01
The Often Unseen Price of Suboptimal Optimum
Defines the core failure mode: local optimization can raise the metric being managed while destroying more value elsewhere in the system.
Research by Ildar “Dar” Fazulyanov
Measure what the system actually does.
Paxology studies when locally rational optimization damages the systems it depends on, why the damage becomes hard to see, and what it takes to measure it.
Start here
The Often Unseen Price of Suboptimal Optimum
A local metric can improve while the larger system gets worse. This paper defines the Suboptimal Optimum, formalizes the pattern, and develops a measurement program for the value lost beyond the local boundary.
PRELIMINARY WORKING PAPER. FORMAL RESULTS WERE MACHINE-PRESCREENED BUT HAVE NOT YET BEEN REVIEWED BY A HUMAN MATHEMATICIAN. COMMENTS AND REPLICATION ARE WELCOME.
Five working papers
The research program
Direct copies of the submitted July 2026 working papers. Every file is mirrored here from the local submission corpus and recorded in a public hash register.
Read PDF ↗
02
Past the Goodhart Horizon
Shows that, under a stated observability condition, monitors built only from a system’s trajectory cannot distinguish a benign welfare world from a collapsing one.
03
The Goodhart Escape Companion
Specifies the exogenous, welfare-touching measurement architecture needed to escape that blindness, and starts pricing the audit burden.
04
Welfare Observability
Models welfare coverage as a budgeted measurement problem and derives a cost-coverage frontier under passive and strategic conditions.
05
Net Systemic Value
Defines Net Systemic Value for AI-augmented firms: value created for downstream units minus value destroyed within them, net of the part’s own cost.
Plain-language research notes
Start with the idea
The papers carry the formal claims. These notes explain one testable idea at a time and state where the work remains preliminary.
NOTE 01
When both sides win and the system loses
Why the ordinary A/B win-win classification cannot distinguish Integration from a both-party win that degrades the surrounding system.
From papers to practice
Help test it. Help build it.
Researchers
Stress-test the formal claims, assumptions, and measurement boundaries. Replication and serious criticism are more useful than praise.
Company leaders
Test whether Systemic Value catches downstream losses that agent ROI, process mining, or local scorecards miss.
Builders and funders
Explore the measurement infrastructure, operating team, and company needed to make this usable in real organizations.
Publication status
These are independent working papers, not claims of settled science or peer review. The public copies are preserved exactly as submitted in July 2026. File sizes, page counts, and SHA-256 hashes are available in the paper register.