Research

PKOS explores infrastructure capable of preserving reasoning continuity in environments where human–AI collaboration accelerates knowledge production.

This research program investigates Navigational Coherence — how reasoning can remain visible, accountable, and extendable across time under conditions of acceleration.

Rather than presenting a finished theory, this section provides structured entry points into an evolving system of concepts, experiments, and publications.

Continuity Under Acceleration

Across domains, systems are experiencing increasing difficulty maintaining continuity of reasoning as knowledge production accelerates.

Historically, systems such as law, education, and science have relied on the accumulation of reasoning across time. Under current conditions, this continuity is increasingly strained.

The Continuity Threshold describes the point at which systems shift from cumulative understanding to increasing dependence on reconstruction. The System Recovery Principle describes how systems respond: either by reconstructing reasoning at rising cost, or by restoring continuity through new structures.

→ The Unattained Gains — what recovering continuity would give back

This research program investigates whether reasoning infrastructure can sustain continuity of understanding under accelerating human–AI collaboration.

AuditedState.app has placed significant emphasis on experimental exploration of what we call continuity of understanding — including a central question:

Can reasoning orientation transfer across instances, agents, and time without requiring deterministic reproduction?

One experiment suggests that continuity structures may influence how ambiguity is resolved across independent reasoning instances, preserving not identical outputs, but continuity of interpretive gravity. This is a single result, not yet a pattern — worth reading as a first data point, not a trend.

This experiment explored whether shared reasoning substrate and reflective review could visibly alter reasoning trajectories across independent AI instances:

→ Transferable Understanding — Experimental Report

Publications

Research outputs are organized across two layers:

Essays — exploratory reasoning across domains, organized into thematic clusters
Papers — formal proposals and system structures, published on SSRN

Together, these form a structured approach to understanding continuity under acceleration.

→ Browse all publications

Entry Perspectives

The PKOS framework intersects multiple research traditions. Do you have a very specific interest or expertise? There are pages written for specific fields and roles, from private individuals to regulators and funders.

→ Research Entry Points

Governance Across Time

Governance is not a single mechanism. It operates across time: before, during, and after action.

This research explores how governance must evolve when reasoning becomes distributed across humans and AI systems.

Before — Structure and Intent

Governance before action defines:

policies and constraints
acceptable reasoning structures
conditions for participation

This corresponds to: Architecture and Structure & Flow.

During — Execution and Responsibility

Governance during action focuses on:

decision-making under uncertainty
responsibility boundaries
traceability of reasoning

This corresponds to: Reasoning Vehicle and Labs.

After — Audit and Learning

Governance after action ensures:

decisions remain inspectable
errors can be corrected
systems improve over time

This corresponds to: Audit & Repair, Continuity, and Continuity Bridge.

A foundational question emerges across all layers of governance:

Is your system even legal?

If systems produce outcomes without preserving the originating intention, they break the same condition required for judgement in law — where action alone is not sufficient without a "guilty mind."

→ Why Intention Matters for Judgement

Governance is not control of behavior. It is the preservation of understanding across time.

Research Context

PKOS draws on ideas from several established research traditions, including cybernetics, information theory, organizational learning, and institutional governance.

These traditions raise a shared question:

How can complex systems accumulate understanding across time without losing the reasoning that produced it?

A recurring theme across disciplines is that systems can accumulate vast amounts of information while losing the continuity required to understand it.

Open Questions

Big claims deserve open questions stated as plainly as the claims themselves. Here are the ones this research program has not answered yet.

Can we even measure understanding, or navigational coherence, directly? Almost certainly not — which is the actual argument of Measurement as Convention: measurement choices are prior theoretical commitments, not neutral observations. What can be measured are proxies for its presence or absence, not the thing itself.

What instrumentation would actually give us more insight? Candidate proxies exist but remain unformalized: four navigational proxies proposed for the economic cost of fragmentation, and — for the ambiguity/coherence side specifically — ambiguity retention capacity, ambiguity half-life, premature closure frequency, and coherence gain. None of these are settled instruments yet.

How would we know if a proxy were measuring the real thing, rather than something that merely correlates with it? This is not yet answered anywhere in the current research program, and it should be treated as open rather than assumed away.

The Transferable Understanding experiment above is one data point toward these questions, not a resolution of them. Further experiments — and independent replication by anyone willing to check this work — are the actual next step, not a formal theory declared in advance of them.