At Uncorked, we have a ton of opportunities to make things with new tools and technologies in areas where standards aren’t fully defined and there’s still a lot of blank space.
Most recently, we found ourselves developing Skills for Alexa. If you aren’t familiar or haven’t been spending any time talking to a lighted column or a hockey-puck-looking device somewhere in your home, Skills are essentially apps for your Alexa.
You can browse Amazon’s Skills Library and simply ask, “Alexa, add the Lyft Skill.” From there you can have Alexa summon her new Skill at any time by asking: “Alexa, ask Lyft to call a ride,” to which she might reply, “OK! Larry in a Ford Fusion is five minutes away.”
Skills are the primary way, beyond Amazon’s built-in interactions and services, that people can use Alexa to enrich their lives.
While chatting about Alexa Skills with Leia Sefkin, our director of technology, she suggested that we make a “Choose Your Own Adventure” type game, to learn more about the platform. When we thought about what people might do with Skills, discussions kept coming back to the type of architecture that Alexa requires.
With the current state of voice AI—Alexa listening to what you ask, and attempting to respond appropriately—still being very young, developers are trying to figure out what to do beyond getting directions, or the weather report, or a ride from Larry. While Alexa meets the mark, for the most part, in these basic things, most Skills aren’t usually meant to last beyond one or two steps, and don’t support complex interactions.
Using the Alexa, in its early environment, made me think of early days of computer games: low-res graphics, still images, not many options. And when you think back to those heady days of floppy discs and terrible sound, one game comes to mind: Oregon Trail.
We had it on the Apple LC’s in grammar school—you remember, the low-cost color ones on top of the pizza-box looking case—and we would play it during computer lab. And when you think about it, the idea works perfectly built solely for voice. Alexa gives you options, you choose, and suffer the ramifications: cholera and dysentery plus a really expensive wagon wheel to replace the one you broke while driving it too fast.
Since we have no visuals and our grisly fate must be told to us by Alexa at every turn, Alexa’s Oregon Trail had to be even more simple than the original game. Like the desolate trail, full of death and hardship, we tried to keep the blabber to a minimum.
How to Play
To start, there’s no family of five. This is a solo trip. But, we’ve got the trail, with similar stops, directional choices, and everyone’s favorite river-crossing methods. Departing from Independence, MO., the player chooses directions, encounters wild fruit, and comes to rivers—needing to decide whether to ford the river or caulk the wagon and float across.
After each of these events the player can “continue on the trail,” or “stop and rest” for a day. Each time the player continues on the trail, we add on a trip day and subtract food by five pounds—then assess their health by the amount of food. We’re still working on some other variables, including how the time of year, weather, and health fall into the equation.
When the player makes it all the way to Oregon without dying alone in a terrible manner, they win!
How it Works
On the technical side, this is all running on Amazon’s Alexa Skills Kit with the default setup. The Skill is registered in the Amazon developer console and the code is run by Lambda using DynamoDB to save the user state.
To that end, everything was pretty straightforward. Though as you can imagine, registering and detecting utterances was a bit tedious. Amazon even sent me a T-shirt for adding a Skill to the library.
How to Get It
The game is still a pre-release alpha, so don’t get too excited about talking your way down the Trail just yet. The reason being that we are still pretty limited in the audio context for now. There are features from the original game that we haven’t figured out an audio version for yet, namely everyone’s favorite: hunting. In addition, other features have proven cumbersome to navigate, such as buying a new axle at the store.
As a developer who mainly makes things for visual and physical interaction, it’s been fun testing the possibilities of voice, and in that finding Alexa’s present limitations. In that vein, here were some of my big takeaways:
If the questions (or answers) are too big, things go south. Smaller chunks are better
The immediacy of Alexa—ordering a car, or turning a light on—doesn’t naturally lend itself to a protracted game experience
It’s not clear how much back-and-forth with Alexa anyone actually wants to have
I’m most excited to see how the best of the visual and voice worlds of development and products collide, and ultimately how adding voice can help guide a user more deeply into an experience. Perhaps, then, our relationship with technology will reach the fertile lands of Willamette Valley.