Published March 3, 2026 | Version v1
Preprint Open

Referential Displacement and Ritual Capture in Organizational KPIs: A Five-Stage Unified Framework for Measuring Epistemic Transfer, KPI Validity Degradation, and Organizational Reform Capacity

Description

Organizations that measure knowledge work with Key Performance Indicators
systematically transfer epistemic authority from practitioners with direct sys-
tem knowledge to administrators who can only access official representations
of it. This paper formalizes that process as a five-stage mechanism — positional
blindness, legibility compression, channel degradation, ritual capture, reform
resistance — synthesizing thirteen theoretical traditions across eighty years
of scholarship. It derives two computable instruments from the mechanism:
V(m, O, t), a structured metric validity score grounded in information theory and
cybernetics, and RD(m, t), a linguistically computable proxy for ritual capture
derived from the ratio of metric-referential to system-referential language in
organizational documents. The paper extends the original four-stage account
by adding Stage 5 — Reform Resistance — grounded in Hirschman (1970)’s
exit/voice/loyalty framework and Bourdieu (1984)’s theory of field heterodoxy.
This extension is motivated by the empirical discovery in Paper II of the RE-
FORMER agent class: practitioners who have already detected ritual capture
and can be identified from organizational documents before Stage 5 suppression
completes. The theoretical implication is that any organization that can iden-
tify its REFORMERs before they exit or become cynics has identified the practi-
tioners with the tacit knowledge and motivation required to reverse epistemic
transfer. The formal model is validated in Paper II: Pearson r(V, RD) = 0.973,
Spearman ρ = 0.800, Mann-Whitney U = 0 (large effect), with triple convergence
confirmed against LLM pragmatic classification at r = 0.958.

Files

the_mechanism.pdf

Files (144.8 kB)

Name Size Download all
md5:e3ce9a3c35bf3ba2da8d3ea2e253960e
144.8 kB Preview Download

Additional details

Software