Moving things to think

Watching expert Tetris players in 1994, researchers noticed they rotated pieces on screen far more than placing them required. They were not being careless — rotating on the screen was simply faster than rotating in the head.

The name they gave it was epistemic action: doing something in the world to make thinking easier. Rearranging cards on a board is not procrastination. It is the work.

That is easy to forget, because many of the ways we organise work assume the opposite. A list, an outline or a folder tree treats arrangement as the record of a decision already made: think first, then file. In the game, at least, the order ran the other way. The moving came first, and the decision grew out of it.

Two kinds of move

Kirsh and Maglio (1994) drew the line between two kinds of action. A pragmatic action brings you physically closer to a goal — sliding a piece towards the gap it will fill. An epistemic action is performed to uncover information that is hidden or hard to compute mentally — turning the piece to see what it looks like the other way round. Both are the same key press. What differs is what the press is for: one changes where the piece will land, the other changes what the player knows.

Much of their evidence came from a set-up that recorded every keystroke of people playing. Extra rotations were common, even among experts, and they often began early in a piece’s fall, sometimes at the very start. A standard account of skill — perceive, decide, then act — has to treat those moves as noise; Kirsh and Maglio argued they were doing work. By their estimate, a physical quarter-turn could take as little as 100 milliseconds; their pilot data put the same turn, done mentally, at somewhere around 800 to 1,200. Even allowing extra time to reach for the button, the screen won.

A second habit makes the point more sharply. Some players slid a piece all the way to the wall, then counted back to the column they wanted before dropping it. That move carries the piece away from where it is going, so it cannot be progress. It is a check, and players used it more often when the piece still had further to fall and the check was less risky. Kirsh and Maglio argued that experts keep the cost of thinking down partly by making more of these moves, not fewer.

The same thing happens away from a screen whenever a list turns into a layout. Six items written one under another say only which came first. Pushed around a table, they start to say which belong together, which one the others depend on, and which one does not fit anywhere yet.

What arrangement does

A year later, Kirsh (1995) widened the argument from one game to ordinary work. Among the cases in his introduction is a cook preparing an elaborate salad, who sliced each vegetable and laid the slices out in rows before arranging them around a platter she had never used. The length of each row stood for the number of slices left. She had put a cue into the world that she could read at a glance as she worked.

From videos of cooking, assembly and packing, and from observation in supermarkets, workshops and playrooms, Kirsh sorted the intelligent use of space into three kinds: arrangements that simplify choice, arrangements that simplify perception, and spatial dynamics that save internal computation. His broader claim was that, aware of it or not, we constantly organise and reorganise the places we work in to help ourselves perform.

Choice gets simpler when the options sit in front of you. Groceries rarely arrive at the till in the order they should be packed, so skilled baggers build side groups among the waiting goods — thin, fragile things in one place, things that must stay upright in another. The next decision becomes a matter of looking, not remembering.

Perception gets simpler when position carries a property that is hard to see. Washed and unwashed tomatoes look much the same, so a common ploy is to keep them in different places. Veteran jigsaw puzzlers sort edges, corners and colours into separate piles, which cuts the time spent hunting for the piece that fits. Nobody in these examples would call what they were doing thinking. Each was arranging things so that there was less to think about.

Shuffling for a new view

The third kind, saving internal computation, is where the Tetris rotations belong. It also covers a move most people already know. Scrabble players shuffle the tiles on their rack, not to play them but to see them differently. Kirsh called this self-cueing: changing what is in front of you in the hope of triggering an association that staring would not produce.

Maglio and colleagues (1999) put it to a simple test. Participants had five minutes to find as many words as they could in seven letters; in one condition they could handle the tiles, in the other they could not. More words came with the tiles in hand, but the advantage depended on the letters — clear for one set, not significant for the other. The authors read rearrangement as a way to reset a search that has got stuck in one region of the possibilities. That is a useful limit: moving things is not magic, just a cheap way to get a fresh view when the current one has run dry.

Clark and Chalmers (1998) leaned on both games when they argued for the extended mind. Stripped of the philosophy, one of their examples makes the point: when your choice of word comes out of rearranging the tiles on your rack, the tiles are part of the process that made the choice, not a display of a choice made elsewhere. You need not accept the whole thesis to take the consequence. If the arrangement is part of the reasoning, throwing it away throws part of the reasoning away.

Keeping the arrangement

Outside games and kitchens, the same moves appear whenever someone has more material than fits in their head. A person finishing a long report prints the draft and lays the pages across the floor, and a section in the wrong place becomes something they can see rather than something they have to hold in mind. Someone planning a kitchen renovation spreads quotes, measurements and photographs over a table, and notices that one fixed constraint — where the drain runs — ends up beside three decisions it quietly controls. A student revising sorts topic cards into piles for solid, shaky and lost, then moves a card back when a practice paper proves the guess wrong.

A few habits turn this from accident into method. Let nearness mean relatedness, and keep a visible gap between groups, so that distance itself carries information. Treat each pile as a guess about a category; its value is that it is cheap to break. Use moves as questions. Dragging a note into a group asks “does this belong here?”, and the answer is often easier to see than to argue.

The hardest habit comes at the end: when you stop, keep the layout. The tidy reflex is to turn it back into a list for the record, but a list keeps only order, and the layout held more — what was grouped, what stood apart, what was still unplaced. The more of the organisation you keep outside your head, the less of it you have to hold inside. Flatten it, and next time you begin by rebuilding it from memory.

This is work a canvas supports better than a page, because nothing has to stand in a line. In PlansFlow a project is an infinite canvas: notes, PDFs, links and images sit as blocks you can drag into clusters, connect with a line when nearness is not enough to show how two things relate, and find where you left them when you return. When you want a draft, you choose which blocks inform it — often one of the groups you built by moving things around.

In practice

1. 01

   When a list stops helping, spread its items out and move them until groups appear. Count that time as analysis, not delay.

2. 02

   Let distance mean something: put related items close together, unrelated ones apart, and leave a clear gap between groups.

3. 03

   Treat every pile as a hypothesis. Label it loosely, move doubtful items in and out to test whether they belong, and break it up when it stops fitting.

4. 04

   When an arrangement stops producing ideas, reshuffle it on purpose rather than staring harder at the same layout.

5. 05

   When you stop, leave the layout as it is and add one note on what is still unplaced, instead of flattening it into a list.