Bounded Rationality and Satisficing

The idea in plain words

Classical economics pictured the decision-maker as someone who knows every option, foresees every consequence and picks the best. Nobody decides like that. A product manager choosing a vendor does not evaluate every vendor in the world; she looks at four, finds one that meets the requirements and stops. That is not a failure of rationality. Given limited time, attention and information, it is the rational thing to do.

Herbert Simon called this bounded rationality, and the strategy of searching until something clears a bar he called satisficing, a blend of “satisfy” and “suffice”.

Where it comes from

Simon began with organisations. In Administrative Behavior (1947) he studied how real administrators decide, and found them working with partial information, simplified models and rules of thumb. In 1955 he proposed a formal alternative to the economic model: an agent who has an aspiration level, examines options one at a time and accepts the first that meets it. If options are easy to find, the aspiration rises; if they are scarce, it falls.

His 1956 paper added the idea that made the theory deep. Picture an organism foraging for food in an environment where food is scattered in certain ways. Whether simple search rules work depends on how the environment is structured. Rationality, Simon argued, is a fit between a mind and its world. He later expressed this with an image of scissors: one blade is the mind’s computational limits, the other is the structure of the task, and you cannot understand the cutting by looking at one blade alone.

Simon received the Nobel Prize in economics in 1978. His ideas split into two traditions. One, led by Daniel Kahneman and Amos Tversky, emphasised the systematic errors that shortcuts produce: biases. The other, led by Gerd Gigerenzer, emphasised how well simple rules can work when they match the environment. Gigerenzer and Goldstein showed in 1996 that a rule as crude as “pick the option you recognise” can match or beat elaborate statistical models in certain environments, because recognition carries real information there.

What it does to the mind

Bounded rationality is a description of the mind before it is a critique of it. Attention is the scarcest resource in any decision. Simon put it memorably in a later talk: a wealth of information creates a poverty of attention. Every decision process is therefore a way of rationing attention: what to look at, when to stop, what to ignore.

The danger lies in the mismatch between rule and environment. A stopping rule that works for routine decisions, such as accepting the first option that meets the brief, fails badly when the environment changes or when the decision is unusual. People rarely notice when they have moved from one kind of environment to another, so they keep using the rule that served them.

Organisations satisfice too. March and Simon described how organisations search for solutions near existing ones, in the neighbourhood of the problem and of current practice. This is efficient most of the time and blind exactly when the answer lies far away.

Where it shows up

Much of technology history is a story of good-enough designs beating optimal ones because they fit the environment of the time. Unix, as the operating systems essay describes, succeeded partly by leaving out what its more ambitious predecessor tried to include. The Plan 9 and Lisp machine roads show the opposite risk: designs closer to some ideal that lost to systems that were merely adequate and everywhere.

Amazon’s distinction between one-way and two-way doors, in the 2015 letter, is in effect a theory of how much search a decision deserves: satisfice quickly on reversible choices, search hard on irreversible ones.

What it asks of someone deciding

Make the stopping rule explicit. For each important decision, say how many options you will look at, what bar an option must clear, and what would make you look further.

Match effort to stakes and reversibility. Most decisions deserve less analysis than they get. A few deserve far more.

Ask whether the environment has changed. A heuristic that has served your team well for years may be a fit to a world that is ending. The most dangerous moment for a good rule of thumb is just after the conditions that made it good have gone.