Severability layering

How the instrument decomposes when institutions start cutting, and what each cut costs. This document is strategy, not law: it binds the drafters and the campaign, and the layer-fidelity gate reviews the articles against it. It exists because the legal-basis gate's standing finding is real: the complete instrument's centre of gravity is contestable, and the honest response is to know in advance which parts stand on which legal basis, and what the ask becomes at each stage of trimming.

The four layers

Layer 0: the question (survives everything)

The ask that survives any trimming, including partial registration of a European Citizens' Initiative under the settled case law allowing it:

The Commission is asked to assess, and to propose instruments for, the participation of citizens of the Union in the productivity gains of hyper-automated production, including designation criteria for undertakings whose output is substantially decoupled from employment, and mechanisms by which the gains of such production become broadly owned.

Nothing in Layer 0 requires any particular legal basis, because it asks for an assessment and a proposal, which is what an initiative may ask. Registration of Layer 0 is the Gate 1 floor: if the Commission will not register even this, the project's premise about the legal channel fails and the kill criterion in campaign/GATES.md applies.

Layer 1: designation and transparency (Article 114 TFEU, comfortable)

Articles 1 to 4, 14, 16, 18, with Annex I; penalties limited to the notification and information duties; recitals 1 to 7, 24 to 26, 28 to 30.

A regulation that designates hyper-automated undertakings on harmonised criteria, makes the designation public, and reports on adoption, employment effects and ownership concentration. It harmonises exactly the thing Member States are starting to do divergently, on the DMA's own architecture, and it carries the falsification condition. No warrant, no Reserve, no entitlement.

This layer is defensible under Article 114 on the DMA precedent with no novel doctrine. It is also, standing alone, worth having: designation plus the Article 14 evidence base is what every subsequent instrument, Union or national, would build on.

Layer 2: the warrant (Article 114 TFEU, contested; fallback 352)

Articles 5 to 7, 13 in full, 15, 17, with the company-law derogations and the valuation machinery; recitals 8 to 15, 23.

The obligation to issue the citizens' capital warrant, crystallising on the first liquidity event, with the BRRD-pattern safeguards. The Article 114 case: it removes the incentive for divergent national levies and participation schemes (recital 1), harmonises the company-law derogations the subscription needs, and secures the level playing field between Union and third-country undertakings. The attack (legal-basis gate, first run): the centre of gravity is redistribution, not market-building, and Article 114(2) excludes fiscal provisions; the controlling line is Tobacco Advertising (C-376/98), which demands genuine obstacles or appreciable distortions, not a desirable policy. The defence rests on the non-fiscal structure being real (Article 8's insulation, settlement in shares only), not asserted, and on recital 1's divergence record being genuine.

If the characterisation fails, this layer moves to Article 352: unanimity in Council, consent of the Parliament. That is a political mountain, and this document says so rather than pretending otherwise.

Layer 3: the Reserve and the entitlement (Article 352 TFEU, honestly)

Articles 8 to 12, with Annex II; recitals 16 to 22, 27.

The Reserve as a new Union-level body holding assets for citizens, the universal entitlement arising by operation of law, national vehicles, the raid-proofing. Creating a new body with legal personality and a direct Union-to-citizen property relationship is where Article 114 is weakest and where the subsidiarity objection concentrates. The body-creation precedents cut both ways: Article 114 has sustained Union bodies that serve harmonisation (ENISA, C-217/04; ESMA's intervention powers, C-270/12), but a new legal form standing apart from national laws required what is now Article 352 (the European Cooperative Society, C-436/03), and a Reserve owing citizens a direct property relationship resembles the second more than the first. The honest position: Layer 3 is drafted to be defensible under Article 352, and its unanimity requirement is priced in. The campaign's answer to "you will never get unanimity" is the wealth-tax precedent: the fight is meant for the Council, in public, on the record.

Decomposition rules

1. Every layer must function with the layers above it struck. Layer 1 without Layers 2 and 3 is a designation-and-evidence regulation. Layers 1 and 2 without Layer 3 would park crystallised shares with an independent depositary pending a distribution instrument; that escrow amendment is not in the draft, because in the draft the Reserve exists, but it is the prepared answer if Layer 3 is severed in legislative procedure, and no article may be drafted in a way that forecloses it.
2. No article may create a dependency downward on a higher-numbered layer's existence, other than Layer 2's delivery of shares to the Reserve, for which point 1's escrow is the severance answer.
3. The memorandum must argue each layer's basis separately. A single centre-of-gravity argument for the whole instrument concedes the weakest layer's characterisation to the strongest layer's opponents.
4. At Council stage, concessions run top down: Layer 3 is conceded to a separate instrument before Layer 2 is weakened, and Layer 2 is conceded to Article 352 procedure before Layer 1 is abandoned. Nothing in Layer 1 is conceded; it is the floor above Layer 0.
5. The ECI registration ask is Layer 0 as primary request with the full Regulation annexed as the suggested instrument, so that partial registration, if the Commission insists on it, trims the annex and not the ask.

What this costs and why it is worth it

Layering is a concession in advance, and hostile readers will quote it as lack of confidence. The alternative is a monolith that dies whole at its weakest point, which is how most redistributive proposals at Union level have died. The book's own analysis is that the mechanism matters more than the vehicle: designation and evidence (Layer 1) create the factual record; the warrant (Layer 2) is the mechanism; the Reserve (Layer 3) is one distribution architecture among several possible. Losing Layer 3 to an intergovernmental or successor instrument loses elegance, not the point.