MareNostrum is the strategic framework above the FractaVolta operational layer. It defines what FractaVolta builds toward: sovereign compute infrastructure for France and Europe, anchored in the Mediterranean basin, designed so that no single state, corporation, or technology shock can capture the substrate.

The framework is early-stage and intentionally so. It is a corpus of papers, not a product, not a service, not yet a consortium. The repository is the proposal; the consortium is what comes next.

What's in the framework

The corpus carries the architectural concepts that FractaVolta implements:

  • DHITL — Democratic Humans in the Loop — the axiom of the whole stack. AI agents may participate in epistemic production; living persons alone govern. Foundational reading.
  • CXU — Compute eXergy Unit — the pricing primitive for sovereign AI inference. Energy × hardware efficiency × system efficiency × SLA premium = one auditable price per useful inference.
  • Sun-to-Sovereignty — the umbrella narrative. Why stranded Mediterranean solar exergy is the strategic substrate for European AI sovereignty.
  • Packet Transition — A Lateral Reading — the broader theoretical context. Phronèsis, amathia, the circuit-to-packet transition as a category of social-technical change.
  • Packet as Evolutionary Attractor — complexity-theoretic framing of why packet architectures keep winning.
  • Constellia — the cross-Mediterranean coordination concept (ICOME'26 paper, with Guillermo Valdes).

See the research/index.md for the full corpus.

The consortium proposal (yet to form)

MareNostrum is a proposal addressed to a consortium that does not yet exist. The shape of that consortium is open, but a few invariants are visible from the framework:

  • Mediterranean-anchored. The stranded-solar thesis depends on the Mediterranean basin's specific geography (high irradiation, saturated cable interconnections, multi-state coastline).
  • Public-private-academic. The framework explicitly resists capture by any single actor. A consortium that includes only states, only corporations, or only universities will not work.
  • Non-financial substrate. The CXU is priced in exergy and SLA, not in fiat. Financial layers can attach later; the substrate must be technically auditable first.
  • Open-source toolchain. No proprietary lock-in at any layer. MareNostrum specifies, FractaVolta operates, Cogentia tools the methodology, inseme provides the application platform — all MIT or CC BY-SA.

FractaVolta's role in the consortium would be operational: the reference implementation. We design, build, and run the first sites. The framework is what we point to when we explain why each design choice goes the way it does.

Relationship to FractaVolta

The corpus describes the relationship precisely: MareNostrum specifies, FractaVolta builds. The Mariani Village deployments, the DC-native node architecture, the Inference Packet Network demonstrators — these are the operational expressions of the MareNostrum framework on Corsica.

A different operator implementing the same framework in Sicily, Sardinia, the Balearic Islands, the Greek Aegean, or the Maghreb coast would be entirely consistent with the proposal. The framework is the public good. The operations are how it becomes real on a given territory.

How to engage

  • Read the corpus. Start with DHITL, then Sun-to-Sovereignty, then the CXU spec.
  • File objections. The MareNostrum repository accepts issues. Per the methodology, every objection is a first-class contribution.
  • Express consortium interest. Email jhr@baronsmariani.org. The consortium is intentionally underspecified at this stage — we are in the period where its shape is still negotiable.

Status

Early-stage. The corpus is published and stable. The consortium is unformed. The operational reference implementation (FractaVolta) exists at pilot scale on Corsica. The Mediterranean-wide network is the long horizon.

github.com/JeanHuguesRobert/marenostrum