How might a game help people overcome public speaking anxiety?


The Talking Dead is a VR game in which players tell stories for an audience of approaching zombies.

Twitch is a live-streaming video platform with over 15 million daily users. We leveraged the Twitch platform to create a game that allows live audience participation.

1. The problem space

As a first step, I advocated for defining a narrow problem space and exploring current solutions within it.

We wanted to design our game as a research product to test the efficacy of virtual reality as a form of therapy for stage fright. Since we were building a gamified experience, we looked at a number of existing games for inspiration. Here are two examples:


Twitch Plays Pokemon inspired us to think about integrating live audience participation into our design.


Typing of the Dead inspired us to think about the power of thinking on your feet.

2. Our Concept

We chose to employ embedded design, a technique in which the true purpose of a product is buried deep within it.

In our game, a player narrates a story to zombies in a graveyard. We created a game diagram to ground our design and development process for the entirety of the project.

The final game diagram for the Talking Dead.

Here is a video showing footage from live playtests of the game.

Why virtual reality?

Immersive. It gets under your skin.

We chose virtual reality because we wanted to create an immersive experience that would evoke the feeling of being on stage more accurately than a simple screen-based medium. Learn more about VR best practices here.

3. Game Mechanics

Players tell stories to an audience of approaching zombies remotely controlled by Twitch viewers.


The game starts with a prompt that includes a person, a place, and an action.

Game play

Each zombie is associated with a word. The player must use the words in their story to kill the zombies.

Player view

In VR, game stats are displayed on a podium in front of the player. There are no HUDs.

Twitch view

On Twitch, relevant stats, such as time left, are overlaid on the screen. A chatbox allows the audience to participate in the game.

This labeled diagram shows different elements in the Twitch view of the game.

We created a chatbot that would allow Twitch viewers to influence the game in real time.

The game's chatbot accepts two different commands:

  • !word [word]
  • !vote [up/down]

The word command inserts a word into the queue used for generating zombies. We cross-reference submitted words against an English dictionary.

The vote command affects the approaching zombies’ speed. A game-wide speed value is calculated from the last 10 votes.

4. User testing

We conducted experience prototyping sessions as well as playtests to validate our game concept.

A photo from one of our experience prototyping sessions.

We iterated through mutiple versions of the game as moved toward higher fidelity.

Experience prototyping was a scrappy way to get our ideas in front of actual players as early as possible. As our concepts were validated, we moved to higher fidelity playtests.

A photo from a playtest of a game prototype in virtual reality.

The Art of Playtesting

Here are some good questions to ask during a playtest (adapted from Shawn Patton's work at Schell Games):

to throw

What was the most frustrating moment or interaction?

to keep

What was the most delightful moment or interaction?

to add

What was something you wanted to do but couldn't?

wish list

If you had a magic wand, what would you change about the game?

5. Our Process

We used an iterative design process with multiple playtests along the way.

Early on, we used a technique called Round Robin to brainstorm ideas. Based on the brainstorm, we reframed the original problem and placed it in an absurdist scenario, i.e., zombies in a graveyard. Here are two early iterations of the game that prototyped.

In this sketch, the player has a third-person perspective of the storyteller.

In this sketch, the player sees the game through the eyes of the storyteller.

Once our game mechanics were in place, we created a high-fidelity prototype and playtested it in virtual reality. Because of time constraints, we employed the Wizard of Oz prototyping technique for the speech-to-text technology, i.e., we faked the text to speech interaction by typing the words spoken by the player during the game.

I picked and modified assets from Google Poly to create a 3D model of a graveyard in Unity.

The game is built in the Unity engine. We ended up with four scenes: the start screen, the prompt scene, the core gameplay scene, and the game over/score screen. Here's a link to our Github repository.

If we had more time...

We would conduct more playtests to test the efficacy of our game as a form of therapy for people with stage fright.

We would also look into implementing a working speech-to-text translator, perhaps IBM Watson, to complete the game mechanics.

As interaction design lead for the project, I created and iterated on key game mechanics, chatbot design, and high-fidelity game assets. The Talking Dead team also included C. Erdogan, G. Guo, T. Wang, and N. LeBlanc. The playtesting protocol was inspired by Shawn Patton's work at Schell Games.