector be permanently liquidated — zero exploratory wells in 2024, the first time in eighty years.

The wooden nose is poured with concrete into the hole. The farmhouse in the background stands crooked from earthquakes that his cabinet does not properly compensate. The affected residents in the background watch in silence. And the prime minister smiles — because in the party-political ledger, his accounting works out: the climate narrative is saved, the Groningen voter base is too small to cause electoral pain, and the promise of "the strongest economy in Europe" is a frame nobody in parliament dares to refute with NEPK figures. All four levers — E_tv, α, τ, φ — are falling at once. He does not see it, he does not measure it, and he is the first prime minister in Dutch history under whose government an NEPK of 2.95% was recorded.

And this is the point of this scene: the Pinocchio metaphor is not an abstraction. It is not "politics in general" that lies. It is a specific prime minister who, on a specific date — 10 July 2026 — spoke a specific lie, while the alternative was published in a specific article on openvizier.org that he does not read. The wooden nose is not metaphorical. The boreholes are not metaphorical. The concrete is not metaphorical. Only the fairy who arrives at the end to save Pinocchio — that is a fairy tale. No fairy is coming.

Every lie grows the nose. Every denied solution shortens the time until the NEPK falls below 3%.

And unlike Pinocchio, no good fairy arrives here to truly transform the wooden head at the end. No rescue comes from outside. Our political selection based on party background rather than competence is our own responsibility. Our laws from the 1960s, 1970s and 1990s that strangle the sector are our own laws. Our companies that get sold to Korea are our own companies, with our own 5% rules. Our productive core is falling because of our own choices. In the end, we have nothing left. Not because the world did this to us. Because we did it to ourselves by failing to learn.

Conclusion — The Pinocchio nose that never stops growing

Three articles. One architecture. Confession, outlook, learning function.

Confession (article 1, Honesty Makes Government Possible) — the PBW formula that makes visible that €419 billion of Dutch GDP is not productive. Without an honest dashboard, every policy is symptom management.

Outlook (article 2, Deciding Without a View) — the sunset clause that lets laws expire. Without an expiry date, laws pile up like fossils in sediment, and the system freezes.

Learning function (this article) — the zero-base method with 1/5 per year per treatment case, executed by high-knowledge teams with industry-level pay, with a constitutional qualification requirement for parliamentary committees. Without a learning cycle, sunset merely replaces an old law with a similar new one — and nothing is gained.

The three together form a democracy that can learn the way a child, a violinist, a neural network or a Chinese planning commission can learn. And that is exactly what we need now, because the core of what the Netherlands as a country earns — the NEPK — is falling toward 2%.

The Pinocchio metaphor is not incidental

In Collodi's fairy tale, the wooden puppet eventually becomes a real boy. His nose stops growing when he stops lying to himself. In Dutch politics of 2026, the opposite happens. We tell ourselves:

• "The market is simply bad" — quote from the FD headline about Batavia. The nose grows — because the market isn't bad; we failed to connect to the new market.
• "We welcome foreign investors" — cabinet line, July 2026. The nose grows — because every instance of foreign ownership lowers φ and thus the NEPK.
• "We have a great science park" — Wennink Report 2025. The nose grows — because the park has no flagship left, and 60% of the entrepreneurs are preparing their companies for sale.
• "We have good laws" — general opinion. The nose grows — because laws from 1994 (biomass out of waste) were good; laws from 2024 are petrified and punish exactly what we need.
• "We'll fix it with a new regulator" — the Dutch answer to the new EU procurement regulation of 19 July 2026 is a 'National Coordinating Authority'. The nose grows — because a new authority does not solve the NEPK problem; it raises τ (compliance costs) and lowers α (productive time disappears into publication duties for every change above €10,000). Even the Tax Authority admits: "in hindsight, many governments realise they overshot with market thinking." Precisely. But a new regulator is not the answer to overshot market thinking; it is more market thinking in bureaucratic wrapping.

Every lie grows the nose. Every denied solution shortens the time.

And unlike Pinocchio, there is no good fairy here to make the wooden head real in the end. No rescue arrives from outside. Our political selection by party background rather than skill is our own responsibility. Our laws from the sixties, seventies and nineties that strangle the sector are our own laws. Our companies sold to Korea are our own companies, with our own 5% rules. Our productive core is falling because of our own choices.

In the end we have nothing left. Not because the world does this to us. Because we did it to ourselves, by failing to learn.

This does not have to become a fairy tale with a bad ending. But then we must stop lying to ourselves — and start learning.

Appendix — Mathematics and formulas

The NEPK formula, worked out

NEPK = Export-VA × α × (1 − τ) × φ, with the four factors:

• Export-VA = export value-added as % of GDP (source: OECD Trade in Value Added)
• α = productive core share (source: World Bank industry value-added as % GDP)
• τ = total tax + social contributions as % of GDP (source: OECD Revenue Statistics)
• φ = 1 − foreign shareholding percentage (source: Bundesbank, Eurostat, FSS, JPX)

Netherlands 2025: Export-VA ≈ 38%, α = 0.39, τ = 0.43, φ = 0.53. Calculation: 0.38 × 0.39 × 0.57 × 0.53 = 0.0448 = 4.5%.

Singapore 2025: Export-VA ≈ 179% (transshipment!), α = 0.58, τ = 0.17, φ = 0.33. Effective calculation with correction: NEPK 17.0%.

The 1/5 convergence formula

Formula: p(t+1) = p(t) + (100 − p(t)) × 1/n, with n = 5. After k years, the system reaches 1 − (0.8)^k of the gap.

• After 1 year: 20.0%
• After 3 years: 48.8%
• After 5 years: 67.2%
• After 10 years: 89.3%
• After 15 years: 96.5%

The Adam optimizer (Kingma & Ba 2014) uses a standard learning rate between 0.001 and 0.1 per iteration — for annual policy cycles, 1/5 is the equivalent step size.

NEPK effects of the three cases — indicative

Case	Without intervention	With zero-base 1/5	NEPK effect after 5 years
Biotech cluster	Further exodus; 3-4 more major closures	φ-lock, 25% rule, Batavia hydrolysate	+0.3 to +0.5% NEPK
Plastics waste law	€334-500m levy/year, 1.32 Mt CO₂	CRCF differentiation, bio-based industry	+0.1 to +0.2% NEPK
Seveso/investment politics	Arbitrary rejection of €3bn+ investments	Technical expert panel; qualification requirement for council members	+0.2 to +0.3% NEPK
Total (three files, after 5 years)	NEPK falls to 3-3.5%	NEPK stabilises at 4.5-5%	+0.6 to +1.0% NEPK

And that's only three files. Twenty high-knowledge teams for twenty critical files — plastics, biotech, RED III, the Habitats Directive, the Nuclear Energy Act, the Spatial Planning Act, GDP methodology, and thirteen others — could together reverse the trend. From 4.5% declining to 6-7% rising. That is not Singapore (17%), but it is back above the critical 3.5% threshold and thus liveable.

Key findings

• The Dutch NEPK stands at 2.95% in July 2026, already below the critical 3% threshold — 5 to 7 years faster than the early-2026 projection.
• 5 of the 6 major Dutch biotech companies in Leiden are in sale mode or in flight; Batavia Biosciences shut down worldwide on 9 July 2026, with 150 employees made redundant.
• Three concrete cases — biotech, the plastics waste law, Eli Lilly Katwijk — show how incompetence in parliamentary decision-making undermines all four NEPK levers (E_tv, α, τ, φ) at once.
• The solution is a learning function: zero-base with 1/5 convergence per year per file, executed by high-knowledge teams with market-rate pay (€250,000 base salary) and a constitutional qualification requirement for parliamentary committees.
• Without this learning function, sunset remains empty of substance — an old law gets replaced by a similar new one, and the NEPK keeps falling toward 2%.