You are here: Articles --> 2002 -->
Hart’s law: the magical number three, plus or
Vous êtes ici : Essais --> 2002 --> Hart’s law: the magical number three, plus or minus zero
by Geoff Hart
Previously published as: Hart, G.J. 2002. Hartís law: the magical number three, plus or minus zero. Usability Interface 8(4):4–5.
George Miller, infamous for his “magical number seven, plus or minus two”, somehow missed an even more important principle of how the world works: no matter how clever we think we are, it still takes us three tries to get anything approximately right.
You can’t really take any shortcuts and succeed in only two tries, other than by blind chance, and though most of us have proven beyond a shadow of doubt our ability to blunder around and take many more than three tries, the overwhelming majority of us get it nearly right on the third try. Microsoft, for example, produced two “idiosyncratic”[*1] releases of Word 97 until Service Release 2 finally got it more or less right. Similarly, Windows 3.1 was useful but problematic, Windows 95 removed many of its rough edges, and Windows 98 finally got it nearly right. (Lest I be accused of partisanship, I hasten to add that Apple also routinely takes three tries.)
[*1] Let’s be polite.
There are three strikes in baseball, three corners in a triangle (love or geometrical), three prime operators in symbolic logic (and, or, not), three layers in a sandwich, three elementary components of the atom (electrons, protons, neutrons), three parts to defining energy (mass, the speed of light, and a square), and three parts to a syllogism. A coincidence? I think not.[*2] I suspect I’ve stumbled upon a fundamental law of nature here. So what are the important underlying reasons for the magical number three?
[*2] Also three words—the examples are endless!
In our first try at solving a problem, we usually rely on stereotypes, personal opinions, the Psychic Friends Network, and a careful statistical analysis of several weeks worth of horoscopes to come up with a reasonable guess at what should work. Such a rigorous design process generally functions well enough to get us into the right ball park, which is a wonderful result unless we’re metaphorically playing hockey rather than baseball.
Being good designers, we immediately begin soliciting feedback on our first draft—and discover to our dismayed horror that we’ve somehow managed to misunderstand our audience’s needs, and produced only an approximation of what they really need. Sadly, all the “bonus” Ginsu steak knives in the world aren’t going to satisfy them if the salad shooter itself doesn’t work right, so we dive back into the fray with a second try. This new, improved design not only cuts through Coke cans, it responds better to the needs our newly irate and activist audience has energetically brought to our attention. (There are, after all, perils in offering to listen. For one, people start talking…)
Crossing our fingers, we patch up the design, dress it up in new garb brighter than the emperor’s new clothes and equally resistant to fickle fashion trends, then send our creation back out into the hostile world once more, hoping that everything’s now perfect and that our ears will soon ring with the audience’s plaudits. Yet the physics behind the magical number three lurk in wait for anyone so foolishly optimistic…
Inevitably, we discover that we’ve guessed right and solved some of our audience’s problems. That would be a cheering thought, were it not for the fact that earnest good intentions notwithstanding, we’ve also misunderstood much of the audience’s feedback and come up with incorrect solutions to problems we’ve correctly identified, correct solutions to problems that nobody identified as needing a solution, and elegant solutions to entirely different and unexpected problems.
Heaving a patient sigh, we lick our wounds, return to the drawing board, and come up with our third try at solving the new crop of problems. And lo and behold, the third time’s the charm[*3], if I might be permitted to mix the hard science I’ve relied on thus far with a somewhat more mystical reference: we’ve now achieved a usable product that satisfies most of our audience.
[*3] If anyone needed more proof, this idiom reveals an age-old recognition of the magical number three in operation.
Our world would be a happier place if this were all we had to worry about, but sadly, those who develop the products we document are never satisfied with “good enough”, and perversely insist on one more kick at the cat. The cat, predictably, isn’t particularly happy with this design approach, and demonstrates why small furry predators have claws. Not long after, we find ourselves back where we started, trying to design something that will satisfy our audience. Three tries later, Hart’s law predicts, we’ll probably get there, only to discover a new cat, a new audience, and a new manager—and then, haunted by yet another instance of the magical number three, we start the process all over again.
Rather tri-ing, but physics can be that way some days.
©2004–2018 Geoffrey Hart. All rights reserved