Neatness doesn't count after all - tidy vs. untidy desks

Hugh Kenner

There are clean-desk people--you know them, you may even be one--whose working space always looks as if it were scrubbed for surgery. They make a virtue of handling no piece of paper twice--''Do something with it right now. Don't dither. 'In doubt? Throw it out.' '' Any time the clean-H desker takes down a book, it's no sooner snapped shut than back with it to the shelf. Each piece of paper summoned from the files is rebounded instantly to the files. The steady stream from In Basket gets deflected just two ways: to Out Basket, to trash. Promptly at five, the clean-desker smugly departs from a place where the only hint that anything ever happened all day is an overflow-

ing wastebasket. (On noticing a wastebasket's tendency to overflow, the dedicated clean- desker gets a bigger one.) Off duty, clean-deskers measure their vermouth with an eye- dropper, walk their dogs by the clock, succor their spouses by the calendar. Such people exist, and some of them ask fees for training de-centered souls to be just like them.

But there are also souls like me, content amid what clean- deskdom calls unholy clutter. Cleaning up the room I'm sitting in at this moment, to the extent of meeting clean-desk standards, would take a week. The few times I've tried it, useful things have disappeared forever: things I routinely laid hands on back when they were integrated with the mess I fondly manipulate. I am, to put it mildly, an untidy-desker.

But untidy-deskers of the world may take heart. It's we who havemathematical validation. Forget what you may have thought about the swept and tidy world of numbers. Concentrate on the fine randomness of Einstein's hair. We connoisseurs of scrutable chaos have been guided all along by an inscrutable proposition called the 80-20 rule: a special case of Zipf's Law (of whom, and which, more later).

Please observe that mess tends to accumulate for good reasons. Taking down Eric Partridge's invaluable Origins to check, as I often do, the pedigree of a word, I reflect that before long I'll surely be checking another. So instead of returning Origins to the shelf, I leave it (for now) on my desk. And this letter from Alabama about the conference: though it has just now been answered, I'll be needing its dates when I call the travel agent. It may as well stay (for now) on top of Partridge. As for this sheaf of notes for the piece that's due next week, let's leave it here (for now) to remind me to get started. And when I've finished, it may as well stay a while longer, because editors will have queries . . . Thus ''for now'' stretches out and out, and stuff accumulates.

There's a principle behind all this, and it's not sloth. It's this: What you need now you're quite likely to need again. Human experience says so. Conversely, What you don't need now is something you'll quite possibly never need. Human experience again.

Books: every librarian keeps thousands that no one ever seems toask for, and knows that the longer any book goes unwanted the greater are its chances of staying that way. Widgets: the kinds that aren't selling will probably never sell; less rule-bound than librarians, vendors hold clearances. Or language: we've all learned just how much of it we can safely let doze in thesauruses for occasional summoning. But some words--a few hundred, a few thousand--can't be done without, and they come to mind instantly. We make over 50 per cent f our normal talk by recycling only about 100 words.

Parsimony is structured into language itself. Shakespeare's glorious vocabulary extended to 29,066 words that we know of by computer count, yet just 40 of them make up fully 40 per cent of the text of his plays. If words had to be taken down and put away like books, Shakespeare would have kept those 40 piled on his desk, and a pox on ''clutter.''

Time and again such intuitions surface. Many a secretary has noticed that the file she just fished out is the same one she fished out yesterday; if she's shrewd she will stick it (for now) not back in its alphabetical place but in an ''active'' pile where it'll be handy next time. And experience suggests that there will be a next time. (To avoid any look of disorder she keeps the active pile in the filing cabinet, but up at the front of the drawer.)

Long ago someone noticed that the files resting out of place at the front--the ones in heavy use--amounted to about one- fifth of the drawer's contents. Further study produced a rule of thumb: 80 per cent of the action involved 20 per cent of the files. And the 80-20 rule was born. Though no one seems to know who formulated it, lovers of the untidy desk will want it engraved on a platinum bar for enshrinement at the National Bureau of Standards.

I'll rephrase it: the greater part of any activity draws on but a small fraction of resources. That small fraction may as well stay handy. Though the numbers will vary, 80-20 gives us a feel. Anything of that order--80-20, 75-27, 81-11--says ''a lot done with a little.''

So if 20 per cent of the contents of the room is piled on your desk instead of being stowed in the out-of-sight places where clean-deskers try to tell you it belongs, then 80 per cent of your needs can be satisfied by what's instantly within reach from where you sit. At the cost of a little rummaging, of course.

It's a tantalizing proportion, 80-20. In Volume 3, Sorting and Searching, of his classic work in progress, The Art of Computer Programming, Donald Knuth of Stanford mentions the 80-20 rule as ''commonly observed in commercial applications.'' Knuth cites a 1963 issue of the IBM Systems Journal, where we find a man named Heising assuring us that the same principle applies in turn to just the active 20 per cent.

Thus if we keep 1,000 files, of which 200 bear the workaday brunt, then 20 per cent of the 200--that's 40 files, or four per cent of the original total-- get 80 per cent of 80 per cent-- that's 64 per cent--of the use. 80-20; 64-4; by venturing one more step we find that fully half the activity entails only about eight of all those 1,000 files: 51 per cent versus 0.8 per cent. A moderate untidy-desker might prefer to be called a 51-0.8 person. Of a thousand folders, a mere eight scattered on your desk needn't seem unruly.

The reason the IBM Systems Journal took notice in 1963 is that back at the dawn of computerdom, when punched cards were clumsy and memory cost a shah's ransom, searching through thousands of cards for the Widget account was a task to abridge even when steel fingers did it. And the 80-20 rule said, if you've searched for the Widget card once, it's apt to be wanted again; move it up to where the next search will hit on it speedily. Skeptics might have called that fudging, but IBM, adjusting its wide blue ties, pronounced it Computer Science. And it was. The essence of computer science is putting numbers on what really goes on.

Nor has cheaper and fast- er hardware diminished Big Blue's interest. Just recently IBM researchers reported noticing that computers spend most of their time performing about 20 per cent of the instructions they're capable of. Whole new machines are now exploiting that fact.

This is all very well, but you might like some proof. Not long ago, stuck midway in a wholly different project, I did what I often do while getting unstuck: I tinkered at a computer program meant to do something soothingly irrelevant. Soon it was fetching me some modest statistics on the habits of the novelist Henry James.

James kept a clean desk; in fact, at various places in his housein Rye, England, James kept eight clean desks. But en route from desk to desk in his ceaseless flight from clutter, he could never leave behind his considerable vocabulary, and he seems to have used it exactly as I use my mass of papers. The words he kept within easy reach, so to speak--say 20 per cent of them--were the ones he put to use some 80 per cent of the time. They're the same ones we all keep handy. When James did want something fancy he could always hesitate and grope.

In an instance I happen to have handy--2,339 words, most of BookII, Chapter 1 of the authoritative New York edition of his 1903 novel The Ambassadors--there are just 665 different words, all told. With no more than those, James somehow managed atmosphere, and narrative, and dialogue, and several instances of psychic crisis--and how he did that is instructive.

The 665 words seem a meager resource. Spread evenly over the pages like soft margarine, they'd turn up with dull uniformity, each one just three or four times, leaving you conscious of a certain poverty. But James used ''the'' 86 times, and ''you'' 72 times, and ''to'' 62 times, and ''he'' 56 times--you see the pattern. Fully 75 per cent of the chapter's carpentry is done with only 176 words-- 27 per cent of its vocabulary. That left 489 for special effects. When he mentions a fire ''burnt down to the silver ashes of light wood,'' four of the nine words (the ones in italics) in that lyrical phrase are making their unique appearances.

Now look back at the paired percentages: 75-27. Not 80-20, but you see the shape.

Or here's T. S. Eliot's The Hollow Men, a much slenderer artifact, just 417 words. Eliot included 187 different ones, a surprising proportion when you think how his poem keeps recycling its bleaknesses. And (again) fully 71 per cent of the whole is accounted for by just 65 words--35 per cent of the the total vocabulary. Ah, 71-35! Remember 80-20? The remainder--122 words--make one-time appearances, as when three of them lend tang to a single line: ''The supplication of a dead man's hand.''

Thirty-seven years ago, in what he called an introduction to human ecology well before ecology was a buzzword, George Kingsley Zipf (1902- 1950) of Harvard spelled out in explicit detail what it was that yielded such uniform numbers. It was human intelligence, constantly estimating the path of minimal bother: exactly what I'm doing when I don't put things back where they're said to belong.

Zipf, a philologist, gave his book a frank title, Human Behaviorand the Principle of Least Effort, and if he drew many of its data from our behavior in speaking or writing, it was because in that domain statistics were handy. Scholars of language had been extending themselves from compiling word lists to counting the words they'd compiled, and Zipf was the man to see a use for such data. What they illuminated wasn't ''linguistics'' but the way people manage resources like time and effort.

He got much mileage from a project Miles Hanley, an

English professor at the University of Wisconsin, had completed in pre-computer 1937 with $148 of university money and 19 assistants sorting 250 pounds of cards. That was a mimeogrphed ''Word Index to James Joyce's Ulysses,'' in the statistical appendices to which Zipf's eye was caught

by a surprising symmetry. Appearances of the tenth most frequent word: 2,653. Of the 100th: 265. Of the 1,000th: 26! That seems too neat to be true. But it is true. Something was balancing.

It resembles one of Zipf's neatest demonstrations, based on the 1930 population of the 50 largest U.S. cities. The largest was New York. The second largest had half the population of New York, Number 3 one- third the population of New York. And on down to number 50, with, yes, 1/50th of New York's population. A series like 1, 1/2, 1/3, 1/4 . . . is called harmonic. Zipf's Law says that any allocation of resources (people in cities, words in books, tools in a toolbox) will settle down to a harmonic arrangement.

Such tidy balancing had nothing special to do with the verbal fastidiousness of James Joyce. Ulysses simply provided Zipf with a sample the right size for displaying language in long-term equilibrium. He'd already spotted equilibrium in 44,000 words scribbled against deadlines for the Buffalo Sunday papers, and when the Joycean data came to hand he wasn't surprised to find the same pattern exactly. Zipf in pursuit of his law was a hard man to surprise.

The vocabulary of Ulysses happens to be about the size of Shakespeare's--29,899 words. If we think of a circus with that many people, we find ringmaster Joyce, like Shakespeare and the Buffalo reporters, depending heavily not on his high- wire stars but on his roustabouts. Fully half the bustle in the 260,430-word book comes from a mere 135 words (the, of, and, a, to, in). At the other extreme more than half the total vocabulary, 16,432 words, make just one appearance (ventripotence, yak).

What goes on, according to Zipf, is this: Words that say much, like ''entropy'' and ''ecliptic,'' help us be brief. That's why technical terms evolve: the few who know what they mean save a lot of time. But short common words that spell things out are easier on the rest of us, though it takes a lot of such words to specify what ''entropy'' can sew up in three syllables. So a balance is always being negotiated, fifty- dollar words traded against handfuls of penny ones. And the working of language mirrors those trade-offs. The frenzied action goes on in the bargain basement; hence Shakespeare's--and everyone's--40 or so words to do 40 per cent of everything.

If you think I've left my desk top far behind, perhaps to distract your gaze from it, reflect on Zipf's implication that humans use language exactly as they use whatever else gets in its own way out of sheer variousness: papers, books, kitchen tools. What a nutmeg- grater does is grate nutmegs, and that's all. What a paring knife does is so many things that you'd never finish list-

ing them. Likewise, the rare words give dictionary writers no trouble at all. The big Oxford disposes of ''colubri form'' (snake-shaped) in just seven lines. But ''set,'' the Swiss Army knife of the English word-kit, is handy in any thinkable context--get set to set the table with the dinner set, set the alarm so we can set out early, and set things up so we'll not be upset by a prowler but can set our teeth and set a dog on him. The Oxford entry on ''set'' was 30 years in the pondering, took 40 days to digest the examples, and ran two-thirds the length of Milton's Paradise Lost.

By Zipf's reckoning, use always tends to draw in close what's used. He scatters odd instances. For example, a check of Philadelphia marriage licenses in 1931 revealed a surprising gravitation toward the girl next door. The shorter the distance, the likelier the pairing. In the range of up to 20 blocks, a 70-30 rule seemed to be operating: some 70 per cent of the unions sparked within 30 per cent of the distance. And the ghost of 80-20 winks again.

Though Zipf never seems to have mentioned the 80-20 rule, it gives us a ballpark formulation of Zipf's Principle of Least Effort, the law by which he sought to explain absolutely everything.

To make it seem plausible to non-mathematical readers, Zipf invented an analogy I find congenial. He asked us to imagine an artisan (me) with numerous tools (books, papers) on a bench (my desk), this artisan being charged (as am I) to do set jobs with a minimum of total effort. He then showed with ease how the ratios that support Zipf's Law would arise, as the tools got put where they'd be wanted.

My need to minimize my total effort originates with me. I want to save time, thinking I have other uses for it. So the books and papers I expect to have most use for I simply don't put where I'd have to get up and fetch them. And when I judge that I'll be using something again because I'm using it now, both Zipf's Law and the 80-20 rule say that probabilities are very much on my side.

If some things, though, do eventually get put away, that's because the efficiencies of clutter can be offset by the effort of fumbling through it. Remember, it's total effort we're trying to minimize; and the human mind, cunning in judging how much of a mess will really abet efficiency, makes estimates you'd need calculus to describe. That's not implausible. An outfielder solves trajectory problems every time he puts his glove where the ball will be.

Zipf's Law and its quick approximation, 80-20, confront a human predicament (much to do, finite time) that's inextricable from the way we cope with it (keep handy whatever we expect to use). The law reflects our expectation of what will be used most, and experience tends to make the expecta- tion reliable. So Zipf and 80-20 end up sketching what actually goes on, quite as if they were laws of impersonal Nature, like gravity and thermodynamics.

But at bottom--so Zipf assured us--such laws work because they describe the situation we create in the course of intelligent coping. Situations like my desktop.

COPYRIGHT 1986 Discover
COPYRIGHT 2004 Gale Group