Skip to content
Spark & Whetstonethe grown-up commons

← Thinking in Loops

Thinking in Loops

Four short lessons on how things build up and run down: stocks and flows, loops that snowball or settle, why delays make us overshoot, and why every model is an argument.

Spark & Whetstone · sparkwhetstone.org/learn/thinking-in-loops · CC BY-NC-SA 4.0. Answers are at the end, so you can try each question first.

Lesson 1. Stocks and flows

A bathtub is the whole idea in one place. The water in the tub is a stock: an amount you could measure at any moment. The tap and the drain are flows: rates that change the stock over time.

A stock can only change through its flows. If the tap runs faster than the drain, the water rises; if slower, it falls. If they are equal, the level holds steady even though water is moving the whole time.

Once you look for them, stocks are everywhere: money in an account, fish in a lake, people waiting in a queue, trust between two neighbours. Each one fills and empties at rates, and each one takes time to change.

Question. A reservoir is half full. Rain adds water faster than the town draws it out, but the town draws more each year. What happens to the level this year?

  1. It falls, because the town is drawing more.
  2. It rises, because more is coming in than going out right now.
  3. It stays the same until the two rates are equal.

Say it back. In your own words: name a stock in your own week, and the flows that fill it and empty it.

________________________________________________

________________________________________________

Lesson 2. Loops that snowball or settle

A loop is a stock that changes its own flow. Some loops feed themselves: the more there is, the faster it grows. Interest on savings works this way, and so does a rumour, or weeds in a garden. These loops snowball.

Other loops push back. A thermostat turns the heat down as the room gets warmer, so the room heads toward a set temperature and stays near it. Hunger and eating work the same way. These loops settle.

Most real things have both kinds pulling at once. A town grows because people bring jobs, and slows because homes fill up. Which loop is winning tells you more than any single number does.

Question. A new café gets busier the more people see a queue outside, until the queue gets so long that people walk past. Which describes it?

  1. Only a snowballing loop.
  2. Only a settling loop.
  3. A snowballing loop at first, then a settling loop takes over.

Say it back. In your own words: describe one loop that snowballs and one that settles, from your own life or work.

________________________________________________

________________________________________________

Lesson 3. Delays

In a hotel shower you turn the tap, nothing seems to happen, so you turn it more. Then the hot water arrives all at once and you jump back. You turn it down, wait, feel nothing, turn it down more, and now it is cold.

Nothing was wrong with your judgement. You were steering by water that left the heater some seconds ago. A delay between acting and seeing the result makes almost anyone overshoot, then overcorrect.

The fix is rarely to react harder. It is to act a little, then wait as long as the delay before acting again. The same holds for adjusting a budget, a medication with a doctor, or how much to order for a shop.

Question. You water a plant and it still droops an hour later. What does the shower suggest you do?

  1. Water it again now, and more this time.
  2. Wait long enough for the first watering to show, then decide.
  3. Stop watering it altogether.

Say it back. In your own words: where in your own life does a delay make it easy to overdo a correction?

________________________________________________

________________________________________________

Lesson 4. Every model is an argument

In 1969 Jay Forrester published Urban Dynamics, a computer model of a city. It concluded that building low-cost housing could make a city worse off, by drawing in people faster than jobs. The book was controversial then and is argued about now.

Twenty years later the game SimCity drew on the same kind of thinking. Its cities obey rules someone chose: how far people will travel to work, what makes them move away. Play enough and you are learning those rules, not a fact about real cities.

That is true of every model, including the small ones in this course. A model is a careful argument: if these premises hold, then this follows. The useful question is never only "what does it say?" but "what would have to be true for it to be right?"

Question. A model says building more homes will raise unemployment in a town. What is the most useful first question to ask?

  1. Was the model made by a computer?
  2. Which assumption about people and homes leads to that answer?
  3. Has anyone ever built homes in a town?

Say it back. In your own words: pick a forecast you heard recently and name one premise it depends on.

________________________________________________

________________________________________________

At the table: The lemonade line

Players: 4 (or 2 people taking two seats each). You'll need: paper slips, a pencil each, and coins or buttons for crates.

  1. Sit in a row: the stand, the shop, the warehouse and the farm. Each seat starts with 12 crates.
  2. Each turn the stand draws a customer card and sells that many cups. Every seat ships what it is asked for, if it has it.
  3. Each seat writes an order on a slip and passes it up the line. Slips and crates both take two turns to arrive.
  4. After a few turns the stand's customers go from 4 to 8 a turn and stay there. Keep playing for twenty turns.
  5. Lay out every seat's orders in a row and compare. Talk about why the farm's swings were so much bigger than the stand's.

Seat sheets and a run-through: sparkwhetstone.org/lemonade-line

Answers

  1. It rises, because more is coming in than going out right now. — The level rises while inflow beats outflow, even if the outflow is growing. It will only start to fall once the outflow passes the inflow.
  2. A snowballing loop at first, then a settling loop takes over. — The visible queue feeds itself at first. Past a point, the wait pushes people away and the loop that pushes back wins. Both were there all along.
  3. Wait long enough for the first watering to show, then decide. — The first watering may not have reached the leaves yet. Acting again before it shows is how you end up overwatering — the same swing as the shower.
  4. Which assumption about people and homes leads to that answer? — The answer follows from the premises — here, how strongly an empty home draws people in. Change that premise and the conclusion changes. That is where to argue.