Simplicity as a way of finding the real problem

Ornament is a good hiding place. That is the argument around this sentence in “Taste for Makers”, an essay Paul Graham published in February 2002 that tries to describe what good design has in common across fields. Long words, busy brushwork and elaborate code can make weak work look substantial. Strip them away and what remains is either an idea or nothing, which is why simplicity is uncomfortable: it removes places to hide.

The idea shows up in many kinds of work. A writer limited to plain words has to know what they mean. A programmer who cannot add another layer of abstraction has to understand the problem well enough to solve it directly. A founder asked to describe the company in one sentence has to decide what it actually does, and many founders discover in that exercise that they have not.

Constraints do the forcing. Twitter’s 140 characters made every message short and pushed its users to say one thing. A small budget forces a startup to find the single feature that customers will pay for, instead of building ten and hoping. In each case the limit takes away the option of avoiding the hard question.

Graham’s own first company is a small illustration. Viaweb, which he co-founded in 1995, let shop owners build online stores in a browser at a time when that meant installing and configuring server software. The simplest version of the product had to answer one question well: could a merchant with no technical knowledge get a store online in an afternoon. Everything else was secondary.

The line has two limits. Some problems are complicated, and forcing a simple answer onto them produces something simplistic. Tax software, medical records and air traffic control carry real complexity because the world they model does. The goal there is to hide complexity from users where possible, not to pretend it away.

And simplicity can be a style as easily as a discipline. A minimal interface can look clean while hiding the same muddled thinking under fewer buttons. The test Graham points to is whether the simple version faces the real problem. That is harder to judge than whether it looks simple, and more important.