I’m going to be honest and say that I’m not positive what went right and what went wrong on Eidolon. I still think it should be valuable to share my (or our) experiences developing, releasing, and supporting this project. This ended up being a very long post-mortem, but I think getting it all out there has some value. (This is probably more information about the process than literally anyone but myself and maybe my partner have, so enjoy the transparency.) Apologies for the length.
I’ve cut it into a few major sections, so here’s a little table of contents:
DEVELOPMENT PART ONE: Early Tries, 2010-2012
DEVELOPMENT PART TWO: Making Eidolon, 2013-2014
DEVELOPMENT PART THREE: Post-launch Support, August 2014-2015
NOTES FROM 2016
DEVELOPMENT PART ONE: Early Tries, 2010-2012
Eidolon was originally conceived of and drafted in 2010 (then referred to internally as “Scientist” or some garbage) as a top-down 2D survival/adventure game with similar narrative focus and world but far more extensive and central survival/crafting mechanics. I’d been reading academic game design writings and was inspired by the idea of a narrative embedded in such a way that it could be central to the gameplay without disrupting or dictating the emergent narrative of the gameplay itself. I spent a month or more making prototypes and building a design document with friends who would later be a part of the real Eidolon team, Isa Hutchinson and Jacob Leach.
At this point the player character was going to be somewhat Ico-like: a mute boy from a strange tribe abandoned/sacrificed/ritually exiled. We were really into the idea of bearskin coverings for some reason. A bit embarrassed by all this, now.
Then we played Minecraft for the first time. Someone had already made the best parts of what we were envisioning, we thought, and so we walked away from the idea.
Come 2011 I was working on little Flash projects here and there, teaching myself to program for the first time (you can go play my little games on Kongregate if you want). I decided to try to create a version of the survival and crafting systems I had wanted, limited to a single island environment. At this time I saw my games as purely an artistic venture with no commercial side; if this had been released it would have been only on Flash portals and with no marketing. It reached the state pictured below and fell permanently into my backup drive. I think there are a few reasons for this. For one, I wasn’t emotionally ready for such a large project. But also, it just wasn’t that interesting to me. I’m glad I didn’t pursue it, as it looks a bit like a poorly conceived version of Don’t Starve.
In early 2012, I was compelled by my college to declare a major. By that time, I was at Fairhaven College, a sort of ultra-liberal haven within the liberal community that is Western Washington University. They had a program where you built your own major on a single subject with an interdisciplinary focus under faculty advisement. I’d recently decided to pursue games not just as a side hobby, but also academically, if not as a career. I spent a quarter writing my major. My vision for my last year or two of college was that I would make many more small games, slowly developing my skills and finding my voice.
But one of my faculty advisors, the fantastic painter Cynthia Camlin, pushed me hard in another direction. She wouldn’t officially approve my major unless I committed to doing a single, large scope thesis project. I agreed, somewhat begrudgingly but also inspired by a teacher with such a strong vision for her student. I was additionally made to write a plan for what my thesis would be into my initial major, though I could change it later on when I actually had to approach the thesis project. I didn’t have a lot of time to figure out exactly what I wanted to do, but I had this old project that had been repeatedly abandoned. I wrote in the concept pitch for what I’d been calling Scientists, Nomad, or Island throughout the years.
Legally, Eidolon was developed and released by Ice Water Games, my own single-member LLC. The other 9 team members worked on royalty share contracts. A team member who contributed X% of the finished product, is currently getting X% of net revenue paid out to them quarterly by IWG as royalties.
Effectively, Eidolon was my personal baby, my undergraduate thesis project. I made sure from the beginning of official development that I reserved the right to veto any contributions or design suggestions. I very, very rarely used this power but I felt it was important, as the project was envisioned by me and in large part developed by me, and I was nervous about sharing a long and personal creative process with others. (I ended up putting in ~50% of the total hours, and very few aspects of the game weren’t overseen by me in some way.)
I also made the decision early on to try to give the other team members exactly as much room for creative investment as they wanted. This was rocky and strange. Some members invested a few hours a week for a few months at the beginning, then completely moved on from the project. Some didn’t show up until the end. Some faded in and out as their personal interest and other commitments swayed. This was good, in theory at least, because it represented the reality of our situation well: nobody was making any money up front; nobody was obligated to continue on; we were all just working to the degree that we were excited to continue working. It made calculating revenue splits a complex and long conversation, but I firmly believe that every member of our project is happy with how that ended up (so we did a good job on that regardless).
In reality, I feel that the fact of the project being fundamentally my baby deterred others from investing as much in it as I would’ve liked them to. I wanted the project both to be mine (I was afraid to open up) but also for others to see it as theirs (I wanted the feeling of mutual investment). It’s hard, maybe impossible, to make both happen.
DEVELOPMENT PART TWO: Making Eidolon, 2013-2014
When Eidolon Was Almost a Minecraft Mod
Eidolon’s actual development cycle began at the very start of 2013, with an independent study in which I wrote speculative short stories, under the tutelage of poet Mary Cornish, to flesh out the Eidolon universe. At this time I also approached the first few members of what would later be the development team: Jeff Klinicke (the most prominent world writer), Meagan Malone (the most prominent character writer), and Adam Murgittroyd (the most prominent codefriend).
Adam and I talked a lot about the technical problems around making the game, then referred to as Nomad, and decided it would be a stronger experience in 3D than the 2D I’d been using. Specifically, it needed a horizon. There is something both instinctive and powerful about being able to see something small and unclear and far away, and then to be able to spend time moving closer to that thing and bringing it into focus and eventually learning what it is. And in a lot of ways, that was the core of what was important about Eidolon: a focus on player-driven exploration and discovery. Failing to mirror that focus in the visuals would have been a tremendous disappointment, especially when the provision of a horizon is so natural and unassuming. We’d never done a 3D game before, but had made a small Minecraft mod just previously, and realized that Minecraft took care of many of our problems for us. I didn’t really care about the survival/crafting as much anymore, and instead was focused on what made our game different—the crafted, non-procedural world, the history, the human aspect. We decided to make Nomad as a Minecraft mod.
That idea didn’t last long. Not even to prototype phase. We elected to switch to a commercial engine, partially because modding Minecraft was incredibly difficult and legally/financially questionable, but also under the pretense that we might want jobs once we graduated, and we knew Unreal Engine was popular in ‘the Industry.’ So we got to work in the Unreal Development Kit. At this point, I was interested in the idea of somehow blending highly textured models and an abstract aesthetic. I really didn’t have a concrete vision for what I was doing, and just found myself floundering uselessly with leaf textures and tree styles.
Finding Our Visual Style
I had so much trouble with texture that I ended up slowly moving towards softer and flatter textures. And then another thing happened—performance issues forced me to learn about and then implement Level of Detail meshes. I just had two LODs at first, one very high poly and one with as few verts as I could imagine working. The disparity between the two was so jarring nobody around me would let it slide. I had to make the change more gradual. But I realized I liked the lowpoly mesh better, and that it would be more efficient and simpler to just use that everywhere anyway. So I spruced it up and made a dramatic pivot to the style of meshes that are in the final game.
As for how we ended up using flat shading, it was stumbled into while trying to hook up toon shading on the suggestion of a peer. The flatness solved a visual problem: the clipping of leaf planes on our deciduous trees. It also lent the game a more abstract, painterly, and cohesive feel. It made many of the most blatant seams practically invisible.
I had been spending the vast majority of my evenings working on Eidolon, and talking about it with everyone I knew, and this apparently had an effect on them. The team structure for Eidolon was always open, so when more of my friends began approaching me asking whether they could help work on the project, the answer was always yes. My partner Zoe signed on to do UI, my professor Michael to do music, my roommate Aron, a historian, to help with the world building, and long-time friend Jacob (who had been around when we were first coming up with the idea in 2010) to develop prototypes for the animal AI and tools. The majority of this team (especially the writers) would meet up once per week for a couple of hours to flesh out the world history and timeline, and eventually to build a cohesive wiki for us to use later in character writing.
Finishing, and Finding Eidolon’s Design
It was around this time that I was really solidifying my thinking around why Nomad was important to make, what was special about it (you can read about it at length here if you really want). I wanted to take optional world story and make it the implicit goal of a game, while also freeing the player’s moment-to-moment decision making from any imposed narrative constraints. This meant that the player character had to not exist; otherwise, the character’s motives could conflict with the player’s motives. Instead, the player character in Nomad had to be the player, or else whomever the player imagined. From this came the final game’s insistence on not showing your hands, not implying character race or origin or age or gender or motive or education level or even corporeality (though there are journals in the game where a character might be writing to you directly, depending on how you interpret it; and there is the implication of a backpack in the UI, which suggests something technologically).
Most of the game was solidly in place, aesthetically and structurally, when I graduated in December of 2013. For the next 8 months, we worked quietly and mostly separately, writing and sculpting and fixing bugs. The final two team members to join up, good friends Isa Hutchinson (writer) and Shadie Hijazi (programmer) came on at some point during the summer, and I ended up working most directly with them for the last couple of months. Before this point, I was working at a coffee table after class while watching bad TV and drinking. This is when my work finally became regular: I would make task lists and work through them on a 9-5ish schedule.
It was also during this time that the game went through a significant pivot in design. There has always been and still is a lingering shadow of our early focus on survival mechanics. This is probably the weakest part of the game, disregarding the bugs, but it wasn’t until late in the development that we fully admitted this, I think. Or at least, we didn’t know what, if not survival, the moment-to-moment experience of Eidolon could even be about.
Eidolon has always presented the player with a broad goal (move slowly in a direction to explore/discover) and micro-goals (turn to look, walk either left or right, jump or not, etc) but there weren’t always easy-to-find, mid-sized goals. The question “What do I do next” still lingered heavily between those two spectra of player choice. We had some survival in there, but survival didn’t fit with our focus on exploration. One solution to this design problem came when we recognized the use of the compass as an interesting mechanic that contributed more-so than our other tools (rod and bow) to the theme of exploration. We realized that we wanted, and could have, more tools that provided explorative functionality. Our first response to this (credit Shadie I think?) was to put in the binoculars, giving players the ability to make slightly broader micro-goals for themselves by being able to scope out their exploration a little bit farther. This diminished but did not remotely obliterate the goal gap.
It had been my plan that players would find a document early and use clues in that document to set mid-sized goals for themselves: “Triya writes about heading north, so maybe I should head north to follow her,” “Ada mentions leaving Oldtown to the south, so maybe I should head south to investigate that.” But the goals had to be so awkwardly written into each text (why do characters always mention where they came from and where they’re going?), and the placement of documents didn’t always make sense in any particular way. I didn’t want to force stilted goal-writing into each document, so instead the branching web of stories to be explored just kind of disappeared from the player’s experience. When confronted with the “What do I do now” playtester, I decided to make this web explicit and give players the ability to make clear choices about what stories to pursue: thus, the tag system. Now, when a player reads Ada’s document, they can pick out which thread is most interesting: Oldtown, Ada, etc, and then, by clicking on the corresponding tag and following the indicator, explicitly seek out the nearest relevant document. This was a tremendous design change for Eidolon. It gave players so much more control over their journey. They could finally actually make mid-sized choices about their exploration. This was the point when Eidolon finally felt like Eidolon.
I had no idea how to market a game when I started trying to figure out Eidolon, but I’d seen media build small cult followings through tumblr, so I figured that was a fine starting place. I’d been posting all of my design articles on a personal tumblr called Game Design Sketchbook (http://icewatergames.tumblr.com/tagged/theory) and had a few hundred strangers following me, so I just forcibly transitioned that blog to a development blog. I also made a devlog on TIGSource and posted there as I went, since I knew I’d get a lot of fresh eyes that way.
Just before we posted our announcement trailer in December of 2013, we got a miraculous write-up by Chris Priestman through IndieStatik, which rippled out to Rock Paper Shotgun and introduced us to the world. Then we put our trailer out, and saw articles from Destructoid, Kotaku, and Kill Screen. This was huge to us. We were a passion project at (what is locally perceived as) a second tier state University. None of us had ever made a commercial game before.
The early successes made me continue posting online and eventually to spread to Facebook and Twitter, and to launch a Steam Greenlight campaign.
DEVELOPMENT PART THREE: Post-launch Support, Aug 2014 - Now
Releasing Eidolon was exhausting and exciting and insane. The game is somewhat buggy, but it was far, far buggier at launch. These days, we mostly just get reports about Mac compatibility issues. Post-launch support was a nightmare and a revelation for me, a student indie with basically no industry mentorship and a hell of a lot of idealism. Turns out, extensive testing is important.
For the months after launch, I had my phone on to yell at me me whenever a bug report came into my inbox, and there were several times when that meant frantically jumping out of bed at 3am to attempt to fix and quickly patch a bug. My memory of this time is honestly hazy, but there were a lot of issues with floating grass (tweaking the heightmap after placing foliage was a bad idea in UDK), and that kind of stuff—little inconsistencies in how the world was represented—just dug into my head and could not be left alone.
As soon as I opened up pre-orders some time in 2013, I had advertised Eidolon as being available on Mac. I had no idea that there were serious differences between the platforms, and thought that surely some light OSX testing would find any issues. All development was done on PCs, and any pre-release OSX testing happened on a couple of old MacBooks.
Eidolon ran okay on old MacBooks, since that’s what it had been tested on. However, it ran abysmally on expensive, brand new, top of the line OSX rigs.
Months of pathetic desperation eventually uncovered what I believe (but cannot ultimately verify) to be the issue: UDK’s optimization of our specific art style in OpenGL (barely implemented before the move to UE4), specifically on Nvidia cards. UDK’s licensing model at the time was such that to get access to source, we would’ve had to pay out more than double what we’d earned by then. So basically, not possible. Basically, the saddest. My ultimate fix was to post as loudly as possible on all storefronts that OSX was only sort of supported, and to continue my incredibly open refund model. I sent so many apology emails.
We patched the game once, with some extra stories that didn’t make it into the first build. That was okay. It was nice to fill in the Kitsap Peninsula, a pretty big geographical absence on the original map.
Eidolon’s total budget before launch was ~$650. No, I am not implying any extra zeroes. This went to UDK ($100 upfront), Greenlight ($100), Flash ($250), Promoter ($50), Dropbox ($10/mo for ~5 mo), and various costs (travel, food) related to the two places we exhibited the game (~$100, Viking Con in Bellingham and Invisible Arcade in Seattle). I had some savings, so I just paid this out of pocket. People’s living situations varied but for most of us this was essentially either a student game, a hobby project, or a personal art project: not something we expected to pay our living expenses upfront, if ever.
We launched in August of 2014 and sales to this point have netted us collectively ~$125,000.
I don’t really care what you take from these numbers. Especially don’t feel entitled to tell me what I should take from this, as some have in the past. For us it is a tremendous success. The amount that I personally have made from this, as Eidolon’s most prominent contributor and beneficiary, would be enough for me to make another game of similar scope with no further income. It’s what afforded the development of Viridi (http://www.icewatergames.com/viridi), which was made by a subset of the Eidolon team: myself, Isa Hutchinson, Zoe Vartanian, and Michael Bell.
NOTES FROM 2016
I played some Eidolon recently, for the first time since working on it. I’ve spent a long time feeling shitty about this thing I made, but now, looking back a little bit clearer, I’m actually really proud of it. Technically, it feels incredibly juvenile. But reading the stories through in the order they come, not remembering the full architecture, that’s been a powerful experience. And exploring a vast world. Watching the clouds and the stars, feeling the rain, encountering a creature or a ruin. Lots of feelings. It’s a really personal game for me, which might seem strange since it was made with 9 other people, but then those are all people I have personal relationships with, and I think that’s part of it.