PoetPainter - Thoughts
Thursday September 12, 2013 / 4 Comments

From Paths to Sandboxes

I’ve been a long time fan of the Montessori approach to education, but it was only recently that I read The Montessori Method. Contrary to popular teacher methods that focus on shaping the path of a student’s learning (consider that most school systems force students down the same learning path), Maria Montessori focused on creating a learning environment that encourages exploration and discovery. To this day, Montessori classrooms around the world encourage independent learning and nurture natural curiosity, which leads to lifelong learners. But, this is not without design and intent. Learning isn’t left to chance, it is designed. Or, as Maria Montessori wrote:

…we have prepared the environment and the materials

It’s this design of environments and the objects (or “materials”) within them that I’m curious about.

What Pinterest, Twitter, and Minecraft have in common

I’m working on a new talk, tentatively titled “From Paths to Sandboxes.” I hope to share the journey I’ve been on, from trying to shape and influence a user’s path, to creating engaging environments where people may determine how and under what circumstances to best engage with a system.

Pinterest, Minecraft, and Twitter logos

The catalyst for this thinking was another talk I gave, deconstructing why systems such as Twitter, Pinterest, and MineCraft are so maddenlingly addictive. Indeed, there are a number of psychological nudges used in these environemnts (here’s my answer to why Pinterest is so Maddeningly addictive). But, put side by side, I saw two common themes:

  1. These are platforms. You can make of it what you want. There is no prescribed way to use the system. Having been an early user of each of these systems, they all shared a common “Huh, why would I do that?” problem. It wasn’t until an early set of users started demonstrating how and what you could do in these environments, and why it might be valuable, that other people “got it.” Which leads to my second observation…
  2. These are social spaces, in which people learn from each other how to use the system. Many of the psychological nudges that follow stem from observing others. While MineCraft is a place for exploration and self-expression (perhaps survival!), it’s watching others that inspires new ideas and creates personal challenges. The hashtag in twitter was an emergent element. It wasn’t until I saw my wife pinning decorating ideas that I saw Pinterest as a visual bookmarking system. These are all examples of “Positive Mimicry,” whereby we learn by modeling our behavior after others.

Sandbox Games

In the gaming world, these are called “sandbox” games, games that “have an open gameplay structure that allows you to ‘play’ in the world and choose to participate in the story at your own pace” [Wikipedia]. With Minecraft, there is no leaderboard, no mission, no points, no badges, no clear objective to the game—none of the usual trappings associated with most games. But, this is the strength of this style game: players make of it what they will. You can play the game to create interesting homes, search for rare minerals, survive threats from creatures, help other players, hurt other players… All players are provided an environment, objects, and rules around what can and cannot be done, but you make of it what you want. All of this was carefully designed to encourage possibilities; contrast that with the typical game that moves you through a series of ever-increasing challenges, each one having you converge upon the same solution.

Possibility Engines versus Exhaustibles

This distinction between games that diverge into possibilities vs games that converge around a common end goal first caught my attention in a talk by Sebastian Deterding. He briefly contrasted “generative” games (also described as “possibility engines”) against “exhaustibles… systems with uses so clearly delimited, they are rapidly exhausted.” Deterding referred to the ancient game of Go as an open, generative game with “a huge possibility space of moves and countermoves and repercussions.” With just a few simple rules there is a near infinite number of possibilities! Contrast that with games like Halo or Candy Crush, which have clear paths and end points. These are fun for a time, until we either master them or lose interest. This latter style of game moves you through an ever increasing series of pre-scripted, linear challenges. Indeed, generative games also have ever-increasing challenges, but they are self-imposed. As you desire to get better at something, you create new challenges for yourself. And this leads to my problem with the gamified apps I’ve seen—they’re based solely on one style of gameplay, the incentives-based, rewards-based, leveling-up, “skinner box” variety of games. Note, this isn’t necessarily a bad thing. If your desired outcome is a prescribed one (increasing registration, training scenarios), a rewards based model could work, though it’s not without it’s share of challenges and is often effective as a novelty.

But what about this other kind of game? This more playful, less directed, form of play? Incentives seem good for training scenarios, but not for habit formation or discovery. Who is designing the playful environments that enable people to create, surprise, and delight others? Where are the environments that people learn from, not because of a scripted series of instructional steps, but through playful discovery, experimentation, and feedback loops? I think systems such as twitter or Pinterest, while not characterized as games per se, might qualify as this kind of playful sandbox environment, or the “possibility engine” described by Deterding.


While I’m not designing full out sandbox games, this thinking has already influenced my design work, in at least two ways:

1. Designing Features (or why not to)

In my own work, I’ve noticed a transformation from designing scripted paths to designing systems through which people can accomplish a variety of things, in their own way. Indeed, this latter form of design is much more difficult, and harder to discuss. We tend think about features and functionality that address particular usage scenarios:
“As a user I need a way to flag interesting tweets for reviewing later…”
“As a user I need a way to give kudos to people for sharing something interesting….”
“As a user I need a way save positive tweets for later use as testimonials…”
Twitter hasn’t specifically built functionality for these use cases, but they do offer the ability to star a tweet. What is starring even for? Twitter doesn’t specify how or why you might want to star something, and indeed there are at least 5, if not more, ways that people have decided to use this functionality. But, that’s the point—it’s precisely this kind of simplicity and ambiguity that allows people to use starring in the way that is useful for them personally. It’s a piece of functionality for which there is no specified usage, though it will satisfy any number of user stories. Designing in a way that says “you can use x to do what you’re asking for” is very different from building out the requested (single purpose) functionality. It’s easy to react to new feature requests. It’s much harder to practice restraint, listen to dozens or thousands of requests, and respond with a thoughtful addition to your software that could be used to satisfy most people. However, if we design in this more thoughtful way, we end up with systems that scale, are easier to maintain (no forked versions for specific customers), and accommodate many more users.

2. Designing Behavior.

Moving from “paths” to “sandboxes” may sound contrary to much of what I wrote about in Seductive Interaction Design. Perhaps. But, I don’t think I’m renouncing any of the persuasive design themes that I opened in that book. Rather, this thinking builds upon and refines that thinking. Take, for example, the familiar “elephant, rider, path” analogy. While “shaping the path” can certainly be an effective means of influencing behavior, I find myself thinking about the difference between short paths and long paths. Nudging someone to complete a sale or fill out a registration page is different from a scripted workflow consisting of many steps.

In the past, I might have designed the “perfect” series of events, anticipating every possible course of action—a long path with many forks and roads laid out before the user. This sounds like hubris on my part, as the designer, but it’s not without some justification. My thinking was influenced by the pseudo AI described by Will Wright, analogous to building a set of dominoes, the sequence of which can vary widely, so long as the links or endpoints match up. My lesson learned from these experiments was a simple reminder that users like to be in control. No matter how much we try to anticipate every course of action and create the perfect journey, people don’t go about doing things in the same way! From these projects, I’ve learned to design in a much more open and flexible way. A specific example: I avoid designing “steps” of any sort. Even the simple process of writing this post in an editor that then publishes these words to the page is too much for me—I’d prefer direct manipulation of the words, as you’ll see them. And don’t even get me started on wizards…

Handing over control

In the end, this is about control–handing over control to your users, without any prescribed outcomes. Which is scary, if you have defined business goals (“increase x by…”). But, what are the outcomes of this approach? More engagement and use? People who love your service, for what it helps them do? (Versus what it does for you). A more active and enlightened society? If we really step back and look at the outcomes of a group of people playing and learning together, this might be the most valuable thing we can build for a society. In many ways, the internet and all the good that is has enabled might fit this description of a “sandbox” game. And it’s not like we’ve giving up complete control—it’s now about directing, by way of setting the conditions and defined boundaries. What happens inside those boundaries is where we can all be surprised, delighted, and amazed.

Where next?

So there you go. This is what I’m thinking more and more about everyday: the open environment, the boundaries of that open environment, and the objects placed within that environment. How do we create engagement by designing sandboxes, rather than paths?

Obviously, these musings are still in the early, formative stages, I’ll debut this as a talk in October, at The University of Waterloo’s Gamification 2013 conference. And as of this writing, I feel as if I have all the pieces gathered (much of what I’ve shared here) but have yet to arrange them in such a way as to have meaning or singular focus, beyond the broad theme of “sandbox applications.” That said, I’m always blown away by the feedback shared with me when I share exploratory thoughts such as this. So, From Paths to Sandboxes, what are your thoughts?

Comments closed for this post.

  1. On Sep 12, 08:04 AM Ben Judy said

    Great article! I’d like to hear that talk.


    1. Core meanings
    One reason sandbox environments may be more interesting is that users are free to do things that align with what is meaningful to them, and what is close to their personal identity values. Indeed, without any guidance at all, I speculate people will likely default to doing things driven by what they find personally meaningful. Contrast this with a directed, goal-oriented environment where the end goal and steps to get there may not align — or even may be contradictory to — individual users’ core meanings and identity values.

    2. Strategic thinking
    Related to that, a “make your own path” environment allows for higher-level strategic thinking. As a user, if I’m not told what my purpose is in this environment, I default to pursuing things that are meaningful to me, and then I must begin constructing a strategy for pursuing goals around those core meanings. (I just made that up; it might be stupid.) But here’s a twist on your discussion of control: people like to have some measure of control over at least one realm, either strategy or tactics. If a system dictates to me both the strategy and the tactics, I might to get bored quickly. As you point out, all I can really then do is achieve mastery, and that challenge is only interesting as long as it exists. Once I’ve mastered the prescribed strategies & tactics, I’ve conquered the world and I move on.

    3. Montessori as an example
    I like how you opened with Montessori. If you open with that for the talk, consider revisiting it at the end. Could make nice bookends for the content. How do Montessori students fare? Are there famous graduates of Montessori education that you can point to as well-known examples of lifelong learners? How does individual, lifelong learning and exploration relate to the contemporary interest in “the learning organization” (reference: Peter Senge)..?

    Great stuff, Stephen.

  2. On Sep 12, 11:47 AM Wade Winningham said

    I find that even if you don’t provide a sandbox, people will tend to use something as their own personal sandbox anyway. Only through observing their actual behavior do you really know what they’ll do. And you obviously don’t have that insight when you’re creating something for the first time. So thinking about it the way you’re proposing makes sense to me.

  3. On Sep 12, 12:15 PM Sebastian Deterding said

    Thanks for the kind shout-out, and I think your link to Montessori is essential here — this concept of prepared environments — prepared, not infinite/incidental, but environments, not linear paths. It is also a pleasure to see how you seem to return to your roots as a teacher :). (BTW, Kurt Squire worked as a teacher in a Montessori school before becoming a scholar on video games as well, noticing the same Montessori principles in well-designed games.

    I also use the phrase “freedom, with rails”: Guiding paths and starting points for beginners who would otherwise feel lost, confused, or afraid of the blank page, but freedom for experts — an implied not forced trajectory of lots to little guidance that users can self-select into (as in jumping the tutorial if you feel safe). Feeling lost or overwhelmed by possibility is just as much a loss of control as feeling constrained — it is about the situational matching of individual plans/projects/ideas/wishes at a given moment, and perceived possibilities for realizing them (that was, by the way, Csikszentmihalyi’s original conception of play, from which his later flow concept, well, flowed). There’s a huge debate within the learning sciences about degrees of guidance, one epicenter of debate being this paper.

    Your observation that these are all “social spaces” triggers another connection in me, namely to what James Paul Gee calls “games with a big G”, that is, the game and the social community of practice around it, though he prefers the term “affinity space”. If “in games, learning is the drug” (Raph Koster), and the best and deepest learning takes place communally, among peers as interest-driven as you are (also fostering experiences of relatedness), then these affinity spaces around games are a big part of their lasting, sustainable appeal. You might have a look at his most recent interview, or some writing about affinity spaces.

    Finally, as to giving up control to the users, that’s in my opinion the true litmus test of generativity: “time to penis“http://www.urbandictionary.com/define.php?term=ttp. How do you create a living community with lived ethics that make your sandbox a playground rather than a cat litter box? Is the big design question here.

    And to leave with a question: Are sandboxes the right thing to everyone, always? When and for whom to build sandboxes rather than paths?

  4. On Sep 30, 12:20 AM Kars said

    It shouldn’t be a surprise I’m in favor of open-ended play. I guess my talk at Playful back in 2008 about playful tools might still offer you some useful input.

    I’m intrigued by the paper Sebastian links to. It might be guidance (paths) are best for instruction and sandboxes are good for performance? Perhaps we should rely on paths if we know the outcome we’re shooting for, but we should switch to sandboxes if we don’t? Sandboxes as generative tools, basically. Along the lines of what I talked about at TEDxUtrecht (you were there, remember?)


commenting closed for this article

Textile Formatting

-deleted text-
+inserted text+

"link text":http://link.com/