# The Marshmallow Challenge — Run It Live with Your Class

Source: https://decka.dev/classroom/business/marshmallow-challenge

Teams of four build the tallest free-standing spaghetti tower they can against the clock, with the marshmallow on top. Every time a team tests, a teammate logs on a phone whether it stood or fell, so the room's own log shows when each team first found out.

## What it teaches

Test the thing that can sink you first, while there is still time to change the design. Students see it in their own room's evidence: every tower stands in the order its team first put the marshmallow on top, and the test log under it shows how much of the build was left to learn from what happened. Each student also sets their own prediction against what their team built.

## Run the marshmallow challenge with every test logged

The marshmallow challenge gives each team twenty sticks of dry spaghetti, a metre of tape, a metre of string and one marshmallow. The team builds the tallest tower it can that stands on its own, with the whole marshmallow at the very top, measured from the table top to the top of the marshmallow. The idea it leaves behind is a habit: test the thing that can sink you first, while there is still time to change the design.

Students pick their table on their phones, which makes the teams, and each predicts how tall their team's tower will be; the predictions stay sealed. The build runs on a clock, eighteen minutes by default. Each time a team puts the marshmallow on top and lets go, any teammate logs on a phone whether it stood or fell, and the server stamps the minute it arrived. When time is called, each team enters one height, or 0 for a tower that is not standing, on a measuring clock, three minutes by default, that then holds the records closed until you move on. You press Next for every other step; allow about forty minutes with the debrief.

While teams build, the big screen shows only the kit and a stopwatch with the time left: no count of tests and no sign of another table's progress, so nothing nudges a team to test early. After measuring, every team's tower stands at its measured height, left to right in the order teams first tested, with those that never tested set apart. Then each team's tests climb under its tower, one mark each, stood or fell. A cut at half time groups the towers whose teams tested by then, after it and never, and reads each group as how many of its measured towers were still standing. A last board sets every prediction beside the tower its team built.

The screen never names a table, ranks the towers or crowns the tallest; the towers on the tables speak for themselves. Each phone sees only its own team's column and its own prediction. Every sentence on the screen is chosen by what the room actually did: if this room's early testers did no better, the board says so, and the debrief asks what else separated the towers.

## Classroom debrief

When did your team first put the marshmallow on top? What did you learn from it, and how much time was left to use it?

If the teams that tested late or never fell more often: what do the fallen towers' test logs have in common? If not, early testing didn't separate the towers in this room. What else did the tallest towers share?

If most predictions ran taller than the towers built: why do plans run optimistic, even for a tower of twenty sticks? If not, what did this room know that most planners don't?

If you had the whole build again, what would you do in its first two minutes?

What is the marshmallow in a project you're doing now: the heavy assumption you haven't tested yet?

## Marshmallow challenge questions, answered

### What do I need besides phones?

Per team of four: 20 sticks of dry spaghetti, 1 m of tape, 1 m of string and 1 marshmallow. Bring scissors and a tape measure, and clear the tables.

### How do teams form?

Each student picks the number of the table they sit at on their phone, and everyone at one table is one team: the team logs its tests together and enters one height. The lesson offers eight tables by default; set anywhere from two to twenty before you run it.

### Are teams named or ranked on the projector?

No. The projector never names a table, ranks the towers or crowns the tallest. Each phone sees only its own team's tests and height, and its own prediction; the towers on the tables speak for themselves.

### What if no team tests early, or every tower falls?

The boards say so. With no test before time, the test log stays empty and the towers stand in table order as one group. With every tower down, the skyline is a row of heaps reading 0 standing. No sentence claims a pattern the room did not produce, and that is worth discussing too.

### Can I change the timings or the number of tables?

Yes. Before you run it, set the build's clock (eighteen minutes by default), the measuring clock (three) and the number of tables (eight, up to twenty). The half-time cut and the test log's minutes follow the build's clock, and the words on the screen and the phones are yours to edit.

### What does the lesson report keep?

Counts and the towers' heights, never which table built which: how many stood and fell, how many tests teams ran and when they first tested, each half-time group's counts, and how far the predictions ran from what was built. It never keeps a student's name or one person's prediction.

### How long does it take?

Plan on about forty minutes: a few minutes to form tables and predict, the timed build, a short measurement, then a quarter of an hour to read the boards and discuss them. The build and the measuring run on clocks; you move every other step on when the room is ready.

## Run the lesson

Present the boards in PowerPoint, Google Slides and Decka. Students scan a QR code or open the session link to join from their phone browsers.
