Models · Model 003 · From Russell's research
Litter Begets Litter
Every flyer on the ground tells the next driver what is normal here.
Three of Russell's students put flyers under the windshield wipers of cars at two southern California grocery stores and watched 223 drivers come back to them. Near an interstate, motels, liquor stores, and boarded-up buildings, 52.0 percent dropped the flyer on the ground. Among well-kept houses and lawns, 31.4 percent did. Here is a lot like theirs. You choose the neighborhood and how much a worker picks up.
01 · The model
One lot, two neighborhoods
Every car that parks gets a gold flyer under its wiper. Rust marks are flyers on the ground. The steel dot is a worker who comes out at the top of each hour to pick litter up. Press Run, then change the neighborhood or the upkeep as the days pass.
Day 1 · 7:00 am
A new lot. Press Run.
The setting
Neighborhood
Houses with lawns and street trees on every side, and a park next door.
For the first 4 minutes of every hour, a worker picks up litter.
A sweep picks up every piece at once. A mess scatters 300 flyers.
Litter on the ground
Pieces, at the end of each hour
Drivers dropping the flyer
Each hourLast 100 driversThe study: 31.4% and 52.0%
02 · The test
The tipping point
Each point is a week on the lot at one level of upkeep, run on 10 lots, each with its own drivers. It shows how many drivers dropped the flyer on the seventh day, and how much litter lay on the ground.
Well tendedLess tendedDrivers ignore the groundRussell's study
440 weeks in all: 10 lots for every minute of upkeep from 0 to 10, in each of the four settings.
The numbers
| Upkeep, min an hour | Well tended | Less tended | Drivers ignore the ground | |||
|---|---|---|---|---|---|---|
| Dropped it | Litter | Dropped it | Litter | Well tended | Less tended | |
| 0 | 57% | 673 | 69% | 801 | 32% | 43% |
| 1 | 59% | 580 | 69% | 696 | 33% | 42% |
| 2 | 53% | 385 | 66% | 525 | 32% | 43% |
| 3 | 43% | 170 | 58% | 329 | 32% | 44% |
| 4 | 31% | 17 | 52% | 147 | 32% | 43% |
| 5 | 32% | 13 | 42% | 25 | 32% | 42% |
| 6 | 30% | 10 | 41% | 17 | 33% | 43% |
| 7 | 30% | 10 | 41% | 14 | 33% | 45% |
| 8 | 31% | 10 | 42% | 14 | 32% | 43% |
| 9 | 30% | 9 | 41% | 13 | 32% | 43% |
| 10 | 31% | 9 | 40% | 12 | 33% | 44% |
At four minutes of upkeep an hour, the model matches Russell's study: about 31 percent of drivers drop the flyer at the well-tended lot and about 52 percent at the less tended one. The model was tuned to match those two numbers, so the match proves nothing on its own. What the model adds is one way such a gap could arise. When drivers ignore the ground, the neighborhood alone opens a gap of about 10 points, half of what the study found. The rest comes from the litter. In the less tended lot more drivers drop the flyer, the worker falls behind, and the pieces left on the ground tell the next drivers what is normal.
Take one minute away and the well-tended lot tips too. At three minutes an hour about 43 percent of drivers drop the flyer, and at two, more than half. Above four minutes, more upkeep changes almost nothing, because the worker already keeps up. The less tended lot tips a minute sooner, below five minutes an hour. When drivers ignore the ground, both lines stay flat: the litter piles up, but no one reads it.
None of the drivers changed. Every run draws them from the same mix of habits. As Russell wrote of the drivers at the two stores, “They just did what the place seemed to tell them to do.”
03 · How it works
Drivers, flyers, and a worker
The lot. A grocery store has 132 stalls in three double rows and is open from 7 in the morning to 11 at night. About 600 cars come in a day, more at lunch and after work, and drivers prefer the stalls nearest the door. A typical shopper is inside for about 20 minutes, some for much longer.
The drivers. Every driver comes back to a flyer under the wiper and either takes it along or drops it. Each has a leaning of their own, set at random: some would rarely drop it and some usually would. The neighborhood nudges everyone's odds. Then the driver looks around the car. Two pieces of litter within seven meters make dropping the flyer more likely, and each doubling of the litter in view adds the same push again, up to about eight pieces. A single piece does the opposite: alone on clean pavement, it looks out of place.
That last effect comes from one of Robert Cialdini's 1990 studies. A large replication in 2021 did not find it, though it did confirm that people litter more where litter already lies. It is the least certain part of the model.
The worker and the wind. For the first few minutes of each hour, a worker walks to the nearest piece of litter, picks it up, and heads for the next one: about 12 seconds a piece, at a walking pace. The wind carries off 2 percent of the litter each hour, night and day.
- Lot
- 132 stalls in three double rows, 71 by 61 meters
- Hours
- 7 am to 11 pm. The night passes in a blink, and the wind keeps blowing.
- Drivers
- About 600 a day, each deciding once, on coming back to the car
- View
- Seven meters around the car
- Worker
- The nearest piece first, about 12 seconds each
- Code
- About 250 lines of JavaScript, running in this page
Well tendedLess tended
04 · What to try
Six experiments
Each button sets up the lot and starts it running. Every lot has its own drivers, so run each one a few times.
-
Move the store
Same lot, same worker, same four minutes an hour, in the less tended neighborhood. More drivers drop the flyer and the worker falls behind. By the end of the week about 150 pieces lie on the ground, and about half the drivers are dropping theirs.
-
Cut a minute
The well-tended lot with three minutes of upkeep an hour. Each day the worker leaves a little more behind, and by the end of the week about 43 percent of drivers drop the flyer, up from 31.
-
Make a mess
One bad night leaves 300 flyers across the well-tended lot. At four minutes an hour it takes most of a week to recover. Set the upkeep to six and make a mess again: in about two days the lot is clean.
-
Sweep and wait
Four days at two minutes an hour have left the lot littered, and it gets a clean sweep at opening. By closing time almost half the litter is back, and four days later the lot is as littered as it was the morning of the sweep.
-
Stop reading the ground
The less tended lot with no upkeep at all, and drivers who never look down. About 500 pieces pile up by the end of the week, yet the share dropping the flyer holds near 43 percent. In this model, litter begets litter only because drivers read it.
-
Put out cans
Eight trash cans in the less tended lot, at four minutes an hour. Drivers parked near a can drop the flyer less often, and the share falls by about five points. The lot still collects litter. Cans help, but they don't replace the worker.
Sources
Reading and code
- The studyRussell Weaver, Littering in context(s): Using a quasi-natural experiment to explore geographic influences on antisocial behavior, Applied Geography 57, 2015
- In his wordsRussell Weaver, The Tended Commons Within, Commoning Consciousness Sunday letter, October 11, 2026
- Norms in viewRobert B. Cialdini, Raymond R. Reno, and Carl A. Kallgren, A Focus Theory of Normative Conduct: Recycling the Concept of Norms to Reduce Littering in Public Places, Journal of Personality and Social Psychology 58(6): 1015–1026, 1990. Drivers dropped a handbill from their windshield far more often when the parking garage was already littered.
- The replicationMagnus Bergquist and others, Replicating the focus theory of normative conduct as tested by Cialdini et al. (1990), Journal of Environmental Psychology 74: 101573, 2021. People littered less in clean places than in littered ones, but a single piece did not lower littering.
- Disorder spreadsKees Keizer, Siegwart Lindenberg, and Linda Steg, The Spreading of Disorder, Science 322: 1681–1685, 2008. With flyers on bicycle handlebars, 33 percent of people littered beside a clean wall and 69 percent beside one covered in graffiti.
- ContextP. Wesley Schultz and others, Littering in Context: Personal and Environmental Predictors of Littering Behavior, Environment and Behavior 45(1): 35–59, 2013. Observations of 9,757 people at 130 places across the United States: litter already on the ground and the distance to a trash can both predicted littering.
This is a teaching model, not a forecast. It was tuned so that at four minutes of upkeep an hour it matches the two shares in Russell's study. Its other numbers, from how far drivers look to how fast the wind clears the lot, were chosen to be plausible, not measured, so the tipping points in the test reflect those choices. The code is plain JavaScript and runs in this page.
Russell's research on littering and neighborhood upkeep is one of the strands in his book in progress, Commoning Consciousness.