Productive performance
Compare usable deliveries, defects, waiting time, rework, and maintenance demand across the process. Does the proposed gain hold up beyond the scheduling agent's own target?
The Full Ledger
Agency, Power, Cost, and Responsibility in the Age of AI
AI expands what we can accomplish—and how far the consequences of our choices can travel.
The Full Ledger examines how responsibility can remain connected as decisions move through people, institutions, and artificial systems. This first book offers a working method for examining those relationships today: what achievement depends on, how benefits and burdens are carried, and whether the responsibilities associated with an action are fulfilled.
Complete manuscript for review
Why this matters
What follows from greater capability?
People, institutions, and artificial agents can pursue different objectives within the same system. Decisions continue through requirements, software, controls, products, and operating practice. Each action can change the conditions under which another must act, carrying productive benefits and burdens onward.
Greater capability creates opportunities for better products, more useful output, and more dependable systems. Understanding those achievements also means asking what they depend on, how their costs are carried, and what happens as conditions change.
A system can meet its own target while moving costs, uncertainty, or correction work elsewhere. The presence of a cost alone does not establish wrongdoing. The book asks whether those consequences are visible, how responsibility is understood, and whether evidence reaches someone able to respond. Design, maintenance, learning, and repair all belong in that inquiry.
An agent that completes its task can leave another with uncertainty, verification work, or repair. Across a chain of decisions, those transfers can create delay and wasted effort. They also raise questions about the responsibilities that accompany each handoff and how those responsibilities are addressed.
The method
Follow one act through the system.
Begin with one bounded decision or action, what it is intended to accomplish, and the evidence available around it. The book develops three connected questions about objectives, consequences, and responsibility. Readers can apply them with a notebook, a review team, or AI assistance, using evidence and expertise appropriate to the case.
Each act is examined in context, with attention to what was knowable, feasible, and within the actor's authority. A chain of consequences may involve several distinct responsibilities. Documented facts, interpretations, competing explanations, and unanswered questions remain visible for human review.
- 01
What did the benefit depend on?
Did achieving or protecting the benefit depend materially on moving cost, risk, uncertainty, or correction work outside the benefiting boundary?
- 02
Were the burdens given effective concern?
Were those burdens proportionately represented and acted on, considering duty, feasible alternatives, and the ability to learn?
- 03
Was responsibility fulfilled?
What appropriate, feasible, and presently due responsibility remains unfulfilled within the defined case and period? Who has the authority and ability to respond?
Applying the inquiry
What does a production gain depend on?
A hypothetical production decision, examined before deployment.
A team proposes a faster manufacturing line. Its scheduling agent can increase starts, but inspection and maintenance must absorb the added demand. If unresolved quality checks travel with the work, another team may inherit a queue, uncertainty, and rework. More starts may still yield fewer usable deliveries.
A bounded review can examine that proposal now: trace the handoffs, ask whether the expected gain depends on transferred burdens, and identify who must address them. The aim is sustainable throughput—useful output maintained over time, with quality, reliability, and the capacity to maintain the process considered together.
Responsibility in operation
Establish who can adjust the schedule, resolve uncertain handoffs, and provide inspection or maintenance capacity. Check whether corrective action reaches affected teams and whether it resolves the burden or moves it elsewhere.
This example illustrates questions the framework can help organize today. It reports no measured production gain. Any claim of improved throughput or reliability needs operating evidence and an appropriate comparison.
Worked public-record example
A technical mismatch revealed a wider systems problem.
Mars Climate Orbiter NASA investigation
NASA lost the spacecraft in 1999 after one ground-software interface reported thruster data in pound-force seconds while trajectory models expected newton-seconds.
The investigation also documented unresolved navigation discrepancies, incomplete problem reporting, weak communication, staffing and training limits, and deficiencies in systems engineering and verification.
What the method adds
It preserves the technical cause while following warning signals, documented acts, feedback, and corrective authority across the surrounding system.
What the evidence supports
These records help explain how the failure occurred. They leave a separate question unresolved: did a protected benefit depend materially on shifting burdens elsewhere? This exercise therefore supports a systems analysis but does not establish a scored responsibility gap.
The book
A narrative argument and a method readers can inspect.
The Full Ledger opens an inquiry into agency, power, cost, and responsibility. Written for the general public and industry, it brings productive achievement and its consequences into the same discussion.
The book includes a working calculus, five public-record demonstrations, and Part IV agent instructions and reference material for inspecting or independently applying the method. The calculations make assumptions and reasoning inspectable. Their scores are provisional signals for further inquiry.
- 1
The argument explains the central question at a human scale.
- 2
The working model defines the method and its safeguards.
- 3
Five demonstrations apply it to bounded public records.
- 4
The reference material supports inspection and independent use.
The book's current working model
The book's current working model follows a bounded act of agency from its objective and action through benefit, burden, feedback, authority, repair, and the changed next decision. Closing the responsibility loop requires evidence to reach corrective authority, change what happens next, and support verification that the burden was addressed rather than merely moved elsewhere.
- Benefit-Cost Transfer
- Did the protected benefit depend materially on shifting cost, risk, uncertainty, or correction work outward?
- Externalized Cost Exposure
- Were the externalized burdens proportionately represented and acted on, considering duty, feasible alternatives, and the ability to learn?
- Unresolved Responsibility
- Given what was reasonably foreseeable or credibly reported, what appropriate, feasible, and presently due responsibility remains unfulfilled within the defined case and assessment period?
Where the evidence satisfies the framework's qualifying gates, these questions inform a provisional responsibility-gap triage signal for deeper review. If a gate remains open, the arithmetic may appear only as Formula Screening—an assumption-based sensitivity exercise, not a framework finding.
AI collaboration and author responsibility
Christopher Capitan developed the framework and directed the book's creation. AI assisted with draft language, analytical outputs, graphics, and production. Capitan made the final judgments about the text and conclusions and accepts responsibility for the work.
This volume lays a foundation for examining how useful achievement can endure as consequences become clearer. Claims about improvements in throughput, quality, or reliability would need operational evidence beyond these demonstrations.
Five demonstrations appear in the manuscript. Twelve additional development exercises are available to invited reviewers. Invited reviewer access
About the author