Hacking the Marshmallow Challenge

A group of 4 tries the marshmallow challenge (graphic by Tom Wujec, http://marshmallowchallenge.com)

I use the Marshmallow Challenge at the beginning of every year. I used to do it the way it’s proposed by its originator, Tom Wujec, and widely written about. (TL;DR: you have to build a tower out of spaghetti, string, and tape, with a marshmallow on top.  Tallest tower wins.) My goals are to :

  • help students get to know each other and me
  • start a conversation about how we learn, and the value of growth mindset
  • invite students to think critically about how “mistakes” and “wrong answers” have been treated in their experience of schooling
  • encourage students to see all of our learning as emergent and imperfect

In previous years, that did sort of happen.  But the other thing that happened was that one group would build the tallest tower, and most of the other groups worked really hard, got really excited, worked frantically for 20 min, and then had their tower collapse at the last second.

We would then watch the video, where Wujec explains that that’s what happens for most people, because they’re so afraid of failure that they avoid putting the marshmallow on until they’re forced to by the buzzer.  And marshmallows are light, but we assume they’re weightless.  Our own wrongness about marshmallows combined with our ingrained fear of being humiliated for mistakes conspire to prevent us from succeeding. Gotcha!

Finally last year I realized something.  (Because sometimes, even when you prototype and you see the same thing go wrong year after year, you keep doing it…).

There is a name for this.  It’s not “discovery learning”.  It’s not “constructivism”.  It’s expose and shame.

This year I made one tiny change.

Modify the Instruction

I gave the students exactly the same materials, same grouping structure, same 20-min timer on the projector, same everything.  It was identical to other years.  Except for one thing.  I gave them this instruction.

“How fast can you get the marshmallow off the table?”

Instantly every group would plunk down a one-noodle-thick layer of spaghetti, put the marshmallow on it (1mm off the table or so), smile smugly like they’d gotten one over on me, and put their hands up.

I would smile genuinely back.  Congratulate them on understanding the challenge.  Then say, write down the time and your current height.  You’ve got 19 minutes and 40s left; can you build another one that’s higher?

Then they’d set out building a tower 1 inch high.  Then 12 inches.  Then 18.

EVERY. SINGLE. GROUP. had not one but 3-5 working designs for standing towers that supported marshmallows.  Some of them were shorter but sturdier.  Some were taller but fragile.  Some used fewer materials. All of them had pros and cons. They got excited about creating their own ideas, but there was also no downside to looking at other teams’ structures, so they wandered and looked and incorporated each other’s ideas. After about 15 minutes, they were getting pretty satisfied that they had exhausted their options, explored the problem space to their content.   They’d look at me like, ok, we’ve learned what we can from this, surely it’s time to move on?

Shifting the Reveal

This shifted the location of the reveal.  Instead of the “big surprise” being their own failure when the timer went off, the big surprise was my explanation.  I would tell them, as they sat there in a room with over a dozen intricate and creative structures, that the vast majority of people who try this exercise succeed in building ZERO standing structures in the ENTIRE TWENTY MINUTES.

They gasp.

They can’t believe it.

It wasn’t that hard, they say.  Maybe they start to go down the road of contempt — most people are stupid maybe.  That’s a good opportunity for me to redirect with our classroom practice of compassion.  “Why might a reasonable person do that?”  Then a good conversation happens.  Maybe they were under too much pressure.  Maybe they were nervous about what people around them would think.  Our conversation about barriers to learning has now shifted from competitive to collaborative, from ego-driven to compassion-driven.  We have now moved the discussion of why people struggle in school safely away from the humiliation of a “failure” that they themselves just went through in public.  We can all talk about those obstacles and humiliations from a safe distance, with compassion for others, and insight from our own lives.  But in the end they’re still dying to know what happened.

So I show the video.

They get excited about how smart pre-schoolers are.  They get some satisfaction about how it shows that people in positions of power are dependent on the work and skills of the people they employ.  They get excited about the specialized skills of architects and engineers, since they are joining the world of engineering themselves.

At the end, I ask them, what does Tom Wujec think makes this so hard for people, and what helps? Having seen the video, have you changed what you think about why this is hard for so many people, and what helps?  Great discussions happen again, about the value of multiple low-stakes attempts, all of which is going to motivate my introduction of Standards-Based Grading later in the week.

Finally I ask, did you notice what was different between the challenge you did, and the challenge done by the people in the video?

Again, they look astonished.

No, they say!  It was identical!

What was the same? I ask.

They tell me.  The time.  The number of sticks of spaghetti.  The number of people in the group.  The rules.  It was the same.

No, I say.  It was almost the same.

The difference, the entire difference between a class full of people who built so many standing structures that they got bored, and a room full of people who built none, the only difference, was how I asked you to do it.  What do you remember about the instruction Tom Wujec gave?  What do you remember about the instruction I gave?


That, I tell them, is my specialized skill.

That is my job.

It is my job to learn the tiny things that make huge differences in people’s ability to learn.

And I am good at it.

Is it ego-driven?  Yes, of course.  But it is also an important way that they meet me. I am establishing a ground of trust. I am going to ask them to do very difficult things.  Things they have been told they can’t do.  It will involve them questioning not just their sense of themselves but their basic assumptions of right and wrong, their understandings of the underlying systems of our society.  It’s going to be scary and unsettling.

If I expect them to even consider coming with me in this process, I absolutely must make sure they have a reason to think I am good at what I do. That the apparently incomprehensible things I ask them to do actually work. That when I say they are capable of things that seem impossible, there’s a chance it might be true. I owe them proof. The books and research I’ve read, the fancy names of pedagogical techniques, will not and should not impress them.  Their skepticism is valid and honourable.  I owe them an evidence-based reason to consider believing me.

Their many towers, and this first hour we have spent together free from shame, is that beginning.

Update July 12: Extension Problems

“Expose and Shame” is a tricky and deeply tempting form of teacher thinking.  Why might reasonable teachers engage in it?  Is there a context where it is useful? Do you think there are places within this interaction (above) where I am still doing it?  If so, what modifications would you propose and why? What generative questions can we ask ourselves about it?  What intrigues you about your own or others’ use of things that look like expose-and-shame? What happens if we “make contact with important aspects of the phenomena, [and] press upon the coherence of [the] explanations with respect to the evidence, arguments, and tools that [we] currently have at [our] disposal“?

Update July 12: Top Tweets

Brian Frank tweets that Elicit-Confront-Resolve (criticized by me and elsewhere as leaning dangerously toward Expose-And-Shame) isn’t inherently the problem.  As usual, he is ingeniously modelling the shift from “Is this thing good or bad” to “what question is this a good answer to…” or, as my students and I ask, “why might a reasonable person do this?”

How students & teachers frame the activity is more important than the structure of the activity. Some structures can help or hurt with the right framing, but are not deterministic. Here’s an example. (Brian Frank)

I hope people will follow and contribute to Lucas Walker’s tweet exploring the idea of “Expose-and-Shame” in inquiry-based physics — especially Brian’s link to the original posts about “fostering intrigue” where I learned this use of the term, and inquires about why some questions seem to be more “generative” of intrigue, while others seem to mostly expose ignorance.  The concept of what makes some questions generative haunts my dreams.

I’d like to start a parallel thread – in inquiry-based physics, where are some other points that the harmful “expose-and-shame” dynamic is likely to surface? (Lucas Walker)


  1. Hi Mylène. Long time since I’ve dropped in. I had not put my finger on it previously, but I agree with your assessment that it is expose and shame. Or show and tell. So few groups engage in the desired iterative process that it’s all about wagging our fingers at them afterward for what they should have done instead of letting them experience how much they could accomplish if they engaged in the iterative behaviour in the first place. Thanks for this modification, I will certainly give it a try and also consider how it can inform other activities.

    • “wagging our fingers instead of letting them experience how much they could accomplish” — that’s it exactly. No need to miss that opportunity. Will be interested to hear how it goes for you — hope you’ll share.

  2. I passed the URL for this post on to the engineering faculty at UCSC. One of them decided to try out the marshmallow challenge with a high-school summer program that he is working with. So your suggestions are having an effect.

  3. Can I ask if the kids still only get the same supplies and have to make multiple prototypes? Or does each prototype get it’s own 20 spaghetti noodles, marshmallow etc?

    • Well, it’s your marshmallow challenge now, so I think you can run it however it best meets your and your students’ goals. If the goal is to help students meet each other, have a conversation about how we learn, and encourage students to see *all* learning (even after it’s “successful”) as emergent, then what does that best? I originally didn’t address this, and students just helped themselves to more supplies. That was both a sensible way to engage the activity, and a chance for them to practice critiquing assumptions of authority. But in hindsight, I’d be curious to simply put the question to the class and see what they come up with. Good luck!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s