The Job Simulator Postmortem

[Music] hey everybody uh welcome to the job simulator postmortem uh we're going to be talking about the Design Tech and business lessons learn from job simulator my name is Alex Schwarz and I'm the CEO and janitor of alchemy labs and I'm Devon Rymer CTO Chief technical owl at Alchemy Labs so uh first off uh if anyone doesn't know what is job simulator we're going to show an abbreviated version of the trailer uh for a couple seconds time to job this looks like it'll taste interesting pistons in the Eng and don't need those delicious you've been doing a really good job lately I think it's about time you got a promotion hooray it's 5:00 woohoo it's time to go home yeah I'm going to have to ask you to come in on

Saturday so shout out to Kirk Garner who did the trailer on this that we uh cut apart for this part but uh all right so we're going to talk about what went right what went wrong and the story of the conception of job simulator and what we learned through out uh so who has 14 hours to sit in this room and go through the whole story uh no it we're going to share some of the most relevant things that we learned uh and hopefully leave time for questions because this is such a uh expansive world of VR from all different topics so we'd love to get some questions at the end so let's roll back in time to a time before VR where we smacked rocks together apparently um so we uh at Alchemy Labs we started the company back in 2010 and we've been working on PC and desktop games for for quite a while um we shipped Jack Lumber and ah and snuggle truck and discourse with a small four-person team before the days of VR and then the oos Rift Kickstarter happened when we received the dk1 we

realized that VR was going to be a thing this time so we dropped what we were doing which at the time wasn't a thing that we just did to drop stuff um but they realized this was so important so we started work on a port of ah to bring it to the Oculus and we called it Oculus uh and it ended up being the first uh VR game on Steam that wasn't made by Valve so what ended up happening was we ended up doing this port and started thinking about what was next what we wanted to make some more VR content but we ended up running this problem is every one of our game ideas had this like seemed like it needed to be compromised to fit into the form factor of VR nothing ended up feeling up feeling right and we knew that we didn't want to make a game with motion sickness at all we wanted to make sure that VR was a thing for everyone so uh that's the moment that we got invited to do valves room demo at uh steam Dev Days 2014 and it was like a

religious Epiphany going into that room and putting on that headset and realizing what it felt like to have a submillimeter tracking that you could move about a room um it really changed everything for us and that was kind of the start where we said 100% of our efforts moving forward are to to try to realize this Vision in some kind of consumer product and build games for it but at this time you've got stickers all over a wall no one would ever do this and kind of walked out of that room feeling disappointed because you're like how would this ever happen in someone's living room it seems so far away and academic and that's when um we got invited to a meeting and Val and HTC told us they would be actually productizing a thing called the Vive and said would you like to drop everything you're doing and try to build some kind of showpiece for GDC by the way we can't get you this Hardware until a couple weeks before GDC so uh would you want to try to do something completely insane and build something for it and we said of course and uh since we were all remote as a small company we uh we

decided to go up to Winnipeg for a couple reasons to Devon's house one was that uh Devon's wife was 8 and 1 half months pregnant with twins and that I don't know if anyone knows if you fly at that level of pregnancy it's like you can't you can't go anywhere you're grounded you're stuck so we all went up to Winnipeg uh in the middle of the winter which was a fantastic move uh yeah no don't clap for that don't clap for that um and so we went and uh we got this shipment of this uh early Vive dev kit and we had one week together to try to build something for GDC so let's talk about how that went so when we said that we were going to go do this we started thinking about what would be interesting in VR when you like had hands and you could move around and the early idea was a warrior Weare like game where it's like a whole bunch of different mini games that means we could explore a lot of different spaces

um the first thing that we ended up doing because when we got the hardware there was literally nothing for it right so we just got this thing and we had to start building what we wanted to actually see and the earliest example we ended up doing was we ended up building three cubes and we put them on a table and we added physics to them so you could pick them up and throw them and instantly it was just this thing it was like wa this is super cool and we ended up feeling that kind of like I am a small child and I'm manipulating objects for the first time and I'm stacking them and I'm doing Lots of interesting things with them like literally that's how it felt because that's what I was doing sitting on the floor here with some cubes so we realized that hands were the key uh they were the simplest input and when you added physics things got incredibly interesting so um once we kind of got past the point of okay we're going to be using our hands to do some kind of mini games we continued forward with our thoughts that we should build a couple weird little mini gamess and so the first thing that uh we had a dock and we

were just listing bad ideas and so we thought okay a circus performer and so we started to build something where you could balance some kind of pole on your hand and throw a ball uh so we prototyped that we also prototyped uh a really quick version of a you're a window washer on the 52nd floor of a skyscraper uh and we thought we were super Geniuses because we put two um little levers which would move you left and right on the building and up and down uh and then you'd pick up a wind you know a little squeegee and you'd wash windows uh by the way that was terrifying um so we had some people buckle and just lay on the floor and start crying can't do it um and then we we thought okay maybe you could be a chef and so uh we started prototyping a quick um area of kind of like where a line Chef would be doing their work um we also U modeled out and started prototyping a bartender um simulation and we thought okay it would be really cool if we have kind of slippery physics

and we could uh end up pouring and then throwing uh you know these kind of glasses down the table uh rot beer Tapper type of thing and so we started to see like what some of the commonality was is that nearfield manipulation where you can physically walk over to something and pick it up and use your hands and grab within the range of where you can be was extremely awesome and so we needed to build stuff where things are all around you and then we kind of realized everything that we were doing you are kind of doing a job you're you're assuming the role of a profession so these are all kind of jobs uh and so I opened a Google doc and I titled it job simulator as a temp joke name and then we went forward with our bad ideas from there uh and the kitchen seemed to have the most promise as a current prototype so we thought let's just kind of not work on everything else and and push forward on making some kind of VR kitchen and see where it goes um and yeah now we just have to kind of build a GDC demo now with 4 days

left uh so what we ended up discovering very early on is that expectations can kill you um we had a huge problem with the kitchen everyone had preconceived notions of how a kitchen would work how you made food and now we need to realize this in VR we realized that we needed to satisfy all these expectations some way or the another and realize that if you can't deliver on something you either need to come up with something really creative or you just need to cut it um so two examples are the RO pin in the knife you want to talk about so first the rolling pin so this was an obvious thing to model a rolling pin in a kitchen makes a lot of sense you can pick up an object you can interact with it um but it had affordances and one of the affordances is that it could flatten things and so people would try to grab this thing with two hands and try to flatten objects and this was a physics nightmare so we ended up being realizing that the the rolling pin just couldn't exist uh and we realize that every single object needs a purpose it's not just enough that it's pickup every single object in job simul has some

underlying purpose y so uh the knife was an interesting one and uh let's go into this so basically uh everyone thinks there should be a knife in a kitchen and so we put the knife in and then everyone wants to cut something so we said okay we need to add a cutting system so we wrote Our Own uh Dynamic cutting system that would raycast through cutable objects build the interior meshes on the thing that got split create rigid bodies for all the pieces that got created and here's the first thing that everyone did is they grabbed the knives and then they chop chop chop chop chop chop chop chop chop and so uh anyone who's done any work in like video game physics would would start tearing up at the fact that that you've now got like 30 interpenetrating rigid bodies that are falling and shaking through the table and now your frame rate's starting to dip because you took a piece of celery and made it into 40 celeries and so uh it's this rigid body nightmare so we thought we uh are going to joke our way out of this one we're going to make awesome knives that when you cut they shatter into pieces

uh so haha we got you turns out that's extremely disappointing uh and so we ended up littering the whole area with lots of knives because that was a good idea so the goal was to have people cut less things and people are like can we just have more knives and it's was like oh it's kind of defeating the purpose here so it's just glass knives breaking everywhere it doesn't work uh and so then we thought later on hey uh let's task one of our Engineers with the concept of cartoonish rubber knives that bend away from items so they can't cut things that was one of our worst ideas ever um and so this was uh during development this is uh some problems um yeah and then this was the extremely creepy end result it's like a rubber Scythe um it was it kind of like made you feel sick to look at it it was

really weird in VR Everything feels different and it's just unacceptable so uh we ended up cutting the knife post GDC build but that's kind of the story of how one item that you think should be in a kitchen can have all of these different affordances that we can't deliver on so we knew we needed a character someone that you could interact with so early on we realized that with the amount of scope and time that we had that we needed to simplify things so the idea of floating robots came up because that was something that we could instant easily move around an environment and have you communicate with this um allowed us to not have to rig and skin complex characters it was very performance friendly because VR is very very import performance intensive and we realized we can actually um yield some empathy through the basic facial structure of a robot so we were in this process of developing this GDC build and we were a few days in and we realized that maybe we needed an intro maybe we needed to figure out where the person was standing understanding give them some story so

then when they ended up going to the kitchen they started to understand what this game was beware early screenshots this is job bot's office uh in this early build and you see there's this pedestal in the Middle with a bell the thought was this pedestal would come out of the floor when you were standing out of the way and then when you rang the bell you'd be positioned in the exact right location for the game and then early super early job bot here would come and tell you some stories about what this uh game was actually like and this also acted like a hub world we had VHS tapes on this table and you could go over and grab The Gourmet Chef one and then you could plug it directly into the face of jobbot jobbot had this VHS hole at one point um so we ended up having that in an early build that we sent over to valve and valve's like you do realize you have three minutes to demo this game right and your intro is like two uh so we realized oh we need to cut this it should have been an expectation and realization that we needed to shorten all the game play and get rid of the intro and not build it to begin with

but it ended up being an incredibly valuable one of the most valuable things that we ended up doing because it ended up leading us to the premise of the game so a game about jobs the original premise was that it was a training Simulator for people to learn how to job but why was that a thing that existed it ended up um through that job Bots office at the time we were doing a lot of thinking about what the future of jobs looked like and we talked about the singularity and it was a lot of things we were talking about like hey what if jobs went away what would the world look like and then we started rolling that into the actual game and the premise is that it's the year 2050 and robots have taken all human jobs and that the job simulator is actually a historical Recreation of what it was like to job there's no longer this simulator trying to teach you how to job correctly it was the robots trying to figure out what it was like to job so that future humans could experience the same thing but with poor Wikipedia skills that they kind of skimmed and said I guess that's how it worked um and so having that strong premise meant that we could have a

better uh process well we had a better design process throughout and writing process throughout so for example our one of our first tasks in the kitchen was you have to make soup and so uh with the Simplicity of what we were trying to do with picking up objects and throwing them we thought okay why don't we just make it so that you toss a bunch of items into a pot arbitrarily and out comes a fully formed can of soup because I guess the robots didn't know how real soup making process worked and they saw the two pieces and put it together in their head um and so that allowed us to be really creative um in how we were able to tackle design questions like we were able to use a microwave and uh kind of like label it as this Elite cooking device that the top chefs would use and so you could make a fully formed cake by putting uh flour bread and you know something into was it flour and eggs uh eggs but it was wasn't really flower it was like a flower uh like with petals so U yes bad puns and combinations led to this so um we got to the point where we're

starting to think okay we have something that's fun now how do we do this how do we do this and it turned out um even the most simple things were extremely challenging in the early days so the concept of how picking up items in the affordances should work visually what is the hand show when you pick something up what are the rules of all the physics interactions inside of VR and so in these days people were mostly just showing the controller visuals and uh not many people had shown hand visuals and the is um like grabbing inside of an item and it interpenetrates and showing your hand going through something it just didn't feel great um and a lot of people were also U building full arm ik on your first person view so you'd look down and you'd see your elbow in a completely wrong spot because there's no way to really predict that that data so it was kind of a mess and uh we thought that maybe we could just do a quick hack and try something where when you pick up an item uh your hand just goes away to to lessen the overlap effect and it

turned out we had no idea it turned out it was an extremely effective way and it had this kind of mental um process that we dubbed tomato presents kind of as a joke where you have hand presence with your hand and then when you pick up an item and you remove the hand visual the Tomato acts as like a visual standin for where your hand is and it's perfect rotation and perfect position so you still have presence it's just your hand presence turns into tomato presence and so since the Tomato was the first thing you encounter in the job that's why we named it that and now uh we've seen a multitude of GDC talks over the past couple years that refer to the official like hiding hand technique as tomato present so write it down everyone it's going in the textbooks tomato presents um and so room scale was kind of this big thing so fun fact when we got the first headset there was no chaperone no chaperon so we were walking into things and we had no ability to tell where the walls were and so we thought well how do we make an environment that doesn't that that makes it so that

people feel like they are in a space where they can access everything and that the room um you know you could stay in a safe area but build something that affords the space that you have and you don't want chapon to be the thing that stops you you want the world to kind of exist in a way that I don't want to go forward because there's a table here or something like that so a lot of thinking about that was uh was happening right now so as I mentioned physics was a super super important thing and we realized that once you end up having some objects that you can pick up you want to pick up everything so there's no hiding that fact anymore if that thing looks like something that you can pick up you need to be able to pick it up which ended up taking a ton of time in job simulator but it is now just the way that we operate so this is I don't know this getting closer to the end of our week period and then we had our Lighthouse fail and we only had one Lighthouse so we talked to Val like we need another one ASAP and they weren't sure they could ship it through customs to Canada fast fast enough so we decided to fall

back to a razor Hydra uh so we ended up building this setup and um long story short it is not even remotely comparable so that ended up being a waste of time but luckily we ended up getting a uh a shipment of another Lighthouse just in time for us to finish up the GDC build y thanks Chad uh so this is uh oh it's you right so this is the final result of the uh the kitchen that we ended up shipping for GDC uh if you've played the actual kitchen you see some of it's the same some of it is radically different uh but this is everything that's uh basically interactable um some of our old sandwich making system thank you human may I have another that one definitely stuck around and so the same thing of like serving things on a plate so we ended up having that basic functionality of the kitchen that carried all the way through um to ship of the game y so uh I mean it looked very primitive uh and job simulator in the early days didn't show well at all in video form uh

but people went into this first experience and again this was the first time that really anyone had had true hand presence in VR and was able to move around with uh with real accuracy and so press came out of these GDC 2015 demos with their mind completely blown uh and a lot of people called out job simulator as this unexpected you know the kitchen thing that I played that I just stood there and I think it was um was time that said I never knew that the most compelling experience I could ever have in a virtual world was throwing a tomato at a robot right like and and so it was blowing everyone's minds and it was an incredible experience to kind of be there at the Genesis of this this new hardware um so it was it was amazing and now it was time to just got to build the full game right uh so at that time we had no idea whether this Hardware would really ship what the price would be whether this Market would be a thing and so it was one of the riskiest decisions we ever made made uh was to just push

forward with this high-end VR game design for platforms we didn't even know would exist or not and just said that's what we're going to do uh whether it works or we go out of business we're just going to push um and that was when our colleagues friends family company advisers said hey uh we're not really sure about this whole VR thing maybe maybe like you should mitigate some risk and like build like a mobile game you should build a mobile game that's probably a better idea um and then we said uh no uh we're going to do VR or we're going to go out of business and because we're that sold that this is the future so um in our early thought process we thought we're going to make uh 10 jobs and we're going to start prototyping them all simultaneously and we're going to just go nuts uh well not all of them but we're going to go with three and push forward and so we started on the prototypes of a chemistry lab and a magician where you're in a backyard doing magic tricks for a group of small robot children um and and a space station repair man

repair person uh which that was just hey that sounds cool to have low gravity let's try something there so then we run into a problem uh something was wrong everything that we ended up making wasn't as good as the kitchen was everyone loved the kitchen but there was something that was off with everything else we built it just wasn't nearly as fun the humor wasn't coming through the gameplay wasn't quite coming through and we couldn't quite figure out what why that was so we started thinking what makes a uh good um job simulator job so we ended up getting some advice and figuring this out that the reason was familiarity that you take your expectations of real life into these things so if you had some understanding of how something was supposed to work then we could play with that then we could subvert that if we wanted or we could use that to our advantage to allow you to play the game more easily the further we got away from something that you understood the more complicated and weird and less funny

things got so for example let's say you're in the kitchen and you're making soup and putting ingredients in and you get a can of soup everyone knows you're not supposed to get a can of soup and it's funny but if you're in this chemistry lab and you're mixing chemicals together and then it creates a crazier chemical concoction did you know what those ingredients were supposed to do in the first place not really so we ended up losing that humor so we realized that it was like that grounding that familiarity that was the the key thanks to Patrick for that heads up there so that brings us to the office this the office is what we ended up showing so many demos of and ended up being kind of that demo show piece of the game uh most people don't know this that was the eighth job we prototyped it wasn't even on our original list uh when we ended up laying out those first 10 jobs and right now that seems like completely obvious but at the time it wasn't because we were thinking about it wrong we were thinking about what's the most exciting job that you can do without thinking about what would actually be interesting so this is a uh a very early prototype of the office

scene so you can see that there's some uh elements here that uh definitely carried over into the final version of the game staplers y with physics so um we thought okay let's again let's move forward on this office so let's think of all the tropes we know about every office thing think about office space we have to open with donuts and coffee there's no other way so we thought well I guess now we have to implement the entire stack of eating and drinking because we got to make this we got to sell the joke that's now needed um and we had avoided all edible Tech in the kitchen because we kind of uh we like ah you're a chef you don't necessarily eat the food you're making dot dot dot we don't need it but now we really need to deliver on coffee and donuts so uh it turns out eating in VR uh as more complicated uh because it you

put things near your face a lot and making a simple trigger that once you get close to it you start eating uh becomes uh unexpectedly you unexpectedly eat things by accident so um we also needed to build multi-stage eating because you don't want to just completely bite the whole thing and have it be gone you want to get halfway into it so that meant our artists had to build bite states of every single object that was edible including the sound effect of juiciness of whether that item was this kind of and that or the color of the particle effect and all that so yes it was very challenging we ended up having one play tester early on that was taking things that are refrigerator and uh and she was a bit shorter so she was grabbing things and every time she grabbed things out she moved it near her face and she just was continuously eating everything she pulled out of the fridge and couldn't figure out what was actually going on so that we have to go through this whole process of building all these systems so um liquid in VR uh this makes eating look like a cakewalk uh so 3D fluid simul ations are generally just not done in video games um let alone VR

where you have all these performance problems uh and with direct input I've seen some fluid simulations where they like hit a and they like kind of pour something or do some sort of Animation but you could directly impact the whole world and how the fluid is actually being manipulated in VR so this is an early um Tech uh we had this uh mesh that we were modifying and trying to uh get to work um uh this was it looked interesting and then just drastically broke down every time that you did anything even remotely complicated I built this terrible Tech uh it was two bones rigged to a skin mesh and it does not work in most cases so once you end up building liquid all of a sudden there's these affordances that roll out of this thing um you expect fluid to boil you expect fluid to be conserved when you pour it from one vessel into another you expect when you end up combining uh different fluids that the colors mixed correctly that when you pour fluid into World elements worlds interact correctly then

we had to realize oh man every milliliter needs to store things like its active temperature because he poured something cold into something hot what the colors were of that what the actual ingredients list of every milliliter of fluid and now I want to pour them into other volumes and now I need to be able to actually have fluid and momentum in that um and this was all to just show a cup of coffee so this is the final result um of the uh the system here someone making a cup of coffee we actually had somebody uh remark that uh this part of job simulator made them really angry when they were demoing to people because they were like people go and they just make a cup of coffee and they drink it and it's cool and they're like you don't realize how complicated that would be to build uh so if you're interested in more there's like an entire talk uh to this um we end up doing an upload VR article on how 850 hours to make that cup of coffee uh so you you can check that out so uh it turned out every new system

that we would add to job simulator as we got through building a job or a task that required some new system it would have fundamental backwards facing changes on everything else we' ever built so uh the implications on the game could be good or bad so now that eating and drinking was in every food-like item in any job especially the kitchen which is filled with them needs to now be edible which means all the bite States and all the art and all that uh every liquid or liquid like or anything that produces liquid or you expect maybe liquid could come out of it through some means that needs to be tracked and drinkable and in the system and so you can't set expectations incorrectly you can't have one job where you could drink something and then you go somewhere else and it's no longer the case um and so our process for finding those missing expectations because we can't know exactly what every set of human beings in the world expects about the world was play testing play testing play testing so here's a first example we had a plant in the office and once we added coffee people wanted to water the plant with

coffee and so people would hold it there and they'd hit the coffee and then they'd wait and they'd look at it and they're like and so now now we have to add that right we have to live up to everyone's expectations no matter how absurd they are because that's the joy of this world um the photocopier was a perfect example uh everyone would put something in a photocopier and expect an item and then they get the item to come out and then they put their hand in and then they'd put their head in and they would get a brain and people would lose their minds uh every single time uh this happened it was like an eruption of laughter I remember the first time we were in our office and there were eight people standing around and everyone hushed because someone started to put their head in and everyone's like it's not going to work it's not going to work and it was just pure silence and then everyone started screaming when a brain flew out it was and we've seen that on the achievement hunter um like video that series that they've done everyone s out because no one thinks that the developer is going to be that insane to

go that far um another good story was the earliest version of the kitchen we had those plates but um and it was Beth was playing and so she took a plate and she threw it on the floor and it did nothing and she goes this game sucks so it was and it was like okay plate breaking is now the top of the list so uh it's just little things like that is if your worldview Miss is a mismatch between real life that it just doesn't work for you so these every bot is like a generic thing we made this generic thing and we started adding costumes to them to make them more interesting and as soon as we did that people were like I need to be able to take the costumes off and then they're like now that I have a pair of glasses I need to wear the glasses so we actually had to go back through the entire game and rearchitecturing to that bot and then put it on to myself

so it's one of those things that as soon as you um allow people to interact with things everything needs to be interactable yep and people use that to tell stories they're like enjoy this cowboy hat idiot for no reason it's just fun so we knew in the convenience store that we wanted to have some sort of see today not your lucky day pal open that safe and give me all your chit up we wanted some sort of robbery but this is what everybody ended up doing that's plenty more where that came from [Music] kid so that was a really challenging scene to make because we knew that we wanted some sort of robbery because it's very like cliche in a convenience store but it's VR is super super emotionally intense and we didn't want anybody being freaked out so we ended up having to go through and we have this you know Bandit bot that's kind of funny and he has this like gangster weird accent and he's using words that are just like not like very very scary words and it has a banana and then you can grab the banana

and then keep doing that and he has so many vo lines to just like keep that funny Once you end up um going through the process of making everything interactable you realize that you now have to redesign the entire world it's like the giant the biggest like industrial design task a blender in real life is actually a really complicated thing generally they have lots of buttons and small inter interfaces which don't really work that well in VR so we ended up having to try to figure out how to solve the blender the blender was actually incredibly challenging to build cuz we didn't want this additional object that you had to snap into that with a lid that you then took off and then you poured and it needed to be pretty big to fit the objects didn't have a knife to cut the objects up into smaller pieces so we had this giant problem and so we ended up going through and looking at things and you see there's like this spck it on the side that then you could pour fluid out of it without having to pick the entire thing up and then having like small buttons and small dials didn't really lend itself very well to

VR so we ended up adding this like slot machine arm um onto the side of it and then ended up having some pretty great implications because as you pulled it it would go faster and faster and we could actually haptic in your hand as you were pulling it you felt like the thing was blending early on when we were going through this process and realizing that we needed to redesign literally every Appliance ever made uh I ended up going to um toy stores and going through because we discovered was that in VR right now at least you're closer to the interaction level of a 2-year-old or a three-year-old I'm just really basic interactions and so I would walk down these aisles and see how um some designers would design toys for children because there's some things that definitely have some overlap here so um as far as like how we were building at this state uh being able to find our problems and well come up with this uh a question and say is it going to work this way or this way we need to

be able to prototype as fast as humanly possible to find whether it should be uh style a or style b or style C that was uh the ninth blender iteration uh and every time we had a new play tester we'd look at how they approach the blender and whether they would know what to do or pause for a couple seconds and then we'd go back to the drawing board to try to make it better and so that meant that everyone who was building the game had had to have a VR headset at their desk so that they could test in editor and iterate as fast as humanly possible and that we have kind of built our company around the fact that almost everyone at the company wears a ton of hats and so if you are working on this blender you need to be able to do some light 3D modeling and some design and some programming and then do some testing and gather some feedback and then go back and iterate and so if you don't have to pass it off between people then you can shorten that iteration time and uh kind of keep things rolling as fast as humanly possible so uh also the humor U as far as how we've been writing The game everyone at the company becomes a writer uh and so people come up to us

and they're like oh who who was the writer for job simulator that was really funny and we're like no everyone contributed because every single thing that goes into the game every poster on the wall every uh like the keyboard at the computer it's all in this world of like what would robots think how how did the world work in through the lens of their uh like poorly done research and so everyone needs to be bought in on that premise and everyone kind of needs to be a writer in that sense when we ended up setting out and starting work on this we ended up building for what ended up being like a 15ish foot across size for GDC because that was what we knew the booth size was but we realized that that was not going to work in the real world um not everyone has a huge space and how are we going to end up handling it so this is the final version of the convenience store in the what we call 3 meter size so it's 3 m across so what we ended up doing was ended up redesigning each one of the spaces for every every Bun's different play Space size so if we go to

the next one if you have a size that is 2 1/2 M or less you end up switching into this size this is all transparent to the end user we take a look at what the actual volume bounds that they have available and we automatically do this so the user doesn't have to select any settings it just automatically happens and this is not something that can be procedurally done you can't just scale the world up and down and be like close enough because everyone understands what the size of an object actually is in real life so as soon as you scale it it'll look off so we actually lay out things differently based on the space based on the available space and change the way that things work so if you go to the and so this is if you end up having just that bare minimum size of 2 m um this is what you'll end up getting and you'll notice that there's some things missing like if you go back here there's a freezer in the front um that ends up going under the counters and we end up moving objects uh systems around and putting them in tool choosers and things like that but from an end user standpoint they don't even know and need to know what's going on and we've had people comment they're like I watched a

a streamer play the convenience store and it looked different than my convenience store did I have a different version of the game and it's like no that was the available space that you had versus the streamer so uh I mean menus were another thing that uh we had no concept of how we should approach and we had nothing we could really draw from uh and we really felt like it would be a copout to just make a 2d plane where you would interact with everything kind of in the method that we've been using for the past 30 years in in games and interaction so we thought okay how are we going to use inor interactions something that we already do as a verb in the game to do our menus um and so we're thinking okay we hit a bell in the kitchen maybe we could do something where you hit a bell and then you get the menu somehow uh or you throw a mug at something or maybe you smack some 3D button and so our first design was let's hit a button on the controller which there was a designated menu button we needed one additional button by the way there's one button in the game you just pick things up that's it and so we had to have one more thing so that you could kind of pull up a menu and we thought let's

spawn this this uh thing and it would have buttons on it in all directions and you could hold it like a big like Thor's hammer and you would like poke it and and something uh it was terrible and then uh finally the joke of what about a menu that's like a platter and then you eat the item in the menu and that's where the exit burrito was uh invented and so it was this perfect 3D representation of how do you uh solve for menus that don't exist in 2D um and it also has this this wonderful effect of having the two-stage confirmation because we could use a bite state to ask whether they were sure if they really wanted to exit otherwise you could throw the rest of the burrito on the ground and continue going um and so this ended up being like something that was referenced through a lot of people's designs and um again like we joked and said this is going in the textbooks everyone should exit their menus via a burrito um and it's just silly enough that it it seems I'm hoping it becomes a

term the same thing as like a radio button right it's like an exit burrito yeah when it comes to menus uh early we talked about that job Bots office that ended up getting cut we knew we needed some sort of location that you could end up branching off into individual jobs and then we ended up coming up the premise the idea of a museum tribute to um these past jobs uh was born but this ended up uh playing another role as a hidden tutorial so this is the progression so the jobbot there he floats over to you when you first start the game and passes you a business card so you're now learning how to play the game so you'll have to reach out and grab this card so we've teached pick up now we've teached that there's a small print on that that people will bring it to their face and read it we've told them some story and showed them how to actually interact with the world the next thing is options so we've got these cartridges that they can pick up we then teach a PCH points by having to attach it onto the uh the console and then

finally here we teach pullable levers and other things that interact from a fixed point and so this through this whole process that's really quick and intuitive we've actually taught most of the core systems Inside Job simulator before you actually even get into the point of playing and no one would treat this like a tutorial it's like no this is how you start the game but it actually is at a fundamental level a tutorial so there's one more Insidious little tutorial Point that's hidden in there is that when you get the card there's no table around you to put the card down on and so people are like and then they usually just throw it and we're trying to get people to start breaking the rules right away so maybe they'll throw it at jobbot maybe they'll throw it on the ground or maybe they'll do something something with it because there's no outlet for this thing you just got and now what do I do so hopefully they'll start breaking the rules and start going nuts performance and frame rate as I mentioned are a huge concern in VR uh from the very early days we realized that an art style was going to have to

be picked a chosen that would work really well with performance and that's why we picked the art style of job simulator is so that we could um build a game that was highly performant uh we we were lucky in the case that we ended up building a lot of mobile games uh prior to this and so we did that with unity and so we had learned all these different optimization techniques so if you're actually going through some VR stuff from a performance standpoint it actually is beneficial to look at some of the the mobile um stuff that was done in the past because some of those exact same optimizations actually carry over so if you're interested on hearing some more performance related tips because that's once again another talk uh we ended up giving a multiplatform VR talk at Vision Summit 2016 that's up online so uh at this point um we had gone through the process remember we originally thought there was going to be 10 jobs it turned out we launched day one on the Vive uh and were bundled with the Vive which was super super awesome and we ended up shipping with four jobs because we realized that people wanted

much more depth in the environments that they were in rather than an incredibly wide variety or bread of environments so that's just something that we found through play testing is we need they people wanted to go deeper and deeper and find more and more things in that world rather than just have lots of Worlds that scratch the surface um we also launched day one on Playstation VR uh huge challenge when it came to perf especially being CPU bound um and there was a a challenge of having to redesign play spaces like we talked about with the various sizes uh but we have now a very known camera frustum that we can calculate exactly where you can reach things and where you'd go out of bounds and so again hard mode for us was we couldn't scale our world or do anything crazy like that because it's it's a real life Place uh so hard mode was manually rebuild levels to be very specific to so that everyone has an amazing great time they never have to reach out of tracking or see this awesome shiny object and then they reach and they can't get it so so uh a lot of work there to design around that and

then uh yeah we did another whole talk on the whole the process of bringing uh job simulator to PlayStation VR which was at unite 2016 which is up online and you can check out um so psvr was also uh a big risk for us in the sense that we were building for this high-end of two track controllers and uh we didn't know at the time that we decided yeah let's let's go forward with with PlayStation whether move controllers would be packed in with all use uh and we knew that like in general uh controller that's not shipped in the box is a sub segment of a sub segment of a market that's early so um it turned out that launching on Playstation was an incredible uh incredible thing and we're very proud of that but in the beginning it was a it was a really tough decision to make is are they even going to have controllers um and so yeah it ended up that we were the number one selling VR game on the PlayStation Store in all of 2016 which was incredible um and for us it was also like vindicating in the sense that with job simulator and Batman

right at the top it proved that all of our prior Notions of people want hands people want to reach out and grab things and manipulate directly with their hands that that was true and that the numbers finally proved it so that was that was really cool to see um and that's what I just wrote on this slide so Oculus touch came uh next uh so we were also day one there and you can see here we implemented fingers so you can do finger guns and point at things uh it's pretty cool uh so this was our third launch of 2016 and this ended up representing the trifecta being complete that we ended up shipping day one on all three platforms that we had targeted so these three platforms uh we decided that we wanted to be everywhere that had hands tracked hands and a tracked head and we were very careful to not fall into the exclusivity trap and lock ourselves to one particular place cuz we think it's super super benefit I for the industry for as much content to be on as many platforms as

possible when it came to marketing it was very important for us to be there day one at launch but it was just as important to make sure that we were ready to build demos for conferences because all this Hardware was so early and we were all on the bleeding edge of what was working we ended up having to vote a lot of resources on a moment's notice to be build demos for different conferences because we realized that this marketing effort Not only was beneficial for us but it was beneficial for the platforms and pushing the entire medium forward this meant that we had super short lead times on things and needed to be very careful in our development that we didn't all of a sudden um run out of resources and weren't being able to ship the game uh we also uh ended up giving away the demo to all developers um on uh touch and on Vive and that was really scary because it was early versions and we didn't know what would happen and we're giving our game away to thousands of people but um it was really good for a lot of reasons um you know that we ended up making sure that people could

uh see the way that we were attempting to do things and we wanted to see the reverse we want to see how other people are trying to do things and sharing content uh was so important because when you get the hardware and there's literally nothing it's really hard to to know what if there's any best practices what are other people tried and you could save time and not trip over the same mistakes as everyone else so um when it came to actually showing the game in video form uh that's one of the biggest challenges of VR is that it's amazing on your face and then in the 2D view it's it's kind of hard to represent what's really going on and what's magical about it so it just looked like people inside of a weird job um throwing things and so uh we ended up making these four mini trailers before making the actual launch trailer and the mini trailers um which are uh basically they would show super over-the-top gameplay with uh throwing physics around and juggling and like uh trying to catch something in the air behind our back and and all these things and that was very specific because the over-the-top

physics Madness uh read very well as kind of a this frenetic weird game uh whereas that's not how people normally are playing it's just it's good on film um and then thanks to Kurt uh we got the uh the launch trailer with integrated mixed reality which uh was like a first of its kind no one had done this kind of stuff and so filming it with the Fantastic Contraption guys it was uh it was a really big moment and it was really awesome and then from a uh another like showing the game marketing standpoint getting uh let's let's players started to attach to this game in an incredible way and so we we built in spectator mode support so that you can get a third person camera and watch yourself externally because when you look around uh with just the unwarped eye shown on a screen it's really hard to kind of get the sense and people's heads move so quickly you it's hard to watch so building that in really helped people adding uh built-in twitch support was also really helpful for people in VR to understand what their twitch stream you know uh chat was what was going on

in the chat so all of that I think helped contribute to the ability for people to show our game which was our marketing is make it as easy as possible for people to show your game uh the launch reception for job simulator was fantastic uh people really loved it um the thing is we got lower ratings from traditional game reviewers in the very early days so the IGN's and game informers of the World um there was a lot of complaint about price in the early days we launched at $40 we ended up changing it to $30 after a couple weeks um we didn't know what Market expectations would be what would people pay for Content you know where's the ecosystem where are we in relative price to everyone else no one knew we didn't all get together and talk about our price we just had to come up with something and put it out there and so the thing is we were being compared to traditional PC games with 40-hour gameplay and uh you know 1020 million budgets or whatever they were like all budgets whereas um that that was the

only thing we could compare to is only a couple VR games and so how else can you base your reviews you just go off of a previous medium and that's really hard to live up to that kind of uh length of time and uh you know Quality Bar um the reception today however we were receiving actually higher reviews than we were at launch and there's little to no price complaints uh just the expected amount of you know internet complaints that get in a video game um and we're winning fantastic Awards or been nominated for them um which is which is awesome and I think being early to Market has incredible advantages but it's also really really hard because the developers are new to this the Press are new to it and they have nothing to compare to and consumers also don't really necessarily have expectations because who knows what's going to happen when they get VR so um we're all just kind of learning together how this all comes together our predictions in the early days um were that because this was a PC

based VR system with the Vive um and you needed a lot of expensive Hardware that this was going to skew towards a demographic that had a lot of expendable income that that PC Gamer which generally skewes more younger male was going to be the market and that market was very you know shooter driven um audience we chose to not Target that market specifically and instead build our game for everyone um but but why did we end up doing this we realized that the year one use case for VR was going to be that the actual Hardware owner the person that went out and bought the system wasn't going to be the player that was going to play our games the most amount of the time that once you end up buying this Hardware no one's going to take the hardware be like this is super awesome and then hide it in their closet from everyone right everybody wants people to come over and play and show their significant others and their families and their friends and hold parties and then spectate this and put it up on YouTube and all that kind of sharing stuff so we wanted to build um the game

so it worked for everyone because we think that VR is for everyone and we don't want this medium to fall into mistakes of previous mediums VR is truly for everyone and we have to work really hard to make sure that that actually becomes reality it VR is the most accessible platform ever made it's like in job Sim literally two buttons everyone's been interacting the world with their hands their entire lives so that they end up getting that it makes it super easy for them to interact so um you know the overall question of a postmortem is was job simulator a success and so from a financial metric standpoint uh yes uh getting uh pulling in over 3 million in sales in 2016 uh well exceeded our expectations I think uh getting great reception and being in the right spot and being super early uh meant that we were able to have a financial success the game still selling well um and so there's other metrics to success so um basically from an accessibility

standpoint we were able to deliver on something that everyone can play from the extremely uh young audience to people of all ages and that was I think one of our greatest successes that everyone's able to play this game and have a ton of fun alt together uh we I think we delivered on hands and physics because we focused so hard on that one particular realm of VR um and I think from from our standpoint that was a success we could use this tech for our next game and we were able to set a bar for hand interactions um from a getting there early and being prevalent uh the name job simulator is widely known and so we feel like that was a success and one of our goals is to have something that everyone had tried in VR uh from a humor standpoint we think that uh there's not many games out there or companies still that try to push really hard into humoring games and don't take themselves too seriously so we wanted to continue our Trend uh as Alchemy to do things that weren't super super serious so I think hopefully some of the jokes landed in job simulator so we think that that's a a reasonable success and then

from a sharability standpoint I think we made it easy uh for people to share the game and show it to others because that's one of the biggest challenges in VR uh and from a social standpoint I think that uh you know we we pushed in that direction and had some some success there and so very proud of that thank you and we would love to take questions so there's mics on the left and right hello um I loved what you said about the uh affordances and stuff you added with play testing about what percentage of your development time do you think was oh players want to do this we have to make that work 100% all right because it starts a lot it starts with the developers saying oh I want to do this and then we get through as much as we can and then we go okay well what else is there and then play testers play uh I would say the second half of development was much more

focused on reactionary uh fixes because you want this process to be smooth for every person who plays so your task list of what you wanted to do kind of has to be second priority to what is coming up as notes through play tests so it's yeah a trajectory of the end of the game development is all about what users are saying the early part is what we think might be good so definitely budget a lot of time for that oh oh yeah like this is the first platform where you like have like super infinite amount of input like it's not like oh I'm going over to a table and I'm hitting the a button I'm like yeah you could have the choice of an a button or a B button it's like you have the choice of doing anything with your hands at any moment in time yeah and we didn't even get into the QA aspect which would in our previous talks took another half hour but like we had people uh stick their face into a vending machine in the office and eat the items out of the vending machine like there's so many things you could do in 3D space that you wouldn't expect that as a developer it's like I didn't try that but someone did uh that there's so much you'll get from play testing

cool thanks hi congratulations on the game uh you mentioned the that sales are are still strong uh in the last I don't know maybe two months or something like that can you say how much it split between the platforms like how how strong are it selling in like each you know PlayStation okus and so there's so many ndas uh when it comes to sharing sales numbers it was actually really difficult for us to share any number uh so we were able to aggregate all of our sales into that 3 million number uh but we're not able to basically split it up uh or sub portion it okay goad thanks yeah so um this has to do with play testing and you know the challenges of getting your game out there to people because the VR Market's so small how did you guys like tackle um getting a broad amount of information to analyze and kind of adapt your game based upon that input yeah so um are you basically is your question how did you uh take

everything you learned from a play test and turn it into actionable like results well how did you like get your play testing out to a larger crowd to kind of get a more data to effect so we um the VR Hardware at the time that we were doing all this development it wasn't everywhere we couldn't just send builds to everyone and say run this on your PC and let us know how it goes we had to bring we had to have a constant flow of people coming into our office and then we needed to specifically spend time and schedule blocks of time to sit there and not have hire someone who was like the play test coordinator we needed the actual developers and designers sitting there and watching the players go through these play tests because uh yeah you could take notes on person did X but the the one who developed that interaction if they watch someone interact with it and do something wrong or hesitate or the affordance is aren't right they have to be the one who says oh I I think I get it they went here

they didn't look at this item and then when they saw the slot machine arm they didn't know whether they should twist it or turn it and then they could use that to go directly make those changes so it just needs to be a very tight Loop between developer and tester such that they are the same person uh and then just sit there and physically watch and take notes yeah and even with the hardware being readily available we still do a vast majority of our play testing in house just because watching people and seeing their actual reactions is super animportant we ended up having some external play tests where they ended up you know play testing and recording themselves play but we discovered that even then we had to then sit down and watch an hour or whatever the play test length a video anyway so why not just have them in the office and the feedback we get from people playing externally is more challenging and one of the reasons for this I think is that say you're play testing someone's game at least when I do that I have a notepad beside me or I have like you know some sort of browser and when I go and I'm like oh oh man that's kind of weird I go and write a note but when you're VR the process of writing a note is actually super painful cuz it's like I'm in the

middle of a great experience and then it's like he take the headset off put the controllers down go over my desk make a note and I find myself falling into the Trap of just like I'll remember that and you just almost never do because the next time a new thing comes up that erases the previous item so so there's a lot of sometimes someone yelling someone open Trello because they're in VR and they were like right down in the triage column that blah blah blah is broken uh and that's that's the only real like you could automate it or do something else voice commands there's probably some tools you could build for that but you just need to get it written down and our brains aren't good enough to remember all the problems that we play play test through thanks guys uh congratulations on your success um I think uh maybe for a lot of us here it's it's pretty impressive to see uh your early uh valve demo that you did in a very short period of time um can you talk a little bit about you know especially with completely new hardware and new sdks I mean now we have like steam VR we have the lab we have a ton of um stuff on the asset store that's

available uh at least in in unity early on when you started the valve uh demo and you said you had a very short period of time can you talk about any of the support that you might have received from valve and also even more importantly than that like can you talk a little bit about like what kind of things you think set you up for Success uh for that demo uh whether it's the team that you already had or how did you build your team like anything like that so yeah I mean uh basically just getting the hardware on time and uh having them put a lot of trust into US was the key parts of it um it really came down to the fact that this was we had done many years of mobile uh optimization and we had built games as a team together we'd been through that original Ip cycle four or five times and so we could sit down and uh throw out everything we'd ever learned about game design and say what works with this Hardware um and that was really the

design process there's not much you could do to prepare someone for this because it took two years to come together and like figure out what some of the Lessons Learned were and yet none of those existed at that time so it was just give us the hardware give us some time and uh see what comes out of it and that was their CH you know their challenges we don't know who might do cool stuff it might be this lone person in their basement comes up with the greatest idea in VR of all time like it's a really big challenge um and just them getting hard getting Dev Hardware out to as many people as possible is probably the best thing you could do as a platform to solve that from our standpoint the mobile and PC stuff we did before was super helpful but another thing that ended up leading to this was we never made the same type of game twice like it was like we every game we finished it and we're like What is radically different after this and this caused us to kind of always be in that mindset of just like throwing out your previous you know conceptions on things and kind of going at things a different direction which is like super super important in VR whereas if we had built

you know four Shooters in a row previously and that was all the design Focus we'd had previously it would be really hard to sit down and and say okay let's make a game about blocks right you just kind of want to go in a direction you've been comfortable with and our history has been removing comfort from our game design and just starting at new things all the time cool and what was your team size for the Val thing just out of curiosity uh four people okay cool thank you very much and we grew to we grew to 16 by the end of job simulator and then since Job simulator we're now at 25 great thanks I had a similar question to him so I'll just thank you for the game because it's time oh sorry we done uh they were're getting the time but uh what was your question we'll do like to say thank you cuz the first thing my four-year-old did when he played the game was pick up a bottle of wine guzzle It smash the bottle and then try and kill the job bot with it so thank you for that terrifying vision of my child's future uh it great thank

you all right um that is the uh that is the end uh make sure to fill out your evaluations and uh thank you very much