Weaving 13 Prototypes into 1 Game: Lessons from Edith Finch

All right. Everybody ready? Welcome. Please silence your cell phones and yeah, we're ready to get started. So this is going to be a talk on how we created the prototype for Edith Finch and then how those evolved into the broader game itself. And this is the the team at one snapshot. We started the game with three people on the team and then grew to about 15 and then tapered off towards the end. So it's sort of a medium sized team, not not too small, but but certainly not not triple-A. The game itself for those who don't know or it's been a while is a collection of short stories and each of the short stories has a totally different game mechanic different from the rest of the stories in the game but then also hopefully different from anything else players

have experienced. The game is an attempt to try to keep players in this kind of beginners mind of never quite understanding what it is that the game is about so they're always kind of feel like they're discovering new things as they move forward. So we did a lot of prototypes for each of these stories and each of the prototypes also had the constraint that there are no tutorials in the game because the game is about this sense of discovery. So we wanted to create things that people had never seen before but also things that were discoverable that people could learn how to progress by playing the game. And instead of focusing on traditional goals like challenge this game is focused on evoking a feeling and that case it's the feeling of the sublime which talk about in a minute. Here's a quick trailer.

All right. So that's the game and in this talk I'll be beginning with the kind of high-level approach to what how how we approach the overall design for emotions and then talk about how we figure out which emotions we're going to be focusing on in in the particular game we're making and then get into the nuts and bolts of actually making those prototypes to try to evoke these feelings. All of these kind of tie into the

problem of how do you make something when you yourself are kind of a little unsure about it which is part of the appeal as a developer that we're making things that we ourselves are kind of actively exploring and the games that we make tend to be themselves about a sense of exploration because that's kind of what the game development process for us is like. The development process of Edith Finch is kind of a worst case scenario of going in blind but hopefully a lot of these things will seem somewhat familiar for games you might be working on where there are at least you know a few aspects that are not very well sussed out but hopefully you know a little bit more about your game than we did when we started making it. So starting off just to give you kind of a philosophical perspective my sense of the way that the games work in stories and and the emotion and the difficulty there is that it really boils down to attention and what you can expect

players to be focused on and it's difficult in games because there are a lot of things that are going on compared to a play where people are generally like pretty much sitting or going to the bathroom one of those two things for the duration of the performance. But in games the problem isn't just your game it's every other game that players have experienced coming into it in their lifetime has kind of trained them to approach the experience of the game often as a life or death situation. Usually you know if you stand in one spot for too long you will be destroyed in a lot of games and genres or you will lose or whatever it is. Um So players aren't used to taking their time. They're not used to stopping to take a breath. They actually have to be encouraged to do those things. So we try to create a space that evokes you know that feeling and kind of encourages them to approach it

differently than they would a more traditional game. So this is kind of our our high-level plan of attack which is you know that we start off by you know not giving them hopefully too many distracting toys. I mean we try to balance it. There are times in the game where you become like a giant tentacle monster eating sailors but those are not the same moments where we're trying to you know get you to pay a lot of attention to a specific area. You know we sort of alternate these very small scenes with broader sillier moments and then yeah it's this in a nutshell what we are trying to do is to create interesting spaces and tools to explore those spaces and those tools are the gameplay prototypes that we'll we'll be talking about a little bit later. And instead of relying on an explicit story and you know setting up character and conflict and plot in words and dialogue we try to do as much as we can

just with the environment itself. Ideally it's something that you know it's like a silent film where you don't need any audio or text in the beginning and just the experience of being in that space already gets you to ask questions as a player and wonder like what might be happening and get you into the the mood of it and it's a lot easier and more reliable because players don't have to pay attention to context it's something that you know they can kind of absorb through osmosis. And instead of starting again with with like a written more didactic story we start with a very broad goal of evoking a certain kind of feeling and what we found is that when we start with something that is a little more open-ended like that even though there's a lot more uncertainty it's something that actually has a good chance of making it to the finish line because games tend to go through a lot

of changes as you realize you know how many of your initially sound seeming ideas aren't going to work for a number of reasons but when you hold on to a little bit of uncertainty and you bring it with you you know that that gives you enough room to actually preserve the original intent even though what ends up at the end you know is not what you would have expected but it is you know if it's something like a feeling there's a good chance that that's going to carry through. In the case of Edith Finch the game began as a way for me to explore a particular moment that I had in my own childhood when I was scuba diving as a teenager and saw the bottom of the ocean sloping away into like a seemingly infinite darkness and that moment where something was simultaneously very beautiful but also unsettling which in a word is like the sublime. Like this moment where you you feel you know the transient beauty of the

universe but also you yourself you know are acknowledged to be very small and a fragile part of that. And so this game was trying to explore that. Initially it began you by literally remaking that teenage scuba diving experience more or less and then it just like was way too gamey and didn't actually feel like what was in my head and so it went off in a obviously very very different tangent. And then once we have a better understanding of that particular feeling which will be the next thing I'll talk about then we move on to making prototypes because those prototypes are the most difficult part of the whole process and have the most kind of emergent complexity and changes that arise out of the process of making them. Story comes in like we we begin with a best effort at the very start of like yeah okay we think these are the characters and this is you know roughly what's going to happen. It's going to be a story of a family for example in this case we we

knew that it was going to be centered on Edith but um we found a lot it was a lot easier for us to rewrite everything once we had the experiences which is like a classic chicken and egg problem but we would have people play the game you know as it existed in a more or less complete prototype and then ask them about how it felt to play the game and then write more to that experience. So instead of trying to you know make an experience that tracked with the story we just like adjusted the story to track better with what the prototype was that that we had made. Okay. Next up is finding that feeling in this case you know we started with the sense of sublime but it you know had to get a little bit more detailed and then the challenge is of you know once you have that idea how do you communicate that to the rest of the team? Uh, so for game development purposes, I think

it's helpful to boil down things into uh a couple of very concrete ideas. Uh, ideally, in the case of Edith Finch, like we we had three words, sublime, intimate, and murky. And the hope was that these were feelings that anybody on the team could look to for inspiration for how to do whatever it was that they were engaged in that day. So, you know, the lighting artist, you know, might be working on like the fog or, you know, an animator might be working on a jump animation, whatever it is that they could look and say like, "Oh, is there a way that I can make this a little bit in- more intimate or more murky?" And, you know, even before anything has actually been checked into Perforce, people have this collective sense of, you know, what this might feel like. And there's going to be divergences, uh, but that's part of the appeal also. When you have something that is powerful and specific, but also a little bit open-ended, it provides a space for everybody else to contribute their own perspective on these things, uh, you

know, and and add to the richness of the game without hopefully pulling too much in, you know, opposite directions. Uh, and then, once you've kind of settled in on the feeling, uh, I think it's really helpful to try to find as many really good examples of that in other mediums, uh, kind of as a way for you to almost like play the game in your head years before you ship it. Uh, in our case, there was a genre of literature called weird fiction that, uh, like Neil Gaiman and H.P. Lovecraft and a bunch of other folks have, uh, have written really good, uh, stories about the sublime. And a lot of weird fiction deals with the challenges of being in a universe that is stranger than you can possibly imagine. And, uh, so those were really useful tools for us to, you know, kind of imagine what it would feel like to play the game by reading it. Uh, also, soundtracks are great because people have already done the work of saying, you know, what does

it feel like for this, you know, like, in the case of Edith Finch, things like Donnie Darko or Under the Skin, uh, you know, movies that were totally kind of in that ballpark, uh, what might that sound like? And so, for a lot of the project, I just had a couple of those soundtracks like on repeat, and it was a way of kind of getting into this alternate headspace and kind of dreaming in that place. Uh, also, just printing out a whole bunch of references was really helpful and sticking them up all over the office, uh, kind of engaging the visual part of your brain, which I'll talk about a little bit later. And then, uh, similarly, updating your Windows or Macintosh desktop background is a nice thing, just to kind of like put it right in the middle of your face. Uh, so like here's the, uh, the game that we're working on next. Uh, that's a little preview. Um, okay. Oh, we're we're out.

Okay. And, okay, here's our first mood board. Uh, this was, I don't know, maybe like three or four months into development on the project when we still thought there would be magic and combat and, I don't know, like Edith was going to take words and put them together and make spells. This is like a nightmare. Uh, it was totally different game. It had a different name. Uh, it never would have shipped, but this is what it would have looked like. And actually looks kind of like a game we did ship, called What Remains of Edith Finch. And I think it's just like the power of having these, you know, kind of specific, but somewhat open-ended, uh, tonal goals. So, we didn't ship a story about a microscope, you know, but, uh, there is a lot of underwater elements, and certainly the the forest in those, uh, lighting conditions are things that are not too far away from from what we did ship. Uh, and then, next I'm going to show a um, Greenlight trailer that we made, I think about like four months in, that,

uh, is kind of a moving version of this mood board. None of these things actually shipped, but, uh, you know, they're kind of what we were imagining the game as. Maybe some of those books. I think some of those books might have been in the game. You can see it's a little more haunted house-y than the real game, but there is, you know, still a sense of this very large house that you'll be exploring. And, you know, a little bit of underwater stuff, still clinging to that original scuba diving inspiration. Uh, here's an invisible dog that we thought would be a good idea. Uh, we went back and forth a lot on how puzzly the game would be. And, you know, for various reasons, like, you know, we the game could have had more of those kind of delightful physical interactions, but they're just really hard, and they weren't where our heart was when it came down to it. Um, but this is really weird. We had an underwater, uh, sequence in the shipping game, you know, like an underwater house

that, uh, we completely forgot about for several years, but it was there four months in. We had this underwater house, and, uh, you know, I think it just speaks to, you know, the game being about this combination of civilized and natural spaces. Uh, yeah, this also did not actually ship, but, uh, it's a nice idea. And, so yeah, what you're looking for, or what we're looking for, is, you know, a sense of where you're going, but I don't think you really want a map. I mean, the producers probably want a map. It would be easier to work with, but it's really hard to make maps for one. Uh, and they quickly become inaccurate and need to be maintained. And, you know, I think it's easier and it's more fun to to just pick a direction, and it also puts you into places where you can't predict where you're going to be, which I think is part of what made Edith Finch, uh, such a strange experience is that we could not have expected that that's where we would end up. We only kind of

found our way there because we set off in a direction and just, you know, kept marching for, uh, for a lot of years. Uh, okay. But, try and balance this with the practical needs of of making a video game. It's not all, uh, you know, these like layers of abstraction and confusion and uh, we'll figure it out later. Uh, we're always trying to find things that we can put a stake in the ground on and say, you know, even though we are operating in this sea of unknown and this morass of poor decisions and, you know, shaky shaky ideas, uh, we can still pull out a few things that we can have certainty about or just like feel somewhat confident about. And for us, looking at things like weird fiction, this literary genre, um, and other experiences that evoke the sublime, you know, I think it became obvious that, for whatever reason, they tend to involve natural spaces. People feel a sense of sublime most

often in places like the forest or the beach. And so, yeah, we should probably be setting stories, uh, in those spaces where we can. So, it gave us a way to focus of all the potential prototypes we might do, you know, maybe we can try some of those at the beach. Uh, similarly, with weird fiction, a lot of the stories are very short, and a lot of them involve very explicit narrators, people that are telling the story, writing it down, often several layers deep, where someone is telling a story to somebody else. And that's a way of, uh, emphasizing that you're never getting the full truth, you're just getting a perspective on it, and making the universe that much more obviously unknowable to you. So, that was something that we took wholesale for Edith Finch. And there was about, I think, a year or so of us, or me, particularly like wrestling with like, "What does the sublime mean? Uh, like how do we make that concrete?" And

eventually kind of giving up and saying, "I don't really know how to do that, but I do know how to do overwhelmed." And that's pretty close. So, I think overwhelmed became something that's easier, like if an animator is making, you know, like a walk cycle or whatever, like you can visualize that a little better. And it's not exactly what we set out to make, it's not as aspirational as sublime, but it's something that's a lot easier for, you know, to actually make a concrete version of it. And, yeah, so next is talking about how, uh, once there's this idea of this, you know, feeling that you have, how do you communicate that with everybody else? And people like to ask for documents, uh, you know, like written descriptions of how this thing will work, uh, knowing that, of course, these things are going to change. And you know, people kind of read them, they forget about them, the documents, you know, become stale. They have some limited utility, but it does help to

write them, um, you know, partly just for your own benefit. And then, also, later on in development, a lot of people will come on to the project, like various contractors or, you know, like you might have, uh, you know, someone who's only going to be there for a couple of weeks doing, you know, like shrubbery or whatever it is that like that week you really need someone to do. And so, having the corpus of documents actually, even if it's not super helpful for communicating the original idea, is good to just have around, uh, for people that come on late. But, I think what works better is to try to communicate to everybody in the language that they're most comfortable with, which, unfortunately, is different for every single person on the team, cuz everybody is a snowflake. But, you know, kind of broadly speaking, uh, like artists tend to prefer like a more visual, uh, way of describing something, so like a collection of screenshots or, you know, like reference photos, uh, might make the idea of you know like your beach prototype a bit clearer where the artist really wants to know things like like how wet is it what time of day you

know that you might not even thinking about but your reference photos will have you know some concrete version of that like programmers tend to like feature lists but everybody has their own you know kind of thing that they're most concerned about so in terms of communicating a lot of it just boils down to talking individually to everybody on the team where that's possible and in our case that that even extended to task tracking where every single department by the end of the game ended up with their own kind of different task tracking tool because they had different ideas about how much you know work they wanted like how described it should be and so rather than having one monolithic task tracking tool that nobody really liked because we were a very small team it actually was easier for us to say like okay you guys can just work in Google Docs and you know we can use Jira and we can use the bug tracking tool and whatever which seems insane but that's part of the like one of the nice things about being a small team is that you can do kind of insane things and it's only

insane for like six of you it's not like 60 or 600 so you can mitigate that a little bit and then in terms of communicating to everybody else nothing beats like a prototype or anything that is moving a lot of it feels like an optical illusion where you can look at something and you know you swear like oh yeah that's really that's a hole but then you move a little bit and you're like oh no it's just a painting of a hole you know but it only works if you're the camera is like in one particular perspective and that's kind of what like a moving prototype does is it gives you a different perspective on the problem and it lets you ask more painful questions than you might otherwise if it was you know like a series of reference photos or whatever so ideally you know any kind of moving object is a great place for discussions and ends up solving a lot of questions or raising questions that there wouldn't have been raised otherwise

and it's not enough to just communicate initially what the sort of emotional goal is of something it also isn't incumbent on you to communicate that every time it changes which is like all of the time for very good reasons but often kind of complicated reasons like oh well players were getting confused about this thing so even though we really liked it we kind of had to remove it or move it over here and it's hard to tell everybody every time something changes so what we found was helpful were a couple of kind of best practices like team traditions like every Friday we would have show and tell where we'd walk around to each person's desk and the whole team would you know kind of look at whatever they're working on and then people would give like a one to two minute spiel and and show things off which is also really nice for people to brag about what they're working on cuz often if you know an artist is you know making some really complicated prop they could spend a week or two on it and it just you know goes into the asset folder and in the

background and nobody really notices it even though it's very pretty it gives it gives everybody a chance to kind of clap for them and also is a very explicit demonstration of like oh that's funny you're making you know props for this like Egypt level or whatever I had no idea we were going to do an Egypt level and it it just surfaces a lot of things that might otherwise you know go under the radar it's been my experience on both of the games we've made which turned out pretty well overall but there are phases in development where a lot of people don't play the game even though they may like it or they may hate it or whatever they're just long stretches of game go by where people get very busy on whatever you know their part of the game is and don't actually play the rest of the game so it can be helpful to force them to do that and we used to have these dog fooding sessions where the team would kind of eat its own dog food and for an hour or two everybody would

play the game themselves like at their desks and then we would have a meeting usually afterwards and sometimes we would generate tasks often we would just chat about it and it sounds really stupid but just having everybody play the game was a really good way of communicating what had changed about the game and you know what we were working on also similarly you know getting everybody together to watch a play test can be a good way to communicate things out I mean ideally people would know that you know you had added this level or changed this thing before they see somebody actually playing it but that's a good kind of last stop for oh yeah now that thing exists cuz I see it now it's live and similarly showing like for the publisher they'll ask you to to demo at various conventions that's a good place to to get people in front of other players and especially later on in development you can get kind of jaded to the game cuz you've been seeing it so long and getting a chance to see how

happy new players are with the things that you yourself have gotten very tired of can be can be inspirational also even though there are a lot of times where in our development process we have no idea what's going on we do have phases where we can make lists of like oh yeah we need like 300 things done like for the art team of like oh here like 300 problems and it feels really good to make that list and then see it go away and actually like chip away at it there's something that's very empowering about crossing things off and seeing things get completed which when you're working on something where you have this kind of broad direction but not a very specific road map for how to get there you don't have a lot of those opportunities so we would take advantage of things like E3 for example or you know anytime we had a big milestone build to turn that into this kind of smaller phase of certainty

and you know rather than use the sort of already really cumbersome overloaded task tracking tools that we all had that we'd been you know working with forever it was surprisingly powerful just to open up like a new document and say all right here's you know just 200 items that we're going to make today and we're going to see them all you know be done a week from now and you know we're just trying to balance the certainty and uncertainty but okay moving on to prototyping so once you have this sense of okay this is going to be interesting space to explore that creates a problem because there's so many ways to potentially explore it and the prototypes are where the rubber meets the road and they're a way for the ideas to actually start kind of talking back to you and to work with them in a very tangible way and to discover things often through repeated failure and this is the way you're looking for I

think for us you know because we were venturing into these places that we really had no idea like how like how a game about short stories for example could work by making a few of these examples and then watching a bunch of people play that we kind of developed our own internal sense of how these things would fit together and over time there's a way in which the game itself starts to kind of talk back to you or like another way of looking at it is that you've already made a lot of decisions intentionally or unintentionally and now you start to see the repercussions of those decisions and you know a lot of times those can be actually positive moments like for us two of the things that we discovered you know that wasn't that we planned these things but just in watching play tests that really changed the way the game of ultimately feels were you know when we saw play tests we found that a lot of players would get very

focused on whatever their objective was so that's kind of a traditional player behavior that is a little bit frustrating when you might want them to you know kind of open up and take their time and take a breath but you know because it's so ingrained in players to follow instructions we ended up making a number of stories that are kind of about that process of what happens when you get so focused on something that you ignore everything else that's going on in your life just because that's what we saw a lot of players kind of were headed towards without us even you know pushing them we also had just a very kind of garden variety design problem of you know most games are about trying to prevent your death how do you make a game where death is inevitable but players don't feel like you know that's a sign of failure so we tried to make those moments feel a little more positive and in doing that we created these kind of strange moments

where players would you know kind of march happily towards their death which was something that we hadn't experienced ourselves in any other games and felt like a very kind of canonical Finchian moment for us so that became something that we found in a couple of stories and really emphasized so it started as us just trying to solve a particular problem and then you know ended up being something that was more interesting and that we put you know more love into than uh than we'd expected okay jumping into some kind of nuts and bolts prototyping tips from the trenches the first suggestion I have is 3 by 5 index cards are a really good way when you're dealing with a A of ideas to try to get other parts of your brain involved. So, like there's a lot of this talk is really just about dealing with, you know, kind of this combination of complexity and confusion and the unknown. And these note cards are one

way that I personally found really helpful. Just drawing really, really crappy visual representations of whatever, you know, is um here. I think it's like motifs in the game trying to figure out like what are the kind of primary and what are secondary and what are bad. Uh and by making them visual, it engages this other part of your brain that is enormous. Like a huge part of our brain is dedicated to visual processing. Uh but if it's something that, you know, is just, you know, a tech document or in your head, uh there's a limit to what that part of your brain can work on. Like uh I think seven plus or minus two is about the things you can keep in your head at any given time. But if you put it out into the world in some way, uh you can think about a few more things, which is helpful when, you know, you have a game that's about, in this case, like 13 different family members. Uh here's another version of that uh in a spreadsheet form, where uh I I looked at this document like pretty much every day

for the last 2 years of the project. It's just a really terse summary of all of the different stories and what the order is and some of the different, um you know, values here, like the gender of the story and trying to balance those things out. And just trying to kind of get my head around what this whole game was going to feel like. And obviously these pictures, you know, don't really look like the stories, but they're enough of a shorthand that, you know, you can kind of imagine what a whole game's worth might kind of feel like and run through it in your head uh just by making it more visual. Uh also, just a note about the brain. I think it's really helpful to write down all of your ideas because if you don't, then your brain tends to hold onto them. I think your brain is like a second grader who's got their hand up and really wants to like tell you the answer, but if you don't call on that second grader, then eventually they just kind of shut down. And what I had a lot of the time the solution for a problem would come when I was in the shower or

doing something else and not really thinking about it actively. But the only way you get to that point, I think, is if you reward your brain by saying like, "Yeah, I'm listening to you. I'm writing these things down." Even these ideas that don't seem like they're that great, there's a power of just getting it out of your head also. So, even if it's not a great idea, uh like it makes space for something that might be coming up next that that could be great. Uh also, even though game development is very collaborative, I think early on when you're making prototypes, it can be really useful to just do everything on your own cuz as soon as you involve anyone else, there's just like a a lot of complexity that you're borrowing that can be really helpful if it's something very technical that uh or difficult or, you know, that that you really don't know how to do. But often you can get a lot of the way there just knowing a little bit about a couple of programs. Like on Edith Finch, it's like a 3D modeling tool, like a very basic knowledge of uh of Unity and some Photoshop is enough to get a lot of

questions answered positively or negatively before you uh you go out to somebody else. And then there's, you know, beyond that, tons of things in game development that you could be learning that are are really useful for uh for making prototypes. Uh also, when we make prototypes, the very last phase is actually making the prototype itself. Uh a lot of the time is spent mm kind of in the uh I don't like phases where you're dreaming about it in different ways. So, just like downloading a whole bunch of images uh and finding reference is is a really great way of getting, you know, your brain kind of spinning on possibilities. Uh particularly with reference photos of real-world things. Like if you're doing a father-daughter camping trip in the 1970s, uh you know, you might stumble across an image of like, "Oh, right. They had, you know, different kinds of cans for opening soda back then with little peel tabs and stuff that you just like wouldn't normally think about, but finding references can give you a lot of

kind of free ideas and uh and help you to uh think of things in different ways. Uh we also did a lot of work in Unity even though the game itself was made in Unreal. It's uh I don't I think kind of nice to work in a completely different space because there's a lot less pressure in some ways. Like it you know it's going to look really shitty if you did all of the art. And yet it actually like, you know, in this particular case of the prototype for the swing set story, you know, it starts to get to a point where you can ask some, you know, valid, troubling questions about should this story have hands, for example? You know, should the um the rope, you know, be bending at all? Should that rope instead be a chain? Uh a lot of questions that, you know, once you start to get things moving at all, uh rear their ugly head that uh you know, might not have been been visible if you were just on a whiteboard. And then the last thing we do is is engine prototypes cuz they're a little bit more heavyweight. Uh okay. Critically damped springs.

That's um the like one-dimensional mathy sort of value for how you move something like basically like a real spring. That we used everywhere and particularly on cameras, places where you want really smooth motion. They're things that I think worked really well well for prototyping because they're so simple. So, I'll just talk really briefly about them, but here you can see like on the top, this is just like it's a box moving. You can see at the very end it just kind of clunks like that. But a spring is just this tiny little piece of code and it feels much more expensive, much more magical. Uh and the other great thing about springs is that they're really easy to tune cuz there's only two values, like how springy they are and then whether or not they overshoot their target. Uh so, even though they look really nice, they are also really easy to adjust for designers later on in the process.

And here's an example of taking like just like a raw input. Like if you have the analog stick moving around, uh sometimes you want that snappiness, but often it feels better if there's a little bit of a delay uh particularly with cameras. So, here's an example. I think PowerPoint is a little slow there. But yeah, so on the left it's just like what you get if you listen to the stick and you say, "Okay, oh, they moved left. Great. Move the camera left that fast." Uh and you see it a lot of Unity trailers uh for games with cameras that look just like that. Like they're very jerky. Uh and these springs are so simple, they're just a couple lines of code, and they start to feel much more like a finished product. Um it doesn't have to be this floaty. Uh this is, you know, kind of closer to the Edith Finch floaty camera that I personally like. But the nice thing about springs is that, you know, there's two values and you can tune those knobs. Uh and because we're making so many different prototypes, tools like uh

these springs were really useful cuz we could drop them into all kinds of different scenarios and then still have something that, you know, we could easily adjust the feel of. Uh and then music is another thing that was surprisingly useful for the prototypes that we made. Um probably because composers are really good at evoking feeling. They are kind of professionally obligated to listen to people ask them for something, you know, to be 20% more whimsical and then have a cogent response to that. Uh they're much more professional than game developers in that way. Uh so, our composer was really helpful when we had problems in certain stories where it's like, "Yeah, it's it's close, but it's not quite there." Uh you know, there's very surgical level of uh of tuning at that point. And it's also something that is very practical to adjust later on. Like as the door begins to close and you get closer and closer

to shipping, there's a lot of things that you can no longer change. And everyone agrees that yeah, the game would be a lot better if we could, you know, get rid of this thing or that or whatever, but it's like we can't for memory reasons or, you know, cuz it's in a sensitive area. But music is one of the things that's almost free. Usually, I mean, there's a little bit of streaming cost, but uh in terms of technical complexity, it's very low. And yet it has a very significant emotional impact. So, for us, there were a few stories uh like swing set for Gus's Calvin's story. Um that uh that made a really big impact on and and came in late. But I think it's especially true for Gregory's story, where we had, I think, four different gameplay programmers uh try to work on it. And it just never felt right. Like it felt too clunky and we got very focused on the mechanical aspects of it. And it's a difficult problem. Like you're doing kind of 3D platforming in a space where the camera is fixed. You're Gregory in the scene, like you're a baby. It's very

hard to judge depth. And so we spent a lot of time adjusting that and thinking about like what should we be doing and why does this not feel whimsical? And, you know, I think the best answer, I mean, there were a lot of legitimate gameplay fixes that came in that helped a lot. But in in terms of the emotional feel of that story, uh just adding the waltz of the flowers got us a lot of the way there. Uh something that felt, you know, like people have associations with ballet, uh put us into the right mindset of being this, you know, kind of balletic frog jumping around a tub. Uh okay. Play tests are a really nice phase because it's a way of moving out of this kind of forest of the unknown into a place of a lot of practical and tractable solutions and fixes. So, there are a lot of things that don't have clear answers in the design process and you can, you know, stay up very late wondering like, "Oh man, there's so many different ways we could do that." But,

on the other side, there are a lot of issues that are quite amenable to just brute forcing. Like, pathing I think is is a great example of that of if players, you know, tend to like you want them to go say right down a hallway, but they keep going left, then it's something that you can just do a bunch of play tests and keep making little adjustments to that. Like, "Oh, maybe make this, you know, a little bit brighter. Maybe add a little bit of motion." Or, you know, however many revolutions it takes, if you just keep doing more play testing, it's like turning the crank and things eventually get better. And it's really kind of comforting that there are phases and aspects of development that are amenable to that because again, a lot of time that we spend floundering in places that are very murky, but play tests are are a great place to just kind of not worry about these problems. Like, they may be legitimate issues like pathing and, you know, people getting confused about something, but there's something that you can just work hard enough on and it will eventually solve itself.

A few sort of specific takeaways, when people who are playing get frustrated, often they will have an idea about why they got frustrated and they'll have, you know, suggestions for what you should change about your game. Generally, it's a bad idea to listen to them because they're unaware of what has led them to that state. Often, we would see players who would get very frustrated if they, for example, if they're playing Molly's story and they didn't catch enough didn't catch any owls. So, they spent 20 minutes flying around trying to catch their as the owl trying to catch the rabbits and then they would complain about the next sequence of the game when you become a shark. It's like, "Oh, the shark sequence is really boring." It's like, "Well, it's only boring because you like exhausted yourself in the sequence like right before it." But, people usually aren't aware enough about that. So, yeah, it's it's useful to listen to that they are upset, but not usually helpful to listen to why they think they're upset.

In a first-person game, it's also really helpful to look at where their camera is pointing. You can kind of mind read a bit there and and get a sense of what they're concerned about or what they think is important in the scene. Repeating anything tends to be kind of a death knell. I don't think I can think of any genre where doing something a second time is going to pay off, but that's kind of the last step of of desperate players. And then, sometimes people have a lot of trouble with the game and it's because they're very dumb. And it's kind of heartbreaking when you watch them play and like one one of our play tests, someone spent I think like 20 minutes on the downstairs of the house, a sequence that usually takes maybe two or three minutes. They went upstairs, they missed the part they were supposed to go, they went downstairs, went upstairs again, downstairs. And then, I think like the fourth time they they found the thing they were supposed to. And I thought they'd be really upset, but they oops,

getting ahead of myself. But, they were actually really proud of themselves. Oh yeah, no, I found my way through it. And I think, you know, there are a lot of people out there that just have trouble with life and your game is like the latest manifestation of that, but it's not your fault. So, you know, it takes a little bit of um you know, detachment to say like, "Okay, what's my bad and what's just like their fault." But, if they're having a lot of trouble and they're not upset about it, that's not necessarily a bad thing. That just might be like how they go through life. So, that you know, takes an expert's eye to to suss out. And then, we would alternate our play tests between doing kind of functional tests where we'd have one person play through with more what we called focus tests where we would have like up to five or 10 people play at any given time. And those were really useful I think for testing different things. Like, the one person test was

really good for very specific mechanical issues of like what might be confusing to a player where the larger tests were really good for more subjective like, how does this kind of feel? Do these things make sense from a narrative standpoint? So, doing kind of mix of those play tests worked worked well and I think about five players is probably as much as you need. After that, it's diminishing returns. Something that was also really helpful for us was doing batches of play tests at conferences like GDC where you've got thousands of game developers from all over the world who were here for one week and you can, you know, call them up ahead of time. And a lot of people are really happy because they're already in town, you know, to come by your hotel room. So, I think like last year or the year before, we had like maybe 10 or 15 play tests with tons of really great game developers, you know, in in a single week that would have been very difficult if we'd had to organize it in Los Angeles in the normal way.

Okay, and then just a note on frustration in particular is something that we focused on a lot with our play tests because more than anything else, frustration kind of unravels all of the good work that you might be doing elsewhere. It's like the scene in The Matrix when Neo like wakes up in this amniotic fluid. Like, it just completely like rips you out of this world and it's something that can be difficult to get players back into. So, our fixes for that, a couple of them, where it was possible, things like dynamic difficulty. The owl was a case of that where some players had a lot of trouble catching the rabbits. And so, if players, you know, failed a couple of times, then the radius of acceptable you know, triggering got larger and larger. And that's something where we wanted to preserve we didn't want to make it too easy for all of the players cuz I think some players really like to have a little bit of challenge and there

certainly isn't that much of any Edith Finch. So, we the places where we had challenge existing, we kind of wanted to hold on to that. But, there are some players that, you know, would spend 30 or 40 minutes potentially on those sections. It just felt like, "Wow, that was a terrible experience. Let's try to help the people that are having trouble without hurting the people that, you know, are doing just fine." So, dynamic difficulty was a way to do that. There are a few places where kind of like there were two ways or more that kind of made sense. Like, for example, when players come into a ladder that's below them, they might press forward on the stick because they want to go like forward and down the ladder. Or, it might make sense in their head to press down on the stick because they're going down a ladder. Either of those kind of work, but what we ended up doing was supporting both of them. So, when you get close enough to the ladder, whatever direction you press, we assume you want to go down the ladder. So, we just adjust the you know, kind of controls to conform to what players expect and then everyone

gets the experience of, "Oh yeah, I just sort of expected it would work like that." And it did. Rather than 50% kind of thinking it would work and then being, you know, surprised it worked the other way. So, if there's no reason it has to work one way, it can be nice to just let it work the way the players imagine that it should. And then, even though there are some people that struggle and I think it's fair to just say like, "Ah, it's it's fine. You can you can struggle." There are times where people will struggle and then never get through it. Places like actually even in something as simple as the swing set story, most players figure out, "Okay, if I move my right stick, like that'll move my right leg and the left stick will move my left leg." But, some players just never figured that out because they got the feedback of moving one of the legs and they just got so focused on that leg that they never thought of moving the other leg. So, the fix for us was to after I think it's like a minute or two, we eventually spawn some little letters that swirl

around the leg that you haven't moved just to try to get people to kick that other leg. And I think that's something that, you know, it seems really stupid when you watch somebody not figure it out when so many other people have had no trouble with it, but it's the kind of thing where everybody has those moments. I don't think you have to be too much of an outlier. But, there are places where like about 5% of the time, people would just, you know, not because they were necessarily being dumb, but it just it didn't click with them. And for a game like Edith Finch, you know, if you have 13 stories or whatever, like 5% failure rate on any of those stories is silly. That's going to add up over time. So, we tried to find places where if there was going to be some kind of hard failure where people would get really really stuck, you know, we would lend a helping hand if they spent a minute or two without moving forward. And then, some of those cases too where most people will never see that and it won't bother them that, "Ah, you're offering this little hint for something when I could have figured it out." But, after a minute or two of not figuring out something very obvious,

just kind of throwing them a bone helped move things along. And then, the last tip is kind of part and parcel of this the difficulty of working with things that you don't understand is that it can be really effective to change a lot of things at the end when you do understand them. So, I think at the end of the project, you're at this point where, you know, you've been struggling and suddenly things are kind of clear and you also have the ability to affect the changes you want, but there's not a whole lot of time. Although, hopefully by this point, you do have a sense of what are the things that you can change without breaking everything else. So, you can be a little bit more surgical. Also, even though we we did make a lot of changes, it's a kind of rolling phase on the project where as you get closer and closer, then the scope of things that you can, you know, make changes to does get smaller over time. But, I just want to jump into a couple of specific cases of places where I think the game

really benefited from the team continuing to be open to embracing uh those kind of changes. So, uh I can one of the bigger changes was a week before we did our final dialogue recording, we did our last pass on Edith VO and rewrote like a huge chunk, uh nearly half I think or more of what's in the game because we finally understood kind of at a macro level what the concerns were and, you know, what players were confused about. And a lot of it came down to very simple things like just adding mom and uncle uh and like being very clear about the familial relationships. Uh so, we changed a whole lot of dialogue in small ways uh at the end. And we also, not quite at the very end, but a few months before we finished the game, uh we did our last bedroom. So, we kind of worked in stages from the ground floor working up to the top. And or maybe like almost nearly last bedroom uh for Lewis's room. And we decided to

make him uh a little bit more of a character. Like, he's he's more of an overt uh archetype. He's like very much like a, you know, pothead. And we had resisted doing that on earlier bedrooms because we really wanted these family members to feel like distinct, real uh human beings and didn't want them to feel in any way cartoonish. But, you know, by the end, what we realized is that in order to understand someone, like if the challenge is understanding them from their bedroom, then a regular person generally would fail that. Like, I know my bedroom is not that expressive of who I am. I mean, there's like a few details, but it's nowhere near what Edith Finch needed to be in order for players to cruise through and in like a minute or two, you know, pull out a couple of high-level details about who that character was. Uh so, for the very last bedroom, we finally realized like, "Oh, right. This is what we should have been doing the whole time." And then we went back and redid every single other bedroom. Uh and a lot of the characters like Barbara,

for example, you know, went from being just an actor, like someone who likes performing, to being a child star, which is like the, you know, more charactered version of that, but also, you know, a lot easier to design that bedroom for and have people understand immediately who that character was. And then the appropriately, the uh the last thing to talk about is the ending of the game, which uh took up a good year or so of my life uh making terrible prototypes, uh things that did uh seem promising and then just didn't coalesce. And we realized after a while that the problem was essentially we were trying to write the last episode of The Twilight Zone. Like, we had created all of these weird, you know, bespoke, unique stories, and then we wanted something at the end that would kind of summarize all that and put a bow on it. And that just doesn't really exist. Like, the last episode of The Twilight Zone is called The Bewitching Pool, and it's terrible.

Uh it doesn't really answer anything. Um but, that was the the problem we had, and the solution for us ended up coming when we just kind of randomly had a request from uh our friend Jenova to come by and play the game cuz he had a friend in town and uh didn't have anything to do with him that weekend or that afternoon. So, they came by and they played the game. I don't think there was an ending at that point, but, you know, we talked about this problem that we had. And then one of them said, "Well, you know, there's like a lot of death in this game. What if, you know, the ending was something about a birth?" I was like, "Oh, yeah. That actually might work." Uh and, you know, part of me was a little frustrated at that point cuz it's like, "Well, how did they see these things? Like, who are these amazing people that can, you know, solve our problems for us?" And I think that the miracle here is really, you know, what it's like to be at the end of something. Like, you you go through all

these confusing phases, but kind of like there are enough pieces there that someone can come in and, you know, ideally it would be you, but, you know, sometimes it's not enough for you to make that. Like, you can't see it. You're too close to it. But, someone else, you know, who's not been there for several years staring at it, uh it becomes immediately obvious that, "Oh, yeah. Yeah. All these pieces are there." So, yeah, I think there's this kind of miraculous phase at the end of projects where even if you've been working in, you know, the dark for years, uh you know, it helps to be flexible just a little bit longer. And Edith Finch benefited a huge amount from that. Uh it is really difficult on everybody to remain flexible throughout that process. Uh so, you know, huge thanks to everybody on the team for putting up with all of that and uh and being excited most of the time uh to, you know, keep making those

changes. Uh it's also pretty rough on publishers. So, thank you to all of our publisher friends. And uh you know, thanks to the press, too. It's been amazing with Edith Finch to see that even like a year after the game came out, people are writing articles on like Edith's glove. Like, there's a whole article just on her glove. Uh and that level of detail has been really surprising and wonderful. And then uh you know, thanks obviously to everybody who's played the game and you know, really opened themselves up to it and had very emotional experiences with the game that are, you know, it's a very collaborative effort on the part of the game developer and the player. I mean, it only really works when people are coming to the game, you know, and being emotionally open to it, and we sincerely appreciate it. Uh and then we're also in the very, very early stages on on our next game, which is going to be super animation focused. So, if there's anybody out there who's an animator or an animation programmer or knows good uh people of those type, uh definitely let them uh know that we're looking. And I think we've got a few

minutes left for QA. So, if there's any questions? Uh hey, great talk. Thank you very much for this. Uh I had a question about your uh development process. Uh it seems very uh iterative and uh explorative and uh uh obviously, no one has unlimited time and money to make a video game. Uh so, I've been wondering like you you talked a lot about like exploring and prototyping. Uh do you have any tips for when you have to stop and just execute on a plan? And or did you just timebox it? Like, what what's your technique there? Yeah, I mean, our technique was to rely on our publisher to set milestones. Um I mean, working with them, but I think that was one of the really useful aspects of that relationship that we would have things like E3 or like that greenlight trailer I showed, for example, was that was a publisher deciding. I mean, we we talked to them about the scheduling, but

I think it is really helpful to have milestones. And it would be very difficult if you didn't have a publisher like forcing you to those milestones, but that kind of discipline would be really useful just to say like, "Okay, we're going to show at this conference or do whatever." Like, nothing really I mean, it's kind of like the miracle at the end of the project. It's like that for every milestone when you have this big goal. It gets everybody kind of together working on, you know, achieving some nugget of it. But, yeah, I think it just comes down to you need little bits of crisis to bring everything together and get all of the check-ins submitted and, you know, integrated together um cuz otherwise you'd never ship. But, yeah, look for any way you can add little bits of crisis. All right. Thank you. Yeah. Um so, as a publisher, that terrified me cuz looking at the talk, I'm like, "Oh, that's cool. Every team uses their own tracking." And I'm just terrifying. So, while the end product was amazing, like was there sort of like you you say there was milestones, but was there elements of crisis as well where

because it was like not really formalized and kind of rough and kind of very sort of a humanistic process. So, was there elements of terror or was there a bad part of it as well, I guess? Uh yeah, I mean, there were definitely moments where the publisher was very concerned. Both of the publishers. But, I think we it's not like we were working in a vacuum and just kind of like in the dark the whole time. Uh every week we would have new cool things that would come out of it. So, we could show the publisher, "Hey, you know, we thought it was going to take us this long and we would have this thing, but instead it's taking us this long and we have this other really cool thing." So, yeah, I mean, I guess to answer your question, we uh we didn't always hit all of our milestones, but we did always find interesting things along the way. And so, I think that was uh somewhat comforting to the publisher to say like, "Okay, well, it does feel like like good progress is being made and it's not very predictable, but the

publishers also kind of understood going in that that was part of the development process and kind of like what they were interested in making, also. It's like they had signed up to publish a game that nobody really understood. Like, that's part of what was appealing about that. But, that would not have worked if we were doing, you know, a Halo sequel or whatever. Uh you know, where there were much more, you know, fixed ideas. Cool. Thanks. Sure. Um sorry. Oh, is there in the back? Yes. Um Go ahead. Oh, you. Oh, me. Oh, Sorry about that. Um So, you talked about your mood boards a lot and they kind of interest me and I was wondering how much they evolved throughout the project and or were they just kind of like a Were they like your base of like it was like a stone for you guys or was it more fluid? Uh we didn't continue to make mood boards per se. I think that was an early thing. Like it takes a while, you know, to put all those things together and and look at them, but we did something very similar where

like Chris, our level designer, would put together, you know, like a a couple pages of a bunch of different photos and ideas and references for, you know, kind of like what a middle school girl might make for like a crush or just like all these cool ideas about a thing and put them together. A little bit more structured, in Chris's defense. Um but yeah, so they weren't like mood boards per se, but they were kind of in that vein when they were really helpful when design needed to communicate something to art was usually where they would come into play. Okay, so the mood boards were more like What would you say like kind of started with then? Yeah, yeah, that was more like what is this feeling that we're thinking about? Like what are the kind of tent poles that we think are important here? But we didn't keep making them in the same way that we didn't keep making like a game design document for the whole game. That was something like initially you take a pass at it and then the game itself becomes the document eventually. Awesome. Thank you. Sure. Hi. So, in a game like this that it's very abstract and with a lot of feeling based

design ideas, how do you manage to balance keeping the vision that you have as a creative director and not like micromanaging the team and giving some free to the to the team to develop their own ideas? Oh yeah, I don't know if I have. That's not really a question that I can answer one way or the other, but you know, I feel like the attempt was to try to define things, I guess, in very specific kind of directions of like okay, we want it to feel like this, but not necessarily prescribe this is how we get there, but you know, to leave things open. Like I think coming up with three words is a good example of a place where, you know, you can be specific about what you're looking for, but you know, not being too prescriptive and I guess that just, you know, comes down to really like defining the task and hopefully leaving enough room there. But I mean, I think as a creative director, one of the things that I've struggled a

lot with is people that don't want that freedom. Like that's I think it's easier when people really are excited about putting their own stamp on something. Like assuming that they understand what the goals are, but I think a lot of times people on the team, you know, especially in our art department, they really just wanted to be told what to make, which is quite surprising. So, I think more so than like the challenge of giving people creative freedom, it's the challenge of understanding what kind of freedom people really want and then giving that to them. Thank you. Sure. In the back. Hi. Um so, you said don't give them a gun, right? Uh and I understand that part of the feeling that you're trying to evoke, you said, is sort of being overwhelmed and so you're constantly presenting these new gameplay ideas to the player and that helps you tell the story. So, how deep and complex of a mechanic could you

present before you'd say it starts to distract from your ability to tell a story? Hmm. Um Yeah, it's a good question, I guess. Because I noticed it was part of your design philosophy, but it was also part of your emotional message. So, it's like when you design another thing with another emotional message, can you make deeper mechanics? Or will you lose your player? Yeah, I mean, we tended to for our prototypes settle on fairly simple mechanics. In some ways, you know, like they feel embarrassingly simple and you know, like it like Lewis is a good example of a, you know, mechanic that initially was was very complicated when we were on the whiteboard describing like how fish chopping should work and you know, there there was going to be a boss who was going to come around and there were going to be fish that you didn't want to chop and fish that you did. And what we found was just like in play testing it, the player's focus was so divided. So, in that particular story, there just

wasn't a lot of bandwidth. So, I guess, you know, in answer to the question, we just would like try something and then we would see how much emotional kind of head space the player had and trying to be conscious of the things that are taking from that stockpile. Like there's like a limited amount of emotional attention that a player can give. And on Edith Finch, one of the things we were surprised about was when we would do these longer play tests with people playing the game, you know, like beginning to end, that when they came into a story, they weren't even really fresh. Like they were still thinking about the story before and the story before that. And so, you know, in testing them in isolation, we would have a different experience from someone, you know, playing it all the way through. So, like in our case, we tried to encourage players to stop after stories. Like we would open the family tree and Edith would talk a little bit and, you know, there'd be a little drawing just as a way of getting some of that attention back because the stories for a lot of players were kind of emotionally draining.

So, it wasn't that we were like designing like complicated things to a certain level of complexity that we thought would be appropriate for the game. It was more like just being budget. Yeah, yeah, like taking the temperature in play tests and trying to figure out like what, you know, kind of made sense. And in our case, like it's kind of embarrassing how simple we, you know, could make things and get away with it where no one complains that fish chopping, you know, needed to be more complicated because it's not really about the fish chopping. It's that is getting you to something else and you're thinking about all these other things and our job is to kind of set the table for you in this in this game anyway, that it was more about, you know, this context and creating that than it was about, you know, digging really deeply into a particular mechanic. Okay. Thank you so much. Sure. Hi. Thanks for a great talk and for making game where fish chopping is central mechanic. That's amazing. I love that sequence. Um so, my question is was

there any prototype that you really really loved, but that didn't make it into the game because it didn't fit? Um there wasn't anything that I'm heartbroken about. I mean, there are stories that we tried really hard on. Like a story where you lean back and then look at the world upside down and things fall away from the earth where we spent a long time, but I think we got really good after a while painful lessons of realizing what we as a team did well and what was kind of like I think there are some stories that work right away and it feels like you're bicycling downhill. And then other stories, even though they may seem really cool, you just find even after, you know, a couple of days of prototyping and then it's like you're pedaling uphill and you're like wow, this is really fun, but I don't know if I want to keep pedaling for 6 months. So, nothing that is like oh, if only we had time. I mean, Lewis's story and Gregory's story almost got cut. So, those would have been like oh, man, yeah, really wish we had these stories. You wouldn't believe

what we had. But no, nothing that yeah, I regret too much. Thank you. Sure. Okay, last question, I think. Um I really loved what you said about like discovering what the game wants to be through the prototyping process. I'm curious if you had any situations where in early prototyping you had sort of like variations on a theme, like different prototypes that tested well that fit within the overarching story, but were like directly in competition with one another that sort of like pulled it in different directions, but felt equally compelling to you as designers. And if you had that situation, do you have any best practices for kind of like sifting through the data and figuring out what to up res? Yeah, I don't think we had any prototypes where we liked both of them and they were like competing with each other. Usually it was more the case of like like the pedaling downhill thing. Like oh, well, this is clearly like a story that we should be doing more of

and this other one, you know, is really frustrating. But I mean, certainly looking back on the original design, it seems crazy that we would have spells or combat or things like that, but we never really got that far into it. I think we we just kept, you know, we made this pile of ideas and then we would take the most interesting idea off the top of it for the next prototype. So, we just kept, you know, kind of sorting that pile. And if we had been making prototypes from the middle of the pile or the bottom, you know, like the spell casting ideas that we had, like I'm sure we would have run into, you know, more conflictory stuff, but I think just the evolutionary process of what should we work on next was enough to weed out a lot of the conflicting ideas. We also early on, like our first two prototypes, like the large one ended up being the largest, most complicated stories we have in the game, Molly's story and Lewis's story. So, in terms of like how we prototype things, we kind of made stories that we really liked, but

were in hindsight like insane that it would not just be like one mechanic, but oh my god, there's like five different animals that you get to be in this story and, you know, these are like stories that were combinations where we kind of got away from that from a production standpoint. But I mean, it's nice that we had them because we could take those and eventually ship them, but we could only do that because we'd started them really early and we had years and years to keep working on those stories. We couldn't then we continued to make prototypes like that. So in that way like they were successful. We like those stories, but they were also cautionary tales of like let us never do this again. Um we can you know we have space for two of these in our hearts, but that is all. Thank you. Sure. All right. Uh I think that's the show.