Funding & Impact
Assessing this work for funding usually means answering a narrower question than assessing it for citation: not only is this rigorous, but does it matter enough, to enough people, soon enough, to be worth backing — the institutional expression of what this research programme calls Navigational Coherence applied to its own case for support.
The Stakes, Stated Plainly
Governance and AI-safety funding increasingly targets disclosure, transparency, and alignment mechanisms. Those matter. But disclosure and alignment both assume something upstream of them: that the reasoning behind a decision remains recoverable at all, by anyone, once the decision has been made.
This programme's claim is that the assumption is already failing, independently of any single AI system's behavior — and that the gap is structural, not merely procedural.
What the Loss Costs, and What Recovery Would Give Back
Understanding that has to be rebuilt instead of inherited costs more than the effort of rebuilding. It costs everything that effort could have built instead. That unattained gain is the other half of the loss this programme describes, and the half least often counted. Recovering part of it is a gain in itself, and it compounds once it is kept.
This is an argued model, not a measured result. Testing it is where the work would begin: counting what rebuilding costs, and comparing like tasks with and without continuity support.
The Research Behind the Claim
The claim is argued, not merely asserted, across two anchor papers:
- Legible Risk: Why Disclosure Isn't Enough — disclosure and legibility are shown to be different properties; a disclosed risk that cannot be traced to a stable object or a responsible party remains invisible in practice.
- Super-Optimization: When Optimization Outpaces the Capacity for Double-Loop Revision — the underlying mechanism does not require artificial intelligence at all; it is already visible in corporations and bureaucracies, and AI accelerates it rather than originating it.
These sit inside a wider, sustained research programme — over twenty further papers developing adjacent claims (Interpretive Entropy, the Continuity Threshold, Recoverability, and related mechanisms), not a single pitch built around one idea.
Policy Relevance
The EU AI Act is a serious attempt to govern this problem through transparency, oversight, and accountability requirements. This programme's position is not opposed to that effort — it names the gap regulation cannot close on its own, because part of the problem is structural and navigational rather than procedural.
→ How the EU AI Act maps onto this work — structural alignment, not legal advice.
Not a Concept Awaiting Implementation
The architecture this programme argues for is not a proposal waiting to be built. Over roughly a year, it was developed and refined into a working, multi-user system: agent-agnostic by design, with its own execution, audit, and state-management layers, and arranged across distinct deployment contexts — personal, professional, educational, and institutional — under a shared custodial core.
At that point the open question was no longer whether the architecture worked. It was whether to release it immediately or to first build the governance structure capable of holding it responsibly — the same custodianship question addressed below. The second path was chosen. That has taken a further year.
What funding would support now is not invention. It is the empirical validation this kind of architecture needs before wider claims about it can be responsibly made, and completion of the custodianship structure that keeps it from being released without one.
Where This Stands
This is independent research. The intention is for the architecture's governance — decisions about how the shared infrastructure is stewarded, and by whom — to sit with a nonprofit structure, kept separate from any commercial instantiation built on top of it, so that custodianship of the architecture is not captured by growth or commercial incentives. That structure does not yet formally exist; the intention is stated here plainly, as intent, so a funder can see the direction rather than infer it.
→ About — for who is behind this work and how it is currently held.
What Support Would Actually Fund
Concretely: continued independent research (the papers above are a programme, not a finished output), and development of the reasoning infrastructure the research argues for — structured reasoning artifacts, decision lineage, and the architectural mechanisms described under Architecture.