The OSS sabotage manual instructs agents to insist on channels, refer everything to committees, haggle over wording, and demand written orders. The Clueless do the same things sincerely every day. The output is identical. This is not a coincidence or an insult. It is a structural theorem: any system that prioritizes control over output will converge on sabotage-equivalent behavior, regardless of whether anyone intends harm.
Simple Picture
Two factories. In Factory A, an enemy agent has been planted. His mission: slow everything down without getting caught. He insists on triple-checking every order, routes decisions through committees, demands written approvals for routine tasks, and questions the wording of every memo. In Factory B, no agent exists. But the management, terrified of liability and obsessed with legibility, has independently instituted triple-checking, committee routing, written approvals, and precise wording standards. Both factories produce at the same dismal rate. The saboteur in Factory A is superfluous. Factory B is sabotaging itself with more enthusiasm than any enemy could muster.
The Convergence Theorem
Why does sincere institutional behavior converge on deliberate sabotage? The answer runs through three garden notes that, taken together, describe a single mechanism from different angles.
The Theory of Constraints provides the first piece: any improvement not at the bottleneck is an illusion, and in an interdependent system local optima actively make things worse. When every non-constraint team “stays busy” to justify their headcount, they flood the constraint with work, overhead, and interruptions. This is exactly what the sabotage manual prescribes — keep everyone frantically active doing anything except the one thing that matters.
The fog of work provides the second piece: when individual contribution is invisible, the only rational strategy is social manipulation. Big Projects make headcount trackable and individual engineers controllable — which serves management’s need for legibility — but create the fog that makes politics inevitable. The saboteur exploits the fog. The sincere manager creates it.
The lion-fox cycle provides the third piece at civilizational scale: foxes optimize for perception management, narrative control, and process — exactly the competencies that produce sabotage-equivalent behavior when applied to organizations that need to actually build things. The fox elite cannot adopt the lion’s approach (direct action, accepting illegibility, tolerating disorder) without ceasing to be foxes. So they keep adding process, keep demanding legibility, keep managing perception — and the wolves keep coming because wolves do not read position papers.
Why Intent Does Not Matter
The deepest version of this argument: it is not that bad actors exploit good systems. It is that the systems are optimized for a goal (control) that is structurally opposed to another goal (output), and any system sufficiently optimized for control will produce the same behavior patterns as deliberate sabotage, because control and sabotage share the same operational grammar.
Both require:
- Routing through channels — because channels make activity visible to whoever sits above the system
- Committee decisions — because committees distribute accountability, which protects individuals at the cost of speed
- Written documentation — because the written record is the medium of legibility and the weapon of retrospective blame
- Precise wording — because ambiguity is power, and those who control wording control outcomes
These are not corruptions of institutional purpose. They are the direct expression of institutional purpose when that purpose is control rather than production. The self-justifying system runs the same loop: the system produces dysfunction, which creates demand for more system, which produces more dysfunction.
The Dead Sea Effect ensures the convergence accelerates over time. The people most capable of producing output — who are also most frustrated by sabotage-equivalent behavior — are the first to leave. The residue, who are most comfortable with process-as-work, entrench. The institution’s talent pool evolves toward people for whom the sabotage manual describes not dysfunction but normalcy.
The Minsky Version
The Minsky cycle reveals the same convergence in financial systems. Stability breeds complacency breeds leverage breeds fragility — and each stage of optimization looks brilliant right up to the collapse. The bank that eliminated buffers for efficiency is running the sabotage manual on its own balance sheet: every buffer removed is a vulnerability introduced, and the removal looked like competence because it improved returns during the stable period.
The non-ergodic frame sharpens this: the ensemble average says the optimized system outperforms. The time-series reality for any individual institution includes the possibility of total collapse — and the optimization that produces ensemble outperformance is structurally identical to the optimization that produces individual ruin. The saboteur and the optimizer are using the same playbook. The only difference is that the saboteur knows where it leads. The mimetic gradient at the elite level is what lets the convergence reach institutional saturation: fox elites mimetically synchronize around whoever most thoroughly embodies the control paradigm, and dissent is eliminated not through censorship but through the gradient itself — the independent judgment that would recognize sabotage-as-competence for what it is cannot survive outside the consensus that produced the paradigm.
The Paradox of Competence
Here is the twist that makes convergent sabotage genuinely dangerous: the better an organization gets at control, the more perfectly it replicates sabotage behavior, and the less able it is to recognize what is happening — because recognizing it would require questioning the very competencies that define institutional excellence.
This is paradigm-lock-in applied to organizational design. The framework that makes someone a good manager (process-oriented, documentation-heavy, accountability-focused, control-seeking) is the same framework that makes the sabotage manual’s instructions seem like common sense. The most locked-in person is the one who reads the sabotage manual and thinks: “yes, this is just good management.” They are not wrong about what the manual describes. They are wrong about what it produces.
The reflexive loop closes the trap: the story that process equals competence creates the investment in process that creates the complexity that confirms the need for more process. The story is load-bearing. Removing it — admitting that the carefully constructed system of controls is producing the same output as deliberate sabotage — threatens every career built on process expertise.
Dimwit / Midwit / Better Take
The dimwit take is “bureaucracy is the enemy — burn it down and let people work.”
The midwit take is “some process is necessary — the answer is better process, not less process.”
The better take is that convergent sabotage is not a process problem but an optimization target problem. The organization optimized for control will always converge on sabotage behavior because control and sabotage share operational DNA. The organization optimized for output will tolerate the illegibility, disorder, and loss of control that output requires — which means accepting that managers cannot track everything, committees do not approve everything, and some work happens in ways that no documentation captures. The exploration-exploitation frame applies: premature exploitation (staying busy, maintaining control) destroys the system’s capacity for the focused, messy, illegible work that actually moves the constraint. The fix is not better process but a different question: “are we optimizing for control or for output?” Most organizations have never asked because the answer is uncomfortable and the fiction of productivity is holding up too many careers.
Threads to Pull
Ideas, concepts, thinkers, and questions worth exploring further — and why.
- Jerry Z. Muller, The Tyranny of Metrics — Muller documents how metric fixation produces the same pathologies as the sabotage manual across healthcare, education, policing, and the military. The “metric fixation” concept is convergent sabotage measured in KPIs. Would deepen the connection to goodharts-law and the fog of work.
- David Graeber, Bullshit Jobs — Graeber’s thesis that a significant fraction of modern employment is self-assessed as pointless maps directly onto convergent sabotage. The person who knows their job is bullshit is living inside the sabotage manual. What distinguishes Graeber’s “duct tapers” (who fix problems created by the system) from the sabotage manual’s “work sabotage” instructions?
- How does China’s state capacity interact with convergent sabotage? — Beijing’s brute-force solutions are the lion’s approach to problems the foxes would manage with process. When Xi terrorizes the bureaucracy, he is disrupting convergent sabotage — but also disrupting the coordination that bureaucracy legitimately provides. Is there a way to disrupt the convergence without destroying the institution?
- The sabotage manual as organizational diagnostic — Take the manual’s tactics, convert them into a checklist, and score your organization. Each match is not evidence of a saboteur but evidence of convergent sabotage. Has anyone formalized this? If not, it would be a sharp practical tool.
- Stafford Beer’s cybernetic management — Beer’s Viable System Model attempts to solve the control-vs-output problem by distributing control recursively rather than centralizing it. His work in Chile (Project Cybersyn) is the most ambitious attempt to build an organization optimized for output that still maintains coordination. Connect to theory-of-constraints and the question of whether the convergence can be broken structurally.