We at Shiny Shoe knew when we started working on our streaming-only game, Death’s Door, that we would be taking a risk. Creating a game that is only available to play on streaming platforms like Mixer and Twitch through viewer interaction had not been met with huge success. But rising popularity of game streaming platforms and the appeal of community-driven experiences like Twitch Plays Pokémon helped convince us that we were making the right bet. When Microsoft’s Mixer service asked if they could help us publish a sequel on their platform as a torchbearer for their unique interactive streaming content, we knew that bet had paid off. That sequel, Death’s Door: Aftermath is available to play (Fo’ free! Right now!) only on Mixer: https://mixer.com/deathsdoorgame.

Founded in 2011, Shiny Shoe is a group of 15 developers in San Francisco. In the past six years, we have shipped 5 original games and have helped clients ship many other games on a wide variety of platforms. Our cross-functional team contains a mix of experienced game industry veterans and new entrants to the industry. While Death’s Door was our first foray into the world of community-driven games, we’ve since used our experience to work on similar kinds of experiences such as our original games shown at Madison Square Garden in New York City.

We’ve recently announced that our stream of Deaths Door: Aftermath will be ending on November 30th with no current plans to continue (we’re still live right now!), but the two projects that we’ve published on this platform have been successful, constructive experiments in streaming-only, community driven experiences that have helped us to better understand the landscape of interactive streaming games and the possible future for new content developed on Mixer.

The Concept

Our back of the box description: “Death’s Door: Aftermath is a roguelike adventure inspired by classic RPGs and choose-your-own-adventure books, with one big twist: the whole community plays together via a live stream. Your vote helps guide The Revenant through a post-apocalyptic wasteland dotted with interesting encounters and tough enemies. Can you survive and defeat the ancient evil that destroyed the world?”

Death's’ Door: Aftermath and its predecessor, Death’s Door, were conceived as streaming-only games where a community of people would adventure through our world together. We could scatter secrets throughout the game knowing that the audience couldn't brute force or data mine the game for its secrets but would uncover them together. In both games, viewers come to our channel and watch the game unfold like any other stream. However, our channel presents the viewers (who are also players!) with choices to engage with directly as its sole input mechanic. Players can vote for one of the options presented and the majority vote will be picked to continue the game with the consequences shared amongst everyone watching.

Because there is only a single instance of the game running, all player state and progression are experienced by everyone at the same time. Secrets can be more deeply embedded into the game’s narrative with the assumption that players will crowdsource the solution and everyone will reap the reward once these mysteries have been cracked. Death’s Door and Death’s Door: Aftermath pull from different areas of design ranging from tabletop RPGs to large-scale social experiences and ARGs. And even with this weird combination of ideas, players have latched on to each game with a steadfast following and a hunger for more.

The original Death’s Door launched on October 31st, 2017 followed by sequel Aftermath launching on June 25th, 2018, exclusively on Mixer, and we’ve streamed consistently every Tuesday through Sunday from 9AM-10PM PDT since then. From the conception of the original to the sequel now in the twilight of its run, here are some of the things that we think went well... and some things that went a little less so.

The Good

With voting for options as our sole input mechanic, we put a premium on engaging decisions in Death’s Door: Aftermath. Each choice needed to feel meaty and consequential with any decision feeling valid given the current context of the game. Because the game would be streaming almost 80 hours a week and replayed constantly, it was only natural that we present these choices in a procedurally generated fashion, so we built the game with that in mind. We authored individual pieces of the game -- one-off events, battles, longer “dungeon” sequences -- and would remix and recombine the pieces in new ways with every run. While we tried to make each individual choice compelling on its own, it was the interaction within the context of all these other choices that provided the replayability that we needed and another layer of engagement to the game.

With our simple democratic approach to making decisions in the game, the audience could (hopefully) take responsibility for the choices and consequences made throughout their adventure. Different decisions would lead to lots of arguing in chat (inventory management is a particularly sensitive topic), but the aftermath of each decision was shared amongst everyone and added to the collective achievement, or disappointment, of the players.

This communal engagement added even another layer to decision-making with an organic system of social reputation and cachet now playing into which audience members could be trusted when offering advice (“Do they really know what they’re talking about or are they just trying to troll?”).

Misery loves company, but so does accomplishment!

We knew we couldn’t develop an homage to old-school dungeon crawlers and tabletop RPGs without a Dungeon Master! But we also knew players would be able to smell an automated Dungeon Master (DM) from miles away. So, we built functionality to send content to the game in real time allowing us to change the game on the fly or respond directly to chat memes under the guise on an in-game NPC. With some light intervention and clever writing, we were able to give players the sense that the DM was always watching, though players were sometimes unsure if this DM character was an in-game NPC intelligently responding to their actions or if there was a real human pulling the strings behind the curtain.

Regardless, this gave us the opportunity to create more reactive and dynamic experiences in real-time and create unique stories that would only ever play out one time. We doubled down on the Dungeon Master by building a web tool to submit content remotely and gave access to that tool to some of our trusted channel moderators. We also began to advertise a “DM Happy Hour” during weekdays to try and create some type of schedule-based programming. All of this led to players hoping (or not!) for a DM appearance in the game, whether that be to bless them with an OP run or scold them for cheating the system too much.

The Dungeon Master, when present, was also our stand-in “Streamer Personality” and provided some much-needed context and character to the game when large audiences were watching and playing. It was also a common “wow” factor for the game, adding another layer of mystery to what could be hiding behind different interactions.

Mixer’s focus on interactive experiences meant that our game was in line with several other new and experimental channels on the platform. We were one of a few other developers that were pushing the limits of their technology and finding new ways to use and break their tools. This community and the tools provided by the platform made the development of this game all the more seamless. At the time we launched both Death’s Door and Death’s Door:Aftermath, Mixer Interactive was live, an API and collection of web tools that allowed us to surface the game’s voting options as buttons below the stream.

Old Mixer Interactive Design

Our game offers a generous window of time to vote on a decision, but high-latency video playback meant that button updates in the web browser (updated by the Mixer Interactive web API) and information in the stream (updated via the stream output signal from OBS Studio) could sometimes get wildly out of sync creating a confusing play experience. Thankfully, the incredibly low-latency stream playback on Mixer meant that user input into the game felt responsive and easy to understand.

We also integrated use of Mixer’s passively earned currency, “Sparks”. Players can spend their sparks on fireworks that display in game (classified in-game as “Yay” and “Big Yay”). Each buy sends a single firework event to the game, allowing multiple players to send this event at once and celebrate boss wins or funny failures. The release of this feature immediately increased the engagement of players in the game, giving them another avenue for interaction and expression. Mixer has recently released collective sparks tracking on channels, allowing audience members to contribute their sparks to a weekly bucket that pays out cash to the streamer based on the number of sparks received (100,000 sparks for $15, 300,000 for $35, etc.) further cementing this feature as one of the more successful for engagement and returns.

Just recently, Mixer released a new set of tools to expand the kinds of experiences possible through stream interaction with Mixplay. Mixplay is a tool developed by Mixer for creating interactive stream overlays that can be personalized for each individual viewer. While the old API was easy enough to interact with from a desktop web browser, this setup could sometimes prove difficult to interact with from the Xbox and mobile apps. Our switch to Mixplay meant that we could surface voting options directly on the stream for more intuitive interaction and streamline the entire experience across all platforms, including mobile and console. It also led to...

Death’s Door and Death’s Door: Aftermath were both developed as communal-only games and finding ways to improve player retention without the existence of individual progression was always a tough challenge. Early on we played with different ideas of individual player state, (e.g., players could gain levels that would give them more weight in votes). Because of the hard constraint that we could not show players’ personalized information, we never landed on anything that could be communicated succinctly in the stream video and would scale as players popped in and out of the channel.

With the advent of Mixplay we were able to push individual, personalized content to each individual player as a stream overlay. This led us to create the aptly named - if not that sexy - Individual Player Experience (IPE). We focused on existing Free-to-Play models by adding the ability to collect “Minis” (8-bit, chibi-style renderings of existing Death’s Door characters) and integrate a collection mechanic into the main game. By voting and playing through the main game, you could gain coins and occasionally be rewarded a chest for winning a fight or completing a run. In that chest would be a Mini card which would allow you to unlock that Mini or level them up, with some additional coins, for use in a separate combat loop.

With the launch of the IPE we saw an almost immediate uptick in viewer retention and engagement with many viewers engaging more with the new IPE than the main game’s voting. This solidified our thinking that a larger focus on individual progression is needed to make a community-driven game successful.

Individual Player Experience (IPE)

With the launch of Aftermath, we developed an automation pipeline to startup our stream on time every day, report us of outages, and handle shutdown every evening at a set time. This freed us up to develop more content for the game, but it also meant that we were sometimes less present in the channel itself. Thanks to our amazing channel moderators (shoutout to ssg_roguey and AdjectiveBeaver) and an incredibly supportive community, the channel was able to sustain itself with relative autonomy.

We frequently heard from players how they keep coming back to the game to play with their new friends in chat or how supportive and helpful the longstanding members of the audiences have been or how uncharacteristically friendly (for the internet) the mainstay audience members were to newcomers. Through the patience of our chosen moderators and our attempt at good-natured attitude when interacting with players, we were able to foster the welcoming community that exists now. Our game brought people into the channel, but our community helped keep the hardcore fans coming back.

The Not So Good

Being on the forefront of stream interactive games meant that not many people understood that what they were watching was actually a game they could play right there. New players, who can arrive in the game at any moment, need to be able to understand what is going on at a glance. With Mixplay we were able to present a more targeted “tutorial,” instructing players on the basics of voting interaction and how choices are made in the game. But the multitude of different systems, RPG status effects and interactions, story events and consequences create a steep learning curve for new players.

Veteran players in chat could sometimes help mitigate this by explaining newcomers the rules but depending on the state of the game when those newbies first arrive, they could be completely confused by what they’re seeing and fall out of the stream as a result. New players could arrive in the middle of a final boss fight with a fully-loaded out player character, bombarded with all the different icons and stats they’re seeing on screen, and that can be, understandably, frustrating.

Some of the solutions we’ve discussed, but never got around to implementing, were mouse-over tooltips surfaced through Mixplay or more involved tutorial sequences for each individual user. Mixplay completely changed the playing field with what we’re able to do for an individual player and we’ve only scratched the surface in terms of delivering information succinctly and effectively.

Just like players in any MMO game, those with more experience have more power, and because there is little individual progress in Aftermath, experience is power. Experienced players know the consequences of different story decisions and they know how to optimize inventory loadouts for certain bosses. Experienced players want to find the hard-to-find secrets and beat the hard-to-beat enemies using their hard-earned knowledge. But that can be directly at odds with less-experienced players not knowing the optimal approach to a situation and keeping more experienced players from getting there. Or, less experienced players know the correct approach but want to see the suboptimal one since they’ve not experienced it personally.

The stratification between experienced players, newcomers, and everyone in between also means that there’s an inherent mixture of personal goals. Since everyone is playing the same instance of the game, creating opportunities for everyone to achieve their own, specific goals becomes difficult in our current game structure and is an ongoing problem that we’ve made attempts at solving. The IPE, for example, has improved each player’s sense of personal achievement, giving them something to focus on when the main game may not be creating their desired opportunities. So, continuing to invest in individual progression seems to be a strong solution for managing these different levels of experience.

Our lengthy streaming schedule means that players churn through new content quickly and we’ve consistently added new content to keep players satisfied, but the cost for this content creation can be high with returns hard to measure. Our primary monetization sources in Aftermath are channel subscriptions and donations. We received some return from each source with some in-game recognition for either action, but the IPE allowed for direct rewards for donating such as high-level minis and chests. Even though the IPE introduced more incentive for these actions the return rate is still low. Pair that with expensive content creation and you get a difficult-to-justify cost for a content treadmill that is demanding to keep running.

In Aftermath we added character class levels and objectives to try and gate content throughout the game behind game-specific goals (e.g. Kill 3 of a certain enemy, Do 1000 damage, etc.). This automatically lengthened the amount of time it takes for players to experience all of the content we have to offer but makes the game more “grindy” as a result.

As it stands right now, the business model for Aftermath is unsustainable given the cost of development. However, with the recent release of Mixer’s Season 2, monetization is more ingrained with stream interaction on a platform that’s continuing to grow rapidly, and we believe successful models can be found to not only make a sustainable game, but one that thrives.

Most streams distinguish themselves through the streamer’s personal brand and audience members keep coming back to those channels because of that streamer. Being a unique kind of game, Aftermath had no problem pulling people in to check out this new idea, but without a constant streaming personality on top of the experience, there wasn’t that constant human connection with the player to keep them coming back.

We automated many common streamer callouts (e.g. acknowledging followers, subscribers, and donations) and added specific “Subs-Only” and “Follower-Only” votes to encourage subscriptions and follows, but the lack of a constant personality made it hard to compete with the more “traditional” streams. Our Dungeon Master provided helpful character, but DM appearances aren’t consistent and interrupt the flow of the game. Combined with added audience confusion over the true existence of the DM as human intervenor or super-smart AI system, our DM character wasn’t the right solution for this problem.

Aftermath and other interactive stream experiences aren’t meant to replace streamers, but to provide a new sort of streaming content for people to engage with. But that hasn’t stopped us from incorporating more “streamer personality” into our channel. More recently, we’ve tried streaming ourselves on top of the game which has resulted in increased engagement with the channel and another scheduled time for players to return and engage with a real human being.

At the time of this writing, the same instance of Aftermath has been streaming for over 1500 hours. Players have hopped in and out of the stream, but we’ve still designed Aftermath with the idea in mind that it’s possible for individual players to continue to play this game for 1000s of hours more. The roguelike structure of the game provides inherent replayability to the game, but the degree to which this game is replayed is far beyond anything we’ve created before. Add 1000s of individual people playing at once, and you have a game that needs to provide new experiences to a large audience who’s already played many hours all in one, single running instance of the game.

There was a definite learning curve for the development team to create meaningful content that could be replayed over and over and over again. Specifically, in trying to shy away from too much authored content and let the engaging experiences come together naturally through systematic collision. In Aftermath we began to add more player character items that were affected by your actions (e.g. an Axe that would do damage relative to the number of throwing axes you’ve used in a fight or a Relic that would give your player a base defense stat equivalent to the number of times you’ve attacked enemies’ weaknesses). This more systematic approach allowed players more opportunity to “break” the game, but that’s where some of the true fun of Aftermath lies.

But, being a narrative-heavy RPG, the need for authored content still exists and reading through the same boss dialogue for the 100th time isn’t exactly the most engaging. Trying to pivot towards an even more systems-heavy game or one with more procedurally-generated narrative could help to extend the replayability of the game, but each carries its own cost as well.

The Conclusion

Death’s Door: Aftermath is unlike anything we’ve built before. We set out to build a community-driven game that presented interesting decisions and consequences wrapped in a compelling and mysterious world. We wanted to create something that seemed bigger than the box with something else always hiding just a little farther ahead. And, in those ways, we achieved what we set out to do. The game is compelling and grew a dedicated, engaged audience that uncovered the lore and secrets we created for them.

The business case for interactive streaming-only games may not have supported sustainable development at the time of ourrelease. But with more ingrained monetization within the platform and extended support for interactive controls, Mixer’s Season 2 paves the way forward for the next generations of these experiences and shows a promise towards even more development. By taking advantage of this new technology, leaning in to the unique interactions capable on streaming platforms, and providing a clear focus on personal player progression the next wave of streaming-only experiences can break the mold and become a viable new way to do  game development.

In anything we create, there’s always something to be learned; something to be improved upon for the next go around. Hopefully, you can learn from our successes and failures so that we can be playing your game on Mixer soon enough. In the meantime, our stream is running until November 30 so why don’t you hop over to Mixer and give it a go while you still can!

Death's Door: Aftermath: https://mixer.com/deathsdoorgame