Control Nothing, Influence Everything
Replace control fantasies with deliberate influence and causal humility
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 5
- Confidence
- 97%
This mental model distinguishes control from influence. In a complex world, no person can guarantee an outcome because every action interacts with other people, environmental conditions, prior history, and chance. Yet this does not imply passivity: every decision changes probabilities and creates ripple effects. The method asks practitioners to define their intended influence, identify uncontrollable dependencies, choose proportionate interventions, and evaluate consequences across different time scales. Power and hierarchical position amplify immediate influence, but ordinary actions can also matter through longer causal chains. The resulting stance combines responsibility with humility: act deliberately, avoid claiming ownership of the entire result, and remain prepared for outcomes your model could not foresee.
Origin
Extracted from Masters in Business, where Brian Klass repeatedly summarized the philosophy of Fluke with the phrase that people control nothing but influence everything.
Core principles
- 01Absolute control is unavailable in complex systems.
- 02Every action produces ripple effects.
- 03Influence varies by position, reach, and time scale.
- 04Uncertainty does not eliminate responsibility.
- 05Humility supports better intervention than overconfidence.
How to run it
- 1
Define the desired influence
State the change you want to encourage without framing it as a guaranteed result. Express success in terms of improved probability or direction.
Pro tip Use language such as increase, reduce, encourage, or protect rather than ensure.
Watch out A target framed as total control invites overconfidence.
- 2
Separate agency from dependency
List what your action directly changes and what remains dependent on other actors, timing, and chance. This establishes realistic boundaries.
Pro tip Include dependencies that appear trivial but could become pivotal.
Watch out Do not equate an unmodeled variable with an irrelevant variable.
- 3
Choose a proportionate intervention
Take an action that improves the desired outcome while limiting irreversible harm. Preserve options where uncertainty is high.
Pro tip Favor reversible experiments before large top-down interventions.
Watch out High confidence based only on past patterns is not proof of control.
- 4
Trace ripple effects
Consider immediate, secondary, and longer-term consequences for affected people and systems. Look beyond the headline outcome.
Pro tip Ask how the manner of achieving the result changes what follows.
Watch out Binary success metrics can hide consequential differences in the pathway.
- 5
Learn without inventing inevitability
Review what happened while preserving uncertainty about causation. Update future actions without constructing a neat story unsupported by evidence.
Pro tip Record what remains unknown alongside lessons learned.
Watch out Hindsight bias can turn luck into a false recipe.
In the wild
A manager wants a team to surface risks earlier. Instead of mandating perfect forecasts, she creates reversible pilots, rewards dissent, and schedules premortems. She cannot control which shocks occur, but she changes the probability that the team notices and absorbs them.
→ The team becomes more adaptive without pretending uncertainty has been eliminated.
Common mistakes
Confusing humility with passivity
Lack of control does not mean actions are meaningless. The model explicitly treats every action as an influence on future possibilities.
Claiming sole credit
Outcomes arise from many interacting causes. Taking complete credit obscures dependencies and encourages future overreach.
Ignoring delayed effects
An action with little immediate impact may create major consequences over a longer time scale. Evaluation should not stop at the first visible result.
Is it for you?
Best for
It is best for leaders and individuals making consequential choices under uncertainty.
Not ideal for
It is not ideal for tasks where direct mechanical control and deterministic verification are genuinely available.
From the transcript
“we control nothing, but we influence everything.”
“There's nothing that we have absolute control over, but everything that we do has ripple effects.”
“everyone is influencing the future all the time.”
From the episode
Cause and Effect on a Global Scale