How We Created Mark of the Ninja Without (Totally) Losing Our Minds

[Music] so I was asked to introduce myself first so I'm Jimmy Chang I found clay in 2005 and we've been building independent games since then our last two games with marking the ninja and don't starve and I wanted to talk about the the development process today a little bit I'm the creative director at clay entertainment there's pretty much the art dude okay so I want to talk about what we want what clay is about and you know what we've been trying to build over time and you know for me I wanted clay to be a place where we're continually building amazing games over and over and so Jeff and I

have been doing this for quite some time and the only way that we can actually see this actually happening is if the process is same is if the process is sustainable so not only do we need to build great games we actually need to do it while still going home and still not not ruining our lives in the process and you know it comes it's basically inedible inevitable I mean for me I actually just flew in today which is why I'm a little out of it I had a 2 a.m. feed I have a two week old at home right now and I flew in today I'm flying out tomorrow jeff has a one-year-old Friday yeah their birthdays on Friday so have to fly out early and all of us are getting older and it just you know for the last three years we've been thinking about this you know that we know that everyone's going to get older everyone's priorities are going to change and we need some changes in order for that to happen and you know for me I find it

just ingenious when somebody says the reason why we worked a whole lot of overtime and the reason why we killed everybody working in this game is because we're doing something new because we're doing art and to hide behind art as a shield to kill you know yourself and your employees and your the people you work with I think is not the right thing to do and I think you're you know going to screw with the development in the future for that so while there's no formula to building a new game how we see it is that there are theories that we can use to help us better understand how game development works so this is a little bit of how you know how much I guess you can call it over time what I would say is insanity happened during the development of our different games and you can kind of see this downward slide and shank one was was the one where basically it was a tipping point for me

it was so stressful and such hard work in such a short amount of time basically I made a promise to myself that I would never put myself or other people in that position again so you can see it's downward slide from there and you can see a little up on mark the ninja and the important thing for me is not that we shouldn't work really hard but that I see that as a failure and I see that you know we can do better so what we end up asking is you know what are the biggest wastes of time as we're working on development and so les it's actually just building the wrong thing so it's not even about the trees or the leaves it's like this analogy where people are in the forest and they're chopping down the trees and they're trying to work really hard and then somebody climbs up to the top of the forest and says hey we're in the wrong forest and the other guys are like but we're making so much

progress down here it'd be a waste if we move to another forest and we laugh because you know we see that alive because we're building these features and we're like well we don't want to cut that because you know we've already spent so much time on it and so you know for us the the the idea is how do we figure out that we're in the wrong force quicker because when you're working on something new the nature of it is that you're gonna do the wrong thing often but the goal is to recognize that quicker and get the hell out of there so I want to start with talking about some of the massive failures that we had and you know between March and July was when we started ninja and even then you know people say that you need to have this vision of your game and you follow your vision and you'll you'll be fine but even then we already had that vision we had the vision that the ninja is a is a stealth game where you feel like a real ninja so you're a glass cannon and you're flitting in the shadows and the enemy never even knows you're there when

we made the first pitch we already had that name written down but the problem was we had no idea how to actually get there and so we had all these assumptions that we were building all that we thought we would work and none of it was working we just we just felt so we actually had light and dark and we had stealth kills we had shuriken as we had the sound all this stuff but it actually felt all mushy and it just sucked and in fact it got so bad that you know around around May we actually thought this isn't actually gonna work at all we can't build a stealth game so for a couple months we decided that Martin a ninja wasn't going to be a stealth game primarily anymore and that it was going to be an action game with stealth elements because we thought this wasn't actually going to work and of course that made the game even worse so then we took it back and we thought again what what is wrong with what we're doing and what we

were wrong what was wrong was is that we were building a whole bunch of stuff based on bad assumptions and so what was going on was we were saying that we think that you know these fire propagation is going to have it was going to work or that if we build more features the game will get better but it was all based on assumptions that we weren't clear about so instead what we did was we went back and we said I'm gonna build testable hypotheses I'm gonna see I'm gonna say this is what I actually believe and that's tested out and so we went through tons and tons of different little scenarios and said I think that for example if I hide behind this this is as vaz and then of a guard comes along I'm going to feel stealthy right and so we build all these scenarios and ways and we were testing them out and and trying to fail quickly and once we started doing that that's when we could actually start to see which of our some sort of right and which our assumptions were wrong so I'm going to show you a little video of something that we did when we as we

started doing that this this is you know somewhere in the summer where we started testing out our hypothesis and and these are the ones that actually worked [Music] who's there a little line up there that's talking about exactly what we're trying to accomplish we're going along here we're trying multi targeting and slowing [Music]

here we're getting people to work we decided was our hypothesis was hey you know people don't know the information but can't learn it that's fun that's that feel sneaky and so we're trying that out here's something not to kill them and just distract them we need to check this out and that's the Canadian Alta I thought I heard something I think we probably built about two dozen of them before we started really realizing that the points of the game is rather mechanically the point is to plan it was to observe the environment work for your information plan your attack actually execute that

and the planning part was the most important part and it was having this predictable cause-and-effect and once that was clear and we could test that and it was it was it was working then kind of the game started designing itself and all of these decisions just kind of worked itself out so for example you know should we have binary light sources yes because that makes it more clear should we have you know should we show the sound rings well when at first when I I remember saying hey we really should show the sound rings and you know there was some some murmur because it was like well MIT does that make the game too easy you know shouldn't you be able to dn't you have to work to figure out how big your distance is and the answer is no because that makes it mushy it makes it it makes it unclear how many steps forward I can plan right and so when you make it super clear then you're able to plan further ahead and do higher-level planning and then another really good example is the AI because at

first we had these really complex systems of what the AI was gonna do and we're having a really hard time implementing it or rather we're having a really hard time getting the player to understand what the heck I was doing and when this idea this theory that people need to know the cause and effect very very clearly became clear it was it was it was very clear that we needed stupid or AI and we needed AI which was really manipulatable so you know this is this is one of those things which actually helps cut out work and I'd say that cutting out you know smart AI cut out probably you know four or five months worth of a programmer worth just there and then this translates totally into the the art process and making things clear to the player like so basically it went from the theory of like okay let's let's take things that make it look cool - okay how do we translate all this information how do you translate all this data to the player so certain things like okay

binary light systems like if you're an artist you know like okay if I walk into a light I could get like shadow up to here and then I'm like partially seen so we actually change that to like if you have one foot in the light you're seeing if you're not if you don't you're unseen so it's basically player knows at all times if they're in shadow or if the non shadow another point was if we had like the UI aiming all the guides that we have in the game we made it like super precise super accurate just so people could plan and execute their their actions to a tee and and the one the one key point that Jamie was talking about too is the AI being able to have predictable AI and from an art standpoint how do you do that how do you translate those emotions and stuff like that so we did like we have really talented artists actually so we did a whole bunch of like facial animation we we scaled up there like expressions to make it readable but we found like the most critical thing or the easiest thing to do was just act but like emotional icons over them so they everybody knows

oh that person's in a terror state that person's in a in a in the investigate state but one of the the big elements here were playing with our theory of making it clear was also like where the ninja can go like we were playing with this idea of like the artists were like oh should we do it like shank where it was like like there's depth to the platforms and stuff like that but the key thing was like we had to get the player to know like oh I could go on that ceiling part I could go on these walls so we decided to keep it completely flat and that was just the result of of just testing these theories like the more the easier is the fear it is for the player the better it is for the game so not only did these theories of like create a theory test the theory and learn from all our failures apply to game design and like what I call visual language for a game it applied to our entire process so that was a weird and awesome kind of learning for us where we could basically test our theories with

production and and and move on and see if like that actually worked if like everybody in the production process enjoyed it and and we created a good effective product at the end so this is an example of us when we had the when we were creating a level so we wanted to test this process our theory and we wanted to fail quickly so like we got nails and the team to do like one level to create all the scenarios great all the cool like stealth moments were you sneaking up and guards going through events and all that stuff but the one thing was it wasn't a world like when the artist got it I even put it like I even tried painting over it and it's like it what like we couldn't make these into like an actual world like is this supposed to be like super vented like like facility like how do you how do we actually make this in the world so we're like the artists we're struggling with it so we're like okay let's let's fail at that like that's let's and let's get the artists to actually create the structures let's get the artists to

create the buildings like knowing that we have to put vents everywhere knowing we have to put like different like rooms and doors and so and then we passed it off to the game designers so the game designers knew exactly how their scenarios worked they had their process planned out but they're able to use it with it within our structure of artists created buildings old bunch of work was like moved around deleted for and pulled to make the final game but at least they had a basis it removed the whole blank canvas structure that the the game designers were struggling with and then the artists went in and painted it up or polished it up afterwards but the key thing the key learning for us here was that like we kind of change their perspectives like people presume that like oh creating a more efficient system for an individual is the better way but with this we actually created more work like the artist did the levels the designers adjusted levels and the artist did the levels again but it also created a more effective process or effective product at the end so it

wasn't like about efficient individual efficiency but more process effectiveness so the failing the theory of failing quickly was was a big thing and and we wanted to incorporate into all our different like in studio processes so batching became one of these things that like we used to do it on Shang where we just like do all the run animation at one time do all the like still animation and then we implement it later we took that idea and we're like okay well let's let's test everything constantly is tested let's make sure everything works so we kind of eliminated batching from our full kind of process and we basically if someone animates like a kill we put it in game test if it works and so they're not just doing like all these self kills and and not having it in game so nothing kind of like lingers and then we're constantly testing our stuff and like consi is testing our assumptions so we're not wasting a lot of work as well and we

kind of get clear so the thing about batching is that it's this is this thing we're in a production process where if you know what you're going to do you can do a lot of it very quickly but if you're doing that based on bad assumptions then you're going to be very quickly doing the wrong thing so that's why we needed to get rid of batching until we really figure it out what we were actually going to open making and now going back to you know some of the the the theories that we're thinking about and and and and you know solidifying what the vision of the game is you know I see you know march of the ninja as like this finely sharpened blade because we just kept shaving things off and in terms of an efficient development process you know in many talks talk about how you need to be able to cut features you need to be as a cut cutting cut so how do you know what to cut is through a solid theory that you've actually tested your assumptions on and because of that

we were able to just keep fine-tuning ninja and it's almost to the point where if we cut anything else that the blade would break and what's interesting is you know as an aside you know our other game don't starve it's about you know where as marking the ninjas are finely tuned blade don't starve it's more like a bonsai tree it kind of grows and grows it's a you know there's all these interlocking systems and as the systems grow we kind of fine-tune those systems but it's growing nonetheless and it gets bigger and bigger and so it's kind of a different process but at the same time using these theories so the theories are flexible whereas you know if you have a formula that's inflexible formula is you put this stuff in and it comes out the other way a theory is when you understand what's going on better and so we were able to use these theories on don't starve as well and I think that even though it's a very different game in a different way of building it it still really helped us out now one other

thing you know that I used to think about process is how the people work together and how all of us work together and at clay we have what we call tea people which are people who are really good at one thing but are you know pretty good at a whole bunch of other things you know I'd say that in a traditional process there's a lot of AI people so people who only know their thing and don't know much about other people right and that's when middle managers come in that's when you need the middle managers because they need to talk to everybody else so that everybody knows what everybody's doing but when you have tea people and they work together they naturally already touch each other and they actually can talk about what they're doing and they actually think about the effect of what they're working on against other people and so the problems actually get worked out way sooner and you actually get more innovative results that way as well so you know iclay we've tried having middle

managers before we thought we needed them and honestly we we don't and it's worked out way better this way so this is this is something that I think about a lot and it kind of comes back you know to when I'm thinking about all these processes it comes back to a state of mind I think you know so so what I'm saying here is you know while shipping a mediocre games is going to screw up your company really quickly you know having an unsustainable process is going to suffocate your company or your your your yourself rather I would say and so you know the message that I want to bring across is that that's not okay and that I hope that you know today I'm presenting a little bit about how we do things and I want to learn from everybody else so that we can create processes that allow us to create art and I think that it it's it's not a you know a juxtaposition or an oxymoron when you have good theories and you have good processes you're actually more free to think about

new things because you're not just flailing around not knowing where you're going and hopefully hitting somewhere good that's kind of a message and I think that's it so time to just write there's a time for questions five minutes questions oh I forgot to say all those things I was supposed to say which is terrific Terry is if you want to go ahead and setup while they're taking questions that would help so questions so you have to come up to the mics over here on both sides if you have any questions no questions there's some questions so you talked about using theories can you give us some examples of the theories that

guided mark of the ninja yeah so I mean the theories were I mean for example that the process theories with some of them were you know tea people make it so that you don't need middle management but from a from a game from a mechanical perspective it was that planning made people feel like a ninja for example they'll be we basically keep drilling down into little theories and testing those things out so with the testing of these theories did you do it internally within the team or did you look outside of the team where you're going back to the same people again and again or was it fresh people every time so Nelson a great job and getting people off Craigslist to come over thankfully we weren't murdered or anything like that but they came in you know twice a week throughout development and sometimes would bringing the same people over and over and we made really clear I mean the distinction is that we use that data to

test our assumptions to say did our assumptions work or not rather than ask them what they wanted because what they actually wanted was to hit harder and to you know to slice people up that's what they said that you wanted right and so the the the feedback for us was oh they don't want itself why don't they want us though hey Jamie don't talk right talk actually did you did you use your hypothesis method to improve other processes besides game design ie absolutely so the idea behind for example the level the level building was that we thought that the best way to build levels was for the designers to build it and then the artist started up in that order right right I guess maybe more on the management side or more business side yeah there was I can't think honey up it's helping my head but for sure that's what I think about all the time maybe after I don't know sorry okay I

think of it right now hey seemed like it was really cool to see those like early prototype LD like testing hypothesis videos of game played it seemed like I was really polished like I wondered did you like did you do an extra effort into doing like a polished prototype of your assumptions or did you just like why wasn't it just blocks jumping around they were just blocks jumping around actually because we built a tiling system for the for the platforms that looked good as soon as you put it down so that was actually my art and I'm not an artist so I just threw down the blocks and it looks good because the artists are awesome but there were just blocks really okay so you did talk about the base of the method of applying

theories and making sure that you're applying them appropriately with your wrong Forest argument so I was curious as to with all these different theories being experimented and you're rejecting the ones that don't work did you still feel an attachment to that kind of mechanic being a fun mechanic that you could apply to a later project I mean do you guys have a notebook or an archive of theories didn't work for mark of the ninja but could possibly make a return in another game we don't actively archive them but I mean it's still cool mechanics like for example we actually had a mechanic where the ninja could counter-attacks we would actually go into attack pose and do like Assassin's Creed stolid you know movements and jumping over and stuff like that it was cool it just made people want to people more well right in Beijing in correlation with my question say for example that was a cool mechanic that you found cool would you ever see yourself applying it in a game that would be more appropriate I agree yeah we but we don't actively archive it it's

not like a process that we actually ready do I guess awesome panel thank you hi regarding the tea people arrangements if you have say an artist designer designer programmer and a programmer producer then for the artisan producer to communicate do they use the designer programmer as a as a middle manager or do they try them interact directly or how does that work we don't have any producers or any far as that kind of if it wasn't producer than say a sound guy then I think everybody like once he created an area in a team that can that has all these multiple skills or like is aware of other people's skills then everybody communicates with each other like everybody sits in the same area so if you have a question there's other people overhearing and then people could pitch in and stuff like that so we basically just create an environment where people to communicate as well as like everybody knows what's going on constantly and I also have you know in terms like more theories and guess is is

you know we see people in certain ways as well so we try to see people where you know for people it's backs to each other because that's the minimum amount of distraction but then they can still turn around and talk to each other you know from it from a space perspective the best space usage is when you have a tea and you're all working inside that tea but then you stare at each other at a time in one one person wait but it stands up everybody notices right so I used to work like that and I was terrible right so you know we actively try to think about all the little things that help the communication thank you thanks guys [Applause] hey we have yeah we have to