---
title: How Small Should You Break Your Work?
url: https://oosioo.com/en/p/%EC%9D%BC%EC%9D%80-%EC%96%BC%EB%A7%88%EB%82%98-%EC%9E%91%EA%B2%8C-%EC%AA%BC%EA%B0%9C%EC%95%BC-%ED%95%A0%EA%B9%8C
date: 2026-10-20T08:38:32+00:00
author: SYSOP
summary: My notebook said: take what you're working on now and break it into the smallest units you can. I followed, with numbers, how breaking work down has changed it: Descartes's second rule, Adam Smith's pin factory, Ford's assembly line, and Galaxy Zoo, where 100,000 people split the classification of 900,000 galaxies. I also gathered research on why small pieces feel easier: children given near goals, schedules that get more accurate once you list the steps, and a coffee card with two stamps already on it. Alongside them are the warnings the same people left behind: communication costs that grow as you divide, a division of labor that makes people smaller, and the urge to rest at every milestone. The nineteenth piece in A Page for Today.
---
# How Small Should You Break Your Work?

![On a kraft-colored background, the title "How Small Should You Break Your Work?" sits on the left. Below it, one large square is divided step by step into 4, 16 and 64 pieces. A lined note on the right reads: "Take what you're working on now and break it into the smallest units you can."](/uploads/972fcac15c13362a.webp =720)

When I opened my idea notebook today, I found this sentence.

> Take what you're working on now and break it into the smallest units you can.

It isn't a new idea. The advice to break work into small pieces is very old, and almost every book about getting things done includes it. Computer science even has a name for it: divide and conquer. I use the method often myself. Once a job is cut into small pieces, a goal that felt overwhelming turns into something I can finish today. Crossing the pieces off one at a time builds up small feelings of accomplishment, and I often end up working without noticing how hard it is. It also becomes easier to share the work among several people.

I once wrote, in an earlier series about my company, about breaking a large project into ninety-one small milestones. That piece was about my own experience. Today I want to look outward. Is this method really that good? Since when have people thought this way? And is it right to break things down, as the notebook says, into the smallest units you can?

## As far as possible, and as far as necessary

The most famous person to put "break it down" into writing is probably Descartes. In his 1637 *Discourse on Method* he set out four rules for conducting the mind well. The second is this:

> To divide each of the difficulties under examination into as many parts as possible, and as might be necessary for its adequate solution.

The French original uses the word *parcelles*, small parcels ([MacTutor](https://mathshistory.st-andrews.ac.uk/Extras/Descartes_Method/)). Rereading it, I noticed one way it differs from the sentence in my notebook. Descartes didn't stop at "as many parts as possible." Right after it he added "as might be necessary." He wasn't telling us to cut things up endlessly. He was telling us to cut until the pieces can be solved.

I'll come back to that difference at the end. First, let's look at what breaking work down has actually changed.

## The factory that made a pin in eighteen steps

In 1776 Adam Smith opened the first chapter of *The Wealth of Nations* with a pin factory. He divided the making of a single pin into about eighteen distinct operations. One man draws out the wire, another straightens it, a third cuts it, a fourth points it, a fifth grinds the top to receive the head. Making the head takes two or three operations of its own, and whitening the pins and putting them into paper are separate jobs too.

The small factory Smith saw had ten workers. Together they made about twelve pounds of pins a day. With more than 4,000 pins to a pound, that is upward of 48,000 pins a day, or 4,800 per person. If each of the same people had worked alone from start to finish, Smith wrote, "they certainly could not each of them have made twenty, perhaps not one pin in a day" ([Cambridge excerpt](https://www.cambridge.org/core/books/poverty-progress-and-population/two-kinds-of-capitalism-two-kinds-of-growth/C376DD9C098D880AD3E94D9274388DA9)). Even if we grant them twenty, that is a 240-fold difference.

A hundred and thirty-seven years later, the same idea took another big step in Highland Park, just north of Detroit. On October 7, 1913, Ford started its first moving assembly line, pulling car frames past the workers with a rope. Each worker stood in place and did one operation. After months of reworking the line and dividing the tasks still further, by early 1914 a Model T was coming off the line every 93 minutes. Under the old stationary method, where the car sat in one spot while it was built, each car had taken 12 hours and 30 minutes ([The Henry Ford](https://thehenryford.org/explore/blog/ford-methods-and-the-ford-shops)).

Up to this point, breaking work down looks like magic. Pins became 240 times faster, cars eight times faster. But both stories have a less familiar side. I'll get to it later.

## Computers ask how to divide all over again

Programmers break work down for a living. You split a big problem into smaller ones, solve the small ones, and combine the answers. Most basic algorithms, from sorting and searching to matrix multiplication, have this shape. Computer science added two more questions: where should you cut, and who reconnects the pieces, and how?

David Parnas raised the first question in 1972. In a short paper, "On the Criteria to Be Used in Decomposing Systems into Modules," he took a single program, one that rotates every line of a document word by word and lists all the rotations in alphabetical order, and divided it two ways ([original](https://cse.sc.edu/~mgv/csce330f24/ParnasCriteria.pdf)). The first is the way anyone would think of: follow the order of processing and split it into input, shifting, sorting and output. The second hides each "design decision that might change later" inside its own module. How the data is stored in memory, when the sorting happens: each decision like that is locked inside one module.

Both programs run equally well. The difference shows up when something has to change. In the first design, changing how the data is stored means rewriting almost every module. In the second, you change only the one module that hides that decision. Pieces of the same size can lead to completely different outcomes depending on where you draw the cut lines. Parnas's paper is still assigned in software design courses.

Frederick Brooks raised the second question in his 1975 book *The Mythical Man-Month*. Brooks, who had led the development of a huge operating system at IBM, wrote: "Adding manpower to a late software project makes it later." His explanation goes like this. People and months are interchangeable only when the work is divided so that the people don't need to talk to each other. If the pieces interlock, each person has to coordinate with everyone else, and the number of channels grows as n(n−1)/2 ([Wikipedia](https://en.wikipedia.org/wiki/Brooks%27s_law)).

![A diagram of how communication channels grow with the number of people. Three people are connected by 3 lines, five by 10, ten by 45 and fifteen by 105. On the right, for contrast, the same fifteen people are not connected at all, each holding a single piece of their own.](/uploads/6a953d0375678a51.webp =720)

Five people have 10 channels, ten have 45, and fifteen have 105. Triple the people and the relationships to keep in sync grow tenfold. Brooks summed it up in one line: "The bearing of a child takes nine months, no matter how many women are assigned."

What happens, then, if the pieces aren't connected at all? There is an experiment that answers that.

## 900,000 galaxies, split among 100,000 people

In 2007 a group of astronomers at Oxford had a problem. They had images of about 900,000 galaxies from the Sloan Digital Sky Survey (SDSS), and someone had to decide by eye whether each one was a spiral or an elliptical. Computers couldn't yet do the job reliably. Kevin Schawinski, the graduate student who first took on the task, classified 50,000 galaxies on his own in a week and said, "it was mind-numbing." At that pace one person would need more than four months, and the result would rest on a single pair of eyes.

So they cut the work down to its smallest unit. Look at one galaxy image and press one of a few buttons: spiral, elliptical, or not sure. You don't need to know astronomy; a few minutes of instructions is enough. They opened it on the web as Galaxy Zoo. Within twelve hours of the July 11, 2007 launch, classifications were arriving at 20,000 an hour. After ten days there were 8 million. In about 175 days, more than 100,000 people had made more than 40 million classifications, an average of 38 per galaxy. The team had hoped for maybe 20,000 to 30,000 volunteers ([Wikipedia](https://en.wikipedia.org/wiki/Galaxy_Zoo), [Lintott et al.](https://arxiv.org/pdf/1303.7118)).

Brooks's law doesn't apply here. The person classifying one galaxy never has to talk to anyone else. The pieces weren't only small; they stood apart from each other. Because each galaxy was shown to many people who checked one another, the results could be trusted without an expert in the loop. By 2017 the project had produced 60 papers. Volunteers even posted about some odd little round green dots in the images on the forum, and those turned out to be a new kind of galaxy, now called "Green Peas."

So when I said at the start that small pieces make it easier to share work, that statement comes with a condition. Cutting the work small isn't enough. The pieces also must not be waiting on each other.

## Why do small pieces feel easier?

Now to the notebook's second reason: small pieces make work feel less hard. Psychology has a good number of experiments that test exactly this.

In 1981 Albert Bandura and Dale Schunk brought together elementary school children who were very poor at arithmetic and disliked it. The children worked through practice booklets on their own, and the researchers split them into three groups. One group got near goals, like "this many pages in this session." Another got a distant goal: "all of it by the end of the sessions." The last group got no goal at all.

Only the children with near goals changed noticeably. They moved quickly, actually learned the operations, and came to believe more strongly that they could do it (self-efficacy). They even grew interested in the arithmetic they had disliked at first. The children with the distant goal did no better than the ones with no goal ([original](https://uploads-ssl.webflow.com/59faaf5b01b9500001e95457/5bc552d85141987915dab842_Bandura%20&%20Schunk,%201981.pdf)). The total goal was the same. The only difference was how close the goal had been brought.

Three years later the organizational psychologist Karl Weick extended the same idea to social problems. In his 1984 paper "Small Wins," he argued that defining problems like poverty or crime on a massive scale actually leaves people unable to act. It exceeds the limits of what people can reason about, and it pushes anxiety too high. So, he said, recast the issue as a mere problem and win concrete, complete outcomes of moderate importance, one at a time. When those small wins pile up into a pattern, they attract allies and deter opponents ([original](https://homepages.se.edu/cvonbergen/files/2013/01/Small-Wins_Redefining-the-Scale-of-Social-Problems.pdf)). The phrase "small wins," so common today, comes from this paper.

Breaking work down has one more, less known effect: you see the true size of the work. When people estimate how long a task will take as a whole, they usually guess too short. Psychologists call this the planning fallacy. In 2004 Justin Kruger and Matt Evans published a paper titled "If you don't want to be late, enumerate." When people wrote out the same task step by step, their estimates got longer and came closer to the time it actually took ([Wikipedia](https://en.wikipedia.org/wiki/Planning_fallacy)). A plan pictured as a whole leaves out a lot of steps.

I find this result rather comforting. Breaking work down doesn't make it smaller. If anything, you learn it's bigger than you thought. But something precisely big is less frightening than something vaguely big.

Among the pieces, the smallest unit of all is probably "the one next action you can do right now." Peter Gollwitzer has spent years studying goals written in the form "If situation X arises, I will do Y." Instead of "I'll exercise," you write "When I get home and put my bag down at the door, I'll put on my running shoes." In 2006 he and Paschal Sheeran pooled 94 tests and found that people who planned this way had a medium-to-large advantage in reaching their goals (effect size d = 0.65) ([Frontiers](https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2021.565202/pdf)).

## The card with two stamps already on it

The most entertaining experiment on why small pieces feel easier comes from a coffee card.

In 2006 Ran Kivetz of Columbia University and his colleagues handed out "buy ten coffees, get one free" cards at a university café and recorded when people came in. The closer customers got to the free coffee, the more often they came. The goal-gradient hypothesis, Clark Hull's 1930s observation that rats run faster as they near the goal, showed up intact in people's coffee habits.

The next experiment is even more interesting. One group got an empty card with ten spaces. Another got a card with twelve spaces and two stamps already on it. Either way, you had to buy ten more coffees to get the free one. Yet the group whose card had two stamps filled its ten spaces faster ([original](https://business.columbia.edu/sites/default/files-efs/pubfiles/1200/goalgradient.pdf)). The remaining distance was the same; the feeling of having already come a little way changed the speed.

In 2012 Ayelet Fishbach and Minjung Koo of the University of Chicago took this a step further. Early in a task, looking at what you've done raised motivation more; near the end, looking at what's left did. Either way, the advice is to look at the small number. A small number feels like it moved a lot even when it moves a little. They called it the small-area hypothesis ([IDEAS](https://ideas.repec.org/a/oup/jconrs/doi10.1086-663827.html)).

If you've cut your work into small pieces, you can always look at the small side. At the start, "one of ten done"; near the end, "one left." Much of why small pieces feel easier may lie less in the work itself than in how the progress shows.

## The warnings from the same people

Having written all that, I'm tempted to just agree with my notebook. But there's still the other side of the two stories I put off earlier.

Adam Smith praised the division of labor in Book I of *The Wealth of Nations*, but in Book V of the same work he wrote something that sounds like the opposite. A man who spends his whole life repeating a few simple operations has no occasion to think up ways of removing difficulties, because such difficulties never come up. So he loses the habit of thinking and "generally becomes as stupid and ignorant as it is possible for a human creature to become." That is why Smith argued that the state should take responsibility for educating workers ([Pitt excerpt](https://sites.pitt.edu/~syd/ASIND.html)). The father of the division of labor wrote down its shadow in the same book.

At Ford's plant, that shadow showed up in numbers. As the work grew monotonous, people started quitting. In 1913 Ford's labor turnover was about 370 percent. To keep a workforce of 13,000 to 14,000, the company had to hire more than 50,000 people in a year ([The Henry Ford](https://www.thehenryford.org/collections/explore/articles/fords-five-dollar-day), [TribLive](https://archive.triblive.com/opinion/featured-commentary/henry-ford-didnt-pay-5-a-day-just-to-be-nice/)). The famous "five-dollar day" Ford announced in January 1914 is often told as a generous move to let workers buy the cars they built, but many historians read it as mostly a calculation to stop the turnover. And the five dollars was a base wage of about $2.30 plus about $2.70 in profit sharing, paid only to those who met the company's conditions.

Psychology has a warning too. In 2008 On Amir and Dan Ariely reported experiments showing that intermediate markers don't always help. When you don't know how far you are from the goal, markers reduce the uncertainty, and both performance and liking for the task go up. But when the distance is already clear, the markers breed complacency. People rest in the relief of having finished one piece, take their eyes off the final goal, and both performance and interest drop ([Duke](https://scholars.duke.edu/publication/810355)). The paper is titled "Resting on Laurels."

Put together, it looks like this.

| When you cut too fine | Evidence | In one line |
|---|---|---|
| People shrink | Smith, *The Wealth of Nations*, Book V (1776) | Repeat the same piece forever and there's nothing left to think about |
| People leave | Ford, Highland Park (1913) | Efficiency went up, and turnover reached 370% |
| Connecting costs grow | Brooks, *The Mythical Man-Month* (1975) | The more you divide interlocking pieces, the more communication grows |
| The cuts go in the wrong place | Parnas (1972) | Cut by processing order, and every change shakes everything |
| People stop halfway | Amir & Ariely (2008) | When the distance is clear, intermediate markers breed complacency |
| Some work doesn't benefit | Kruger & Evans (2004) | For tasks with only a few steps, listing them barely changes the estimate |

![A one-line timeline from 1637 to 2012 of ideas and cases about breaking work down. On the upper side are cases about how work is done: Descartes, Smith, Ford, Parnas, Brooks and Galaxy Zoo. On the lower side are psychology studies: Bandura and Schunk, Weick, Kruger and Evans, Gollwitzer and Sheeran, Kivetz, Amir and Ariely, and Koo and Fishbach.](/uploads/0f9ac7ee442c083d.webp =720)

## How small is the smallest unit?

Back to the notebook. "The smallest units you can." Now I want to add a phrase to it, the way Descartes did: as far as possible, and as far as necessary.

Laying today's cases on top of one another, a good piece seems to meet three conditions.

First, it has to be small enough to start right now. The "this many pages in this session" Bandura's children got, Gollwitzer's "put the bag down, then the running shoes," Galaxy Zoo's "one image, one button" are all like that. They are sized so you never have to wonder what to do first.

Second, it has to be something you can finish on your own. A piece that waits on other pieces, or needs constant coordination with other people, isn't small no matter how small it looks. The communication cost Brooks warned about comes from the threads that tie one piece to the next. As Parnas showed, what you cut along matters before how small you cut.

Third, it has to be big enough that you don't forget why you're doing it. Smith's worker only ground pin heads and never had reason to think about what they were part of. Ford's workers left. In Amir and Ariely's experiments, too, the markers backfired when people took their eyes off the final goal. Even while you're holding a piece, you should be able to see where it fits.

Seen this way, the "smallest unit" isn't the size that can't be divided any further. It's the size you can start right away, finish on your own, and, once it's done, see which part of the whole picture it filled in. Cut any smaller, and managing the pieces becomes bigger than the work.

I wrote this piece the same way. I took one line from the notebook, gathered cases, sorted them into ones to use and ones to drop, and filled in the sections one by one. By the time I finished, it was much longer than I'd first thought. Kruger and Evans were right.

*A Page for Today* is a series in which I open my idea notebook to a random page and lay that day's thoughts on the line I find there. Today's line was about breaking things down, and it helped me understand a little better why this series goes one page at a time.