The Trouble with Optimization in a Complex World What maximizing and minimizing get wrong in complex contexts

Business language often assumes the world is stable, measurable, and controllable. In complex contexts, words like maximize, minimize, and optimize quietly overclaim what can be known or achieved. A more honest approach treats decisions as hypotheses, focuses on direction rather than endpoints, and keeps learning open as situations change.

The language we reach for too easily

In business articles and marketing materials, certain words appear with striking regularity: maximize, minimize, optimize. They are used confidently and rarely questioned.

How do you maximize return on investment?
How do you minimize staff turnover?
How do you optimize team performance?

The problem is not the intent behind these questions. Wanting better outcomes is reasonable. The problem is the language itself. It assumes that the world is sufficiently stable, measurable, and controllable for these ideas to make sense.

In a complex world, that assumption does not hold.

What complexity allows us to know

In a complex environment, how would you ever know that you have maximized, minimized, or optimized anything? You cannot. There is no clear endpoint, no known ceiling, no agreed optimum waiting to be discovered.

All you can ever observe is movement. Something increased or decreased. Perhaps noticeably, perhaps marginally. And even that judgment depends on where you are standing, when you are looking, and what you choose to pay attention to.

This is why so much business language quietly overclaims.

“Simple strategies to maximize profit” would be more honest as “Simple strategies that may increase profit”.
“The best ways to minimize cost overruns” would become “Some possible ways to reduce cost overruns”.
“How to optimize employee performance” might be rephrased as “Some potential ways to improve employee performance”.

These versions sound less decisive, but they are more truthful. They leave room for uncertainty, context, and learning. They acknowledge that outcomes emerge rather than being engineered.

Optimising machines, not organisations

Optimization works when the system is bounded, repeatable, and largely predictable. You can optimize a machine. You can tune a process under controlled conditions. You can often make sensible trade-offs and say, for now, this configuration performs better than that one.

Organizations are different. They are living, social systems. Relationships, histories, power, emotion, incentives, and local meaning shape them. What improves performance in one context may degrade it in another. What reduces turnover this year may store up fragility for the next.

Treating organizations as if they can be optimised in the same way as machinery flattens this reality. It encourages overconfidence and discourages curiosity. It privileges neat answers over careful judgment.

Decisions as hypotheses, not solutions

In complex situations, there is no reliable way of knowing in advance whether a decision is the best one, or even whether it is a good one. Conditions change, time passes, and action itself alters the situation. The value of a decision can only ever be evaluated retrospectively, and even then only partially and provisionally.

Seen this way, decisions are not final answers to be optimized. They are provisional moves in an unfolding situation. Each decision expresses assumptions about how the system might respond, assumptions that can only be explored through action.

Decisions are therefore better treated as hypotheses to be tested, reflected on, and revised, rather than as solutions to be perfected.

A Decision Is Never Final in a Complex World

Why decisions work better as hypotheses than conclusions

In complex environments, nothing stands still. Time passes, conditions shift, and action alters what follows. Treating decisions as final closes conversation and hides doubt. Treating decisions as hypotheses keeps learning alive and allows adjustment as the world continues to change. Continue reading

This also exposes the limits of optimization language. If decisions are hypotheses, there is nothing meaningful to optimize in advance. There are only directions to explore, signals to pay attention to, and consequences to reflect on later, including the unintended ones that inevitably arise.

Living with direction rather than certainty

A more grounded approach in complex environments is to think in terms of direction rather than destination. Not “Have we optimized?”, but “What seems to be improving, and at what cost?” Not “Is this the best decision?”, but “What are we learning, and what might we need to adjust?”

This way of thinking is less comforting. It offers fewer guarantees and weaker claims. But it is also more honest.

In complex situations, progress rarely comes from finding the optimum. It stems from staying attentive, remaining open to challenge, and being willing to revise our thinking as the world responds to our actions.

In complex situations, we should choose our words more carefully and notice what they imply. We should frame decisions as experiments, watch what changes over time, and adjust as we learn. Rather than chasing perfect answers, we should stay attentive to direction, consequences, and the signals the system gives back.