In 2015 I worked for Expii, a crowdsourced educational media startup.

At Expii, math and science enthusiasts write multimedia lessons with images, videos, and interactive quizzes. Because they're written by enthusiasts rather than “Educational Professionals,” we reasoned that their peers would find Expii's lessons especially relatable and insightful.

And for the most part, this was true! We had hundreds of people writing content, but we wanted tens of thousands. So, as we iterated on the acquisition funnel and content creation experience, we started to look for something that would draw a larger audience’s attention.

One thing we did was attend math competitions. At these events, enthusiasts gathered to work on problems and win prizes.

Sure, students attended competitions to pad the extracurriculars section of the Common App and win trophies. But I noticed that students were so enthusiastic about math that they would talk about particularly interesting problems during their downtime!

I realized that if Expii could produce the kind of content that excited the kind of people who attended math competitions, we could capture their attention and convert them into writers.

I wrote these questions to help guide my research:

- Why is math fun?
- Why do math problems go viral?
- How could we design viral problems?

At Expii, I was surrounded by world class mathematicians, people who published research papers in high school and medaled at the International Math Olympiad, while I had scraped by with a C in freshmen year Calculus. So the question was on my mind a lot: Why do people like this stuff?

So, I performed both informal and formal interviews, watched students do “fun” problems, and did problems myself. And I think it boils down to this: the Aha Moment.

The moment when a problem goes from seeming impossible to making complete sense is what makes math fun.

So, we should optimize for this moment as much as possible.

To answer this question, I compiled a set of problems that had gotten a lot of attention. Here’s an example: the Monty Hall problem.

I arrived at three common traits of viral problems, which I called the heuristics for writing viral math. They are:

SET IT UP WITH A STORY

The Monty Hall problem is most commonly explained in the context of a story. The listener is a competitor on a game show. The host says… The door opens… You choose to switch your door…

This format is convenient because it allows the reader to soak in the details of a problem (which can be elaborate) in a familiar and fun format.

USE AN EVOCATIVE MENTAL IMAGE

The Monty Hall problem is memorable because the details are so odd. A 60s game show? A car? A goat? The setting is striking, and it makes the problem stick out for readers.

END WITH A COUNTERINTUITIVE RESULT

Most people get the Monty Hall problem wrong. The conclusion—that you can increase your chances of winning by switching doors—seems to violate basic probability.

This means that once you get it, and really understand why switching doors makes the difference, it’s a major Aha Moment. This makes the problem very fun, and gives the problem its viral potential.

Using the lessons from the viral math problems, I worked with math enthusiasts to write a pilot set of interesting problems.

In parallel, I designed an interface for the problems, and a landing page to help build an email distribution list.

The center of the landing page was an illustration I created to communicate the abstract idea of a math problem set:

I also developed an illustration style that made traditional math problem illustrations look more credible and sleek:

And created wireframes for what the problem solving interface might actually look like:

Once the interface and the problem set came together, we built a rough prototype and recruited ten participants for a week-long diary study. We validated some hypothesis:

- People can do math problems on mobile, and it made sense to target mobile web as a platform.
- People solve at their own pace. Some people solved the problems in a few hours, while others took the full week, and some didn’t finish at all.
- Various interface issues, and some math problems that needed tweaking.

Shortly after the diary study I left Expii to be a full-time student again, but it looked like this when we launched:

Expii saw a significant traffic increase sitewide, attracted media attention from Fivethirtyeight, the WSJ problem blog, and the Atlantic, and sparked a competition partnership with the movie Spirit of Ramanujan.

As of August 2018, over 70 sets of Expii Solve problems had been published.