The Norm Lifecycle
How norms governing advanced systems form, contest, and crystallize across regulatory, judicial, academic, and civil society venues. The paper builds on Finnemore and Sikkink's foundational lifecycle model and extends it for the specific dynamics of advanced-system governance — where technical change outpaces legal change and where binding obligation often emerges from soft-law instruments before legislation catches up.
The lifecycle stages — emergence, contestation, crystallization — are operationalized into measurable signals that distinguish a norm in formation from one approaching binding force. This is the model NormSense uses to interpret what it ingests.
Cognitive Surrender and Constitutive Delegation
The moral architecture distinguishing uses of advanced systems that erode human agency from those that preserve it. The paper introduces an original normative distinction between cognitive surrender — delegation that hollows out the agent's standing as an author of their own decisions — and constitutive delegation — delegation that constitutes and extends the agent's capacities without surrendering them.
The work synthesizes Kant's account of autonomy, Korsgaard's constitutivist ethics, Strawson on reactive attitudes, Clausewitz on friction and judgment, and Nietzsche on the will. Process integrity emerges as the central concept: what makes a use of an advanced system morally defensible is not its outcome but the integrity of the process by which it was deployed.
This is the foundation that lets NormSense interpret a norm not just as a rule, but as a structure that either preserves or erodes the agency of those it touches.
Read the papers
Both papers are published on SSRN and are openly available to read, cite, and engage with.
Existing AI governance frameworks assess individual system deployments in isolation. None systematically track the crystallization of ecosystem-level industry norms or measure the gap between those norms and existing law across jurisdictions. This paper introduces the Individual Agency Impact Framework (IAIF) and its computational implementation as the first methodology to operationalize Finnemore and Sikkink’s (1998) norm lifecycle theory in the domain of advanced systems governance.
This paper introduces a distinction between cognitive surrender and constitutive delegation as the central unresolved problem in human-AI governance. The existing literature correctly identifies cognitive surrender as a behavioral and epistemic problem but lacks a principled normative account of when AI delegation is morally and structurally legitimate. The answer requires a concept the governance literature has not yet named: process integrity, defined as the presence of a constitutively engaged agent whose deliberation has genuinely shaped the structure of a decision prior to its execution.
For academic correspondence or pre-publication drafts of forthcoming work, write to zach@normsense.com.
Theory that does work
Most norm-tracking systems treat governance as a regulatory monitoring problem. NormSense treats it as a normative inquiry that requires both quantitative signal and qualitative interpretation. The two papers above are not decorative academic provenance — they are the interpretive engine that makes the platform's outputs trustworthy to regulatory affairs leaders, in-house counsel, ethics committees, and policy researchers.