Doubt GraphicKnowing our limitations

by Matthew Leitch, 17 May 2002

Bounded rationality, chaotic systems, and imperfect information

Can we think of everything? This question has been answered in convincing theory by Nobel Prize winning economist and psychologist, Herbert A Simon, with his theory of bounded rationality. The answer is "No".

Simon showed that even imaginary problems, free of the messy complexities of real life, quickly create computational tasks that would keep the fastest imaginable computers busy for longer than the expeced future life of the earth. Throw in the complexities of real life and we haven't got a hope.

Even without theory, it's obvious that today's fast moving, complex world is a confusing, chaotic mess. Some of its problems people solve, but many defy solution, with one well meaning change after another failing to work.

Even if we had lots of reliable, accurate information to work with we don't have time to use it properly. But anyway, we don't have good information to work with. Apart from the evidence of our own eyes and ears we rely on a great deal of hearsay (including journalism, friends, work colleagues, and so on), often based on partial information, and yet more hearsay.

Many of the things we regard as "common sense" are no better than dubious received wisdom. Attractive, compelling, but totally wrong beliefs circulate the collective human mind like viruses.

More recently, chaos theory has clarified why some things are extremely hard to predict with certainty - impossible in practice. Some things are extremely sensitive in a particular way. Their behaviour evolves in radically different ways as a result of tiny differences in their starting state. The weather, for example.

I'm not saying that nothing is certain, nothing is predictable, and nothing can be done to solve our problems. But I am saying there are some pretty big limitations we need to be aware of.

The following are all common reasoning errors resulting from not understanding, or choosing to ignore these limitations:

Unreasonable assertions of complete certainty: Some philosophers have argued that nothing is truly certain, but holding to this approach through daily life soon gets tiresome. There are lots of things we can be so confident of that assertions of complete certainty are usually reasonable. But there are also many things where absolute certainty is not reasonable.

When is absolute certainty reasonable?
ExampleReasonable or unreasonable to be completely certain?
"The sun will rise tomorrow"Reasonable (for the time being, and not forgetting the hazards of black holes and asteroids)
"Gravity will keep on working this afternoon on earth." Reasonable, but beware of very precise predictions about its strength.
"These proposals are popular/unpopular with staff." Unreasonable, unless "staff" consists of one person, the speaker. You can get nearer certainty by asking everyone, but human attitudes are extremely difficult to establish with absolute certainty and change too.
"2 + 2 = 4" Reasonable.
"The restructuring will enable us to focus on our core business and provide the world class services our customers demand." Unreasonable. Management speak often includes unreasonable assertions of complete confidence.
"This research shows that people are more stressed now than they were 10 years ago." Unreasonable. Research on human nature is almost never conclusive and the vagueness of the conclusion means it would probably fall apart if analysed critically.

Unreasonable certainty seems to be caused by a number of things, including the effort of holding something in mind as uncertain, the way language tends to favour categories over degrees on a continuum, and the way other people demand certainty from us.

Unreasonable demands for certainty: There are at least three ways to use unreasonable demands for certainty to get your way:

Unreasonable expectations of certainty: Honest but mistaken expectations of complete certainty or of a high level of certainty on all things, not just those where it is feasible, are another form of mistake. Sometimes these expectations may have been created when people took seriously the tricks described above. The problems this can cause include:

Talk of optimisation: Another consequence of bounded rationality is that we can almost never know that we have taken the best course of action. We can find what looks like a better course of action, or select what we think is the best from the plans we have thought of, or keep working on the plan until it seems good enough, but in real life optimisation is not possible. Business leaders often say things like: "This will enable us to optimise our customer service processes and provide the best rewards for our people." This sort of statement is a mistake or a lie.

© 2002 Matthew Leitch