Kingdoms and Castles Postmortem

[Music] alright I'm petering step and welcome to the kingdoms and castles post mortem I'm co-founder at lion shield programmer designer mostly but I do a little bit of art and I've worked on a bunch of different types of games over the years the latest game I've worked on is kingdoms and castles it's a medieval city building game where you grow your Kingdom and defended against Vikings and Dragons and the cool thing about the game is that every single person is individually simulated so if the game says you have 5,000 people you can account for all 5,000 people doing something somewhere in the world it was made by myself and Michael petticord as the main developers and we had a few contractors Jason Taylor for music Steve green on sound effects and also Chad Freeman on sound effects I took about a year and a half of time to make a note here is that all the timelines in this talk are compressed as if we had worked a 35-hour week on it but in reality we average closer to 10 hours a week due to

other jobs and I'm also a stay-at-home dad it was built on unity with some handy plugins and it went really well we've sold about 650,000 copies so far and it has been one of the most fun development experiences I've been a part of and really it was all about using our time as efficiently as possible so let's dig into some of the interesting parts of its development we focused very heavily on market fit testing and for us that meant we needed we wanted to find out our people interested in this game idea at all how many people are interested and how big should we make the game and why are we so interested in market fit testing let's rewind 5 years to my first indie game cannon brah it's a 2d action strategy artillery RTS hybrid it's got a single-player campaign and a competitive multiplayer mode and it had about the same team size myself and Theresa

derringer and it took us about twice as long to make and they're both good games and in fact cannon brawl has a higher steam user score than Kingdoms of castles but kingdoms and castles sold a lot more and it turned out that Rahl had closer to a niche audience and ideally we would have discovered that early and scoped the game to match that audience size and this is where the market fit testing comes in and interestingly during cannon brawls development we did do things that would be considered market fit testing but we didn't take that data to scope the game so for example during cannon brawl I made a bunch of youtube devlog videos and then hundred views but that never grew we made it through Steam greenlight but I would have preferred more yeses and we did an early access launch which got into the steam top Tony which is pretty good but then sales declined in a pretty typical curve and each one of those green arrows is a Content update so we did all these kind of juicy

content updates but it did seem to change anything and we had approached this kind of with the mindset of I hope this update or a YouTube devlog or whatever is really gonna help the game catch on and it's true that each one of these well it is a marketing opportunity and can help you grow the audience for your game oh also informing you how big that market is and so really what the response that we were getting the market was saying well I'm only about this big so investing a lot of development time here is gonna be a risk so by the time we realize this it's it's time to launch the game and the launch ends up being a little smaller than early access and I think at that point we had basically captured the market interested in the game so overall cannon ball sold modestly but it wasn't enough for me to stay in because I was looking to start a family so I get it up getting another job and looking back on it now it would have been wiser to scope that game down a little bit given the response we were getting from the dev vlogs and green light and early access so back to games

and castles this time we're gonna market test early continuously and intentionally and we're gonna we're gonna use this data to find out if we should make the game at all and how big to make it should it be a quick three month project or five year magnum opus and we're gonna use a whole bunch of different things to figure that out so first true we make the game at all we used Twitter and our thought was if we can get something retweeted 25 times we're gonna count that as a success and we chose that number because at that point 25 times was an enormous amount for me so our first tweet got 11 which was act not a success but encouraging and we kept trying this was early so it was it was pretty fast and pretty quickly we got there with the gift of the tree growth system that got 25 and then we had another success after that with the cloud system so something in the look here is starting to resonate with people

and I've seen other people on Twitter kind of do this too like each of these is just a our tacit or little mock-up but it's all it's resonating so if these people were interested in making a game out of these they'd be worth taking to the next level and so the next the next step for us was to make a steam greenlight ideas page which is like a mock steam store page with screenshots and a trailer and people leave comments and the ideas get ranked but then nothing happens after that and so we came up with our success criteria which was get into the top 100 ideas and generate 50 comments by looking at other ideas in the top 100 and how many comments they got and also if they were successful when they eventually launched on Steam and we're using these success criteria as a way to be honest with ourselves and force us to course-correct early if the idea isn't resonating for example say we missed the top 100 and we get 20 comments half of which are neutral or negative we don't want to fall into this like oh well well ten of

them were good like let's keep going on that same track we instead want a bar to measure by that will tell us no go back to the drawing board and figure out why this isn't resonating so at this point the game is about three months in and it is not playable but we've built enough of it that we can get the idea across and screenshots on the trailer so that ends up going pretty well we get into the top 100 ideas and we generate a lot of comments and they're mostly very positive and so judging by the engagement on Twitter and how well this concept page when we can now answer the question should we make this game and we're gonna make it another benefit of doing this market fit testing was that you can course correct early one of the things we learned during the green light concept page was that we were getting a lot of comparisons of stone hearth and that's because we both have a pretty similar voxel art style going on so we redid all of the assets in a low poly style and because we learned this so early there

weren't that many assets to redo so it was a quick change to make so now we're definitely committing to making the game but we need to find out how big to make it is this a three month thing should we quit our jobs and hire people kind of thing and to find that out we're going to run a crowdfunding campaign and then do a real-deal steam greenlight run and at this point we're willing to risk six months six more months of development time to prepare given how well the concept page and Twitter did and so by the time we launched the crowdfunding campaign the game is is playable and so if we just barely squeaked by our funding goal week we can still ship something but the crowdfunding goes better than we thought it's not enough to quit our jobs and and staff up a lot but now we know how much time to invest and how big to make this game we are looking at making some big simulation changes and adding buildings in a military system and we now know that we have the time and budget to do that and and if crowdfunding had not gone as well we could have we could have

just cut that so ten months later it's time to launch and it ends up going really well the game breaks into the top 5 on Steam eventually peaks at number three we were unable to unseat pub G unfortunately but so this market fit testing process helped us avoid sinking lots of time into something that we weren't sure if enough people would buy and the launch was also our last market test which was do we continue expanding the game and for us the answer was yes and we're still updating it today so let's move on now to a concept I think helped it succeed in a market fit test which was that it has a very clear player fantasy and for this talk I'm going to give a special definition for player fantasy which is the ability of the player to imagine what the game is like before having played it at all then this is communicated through most often trailers screen - verbal description and your player fantasy I think can be anything but it just has to be very clear what it is so

for example Assassin's Creed the player fantasy would be your cool assassin in ancient times and it's like communicated in the title of the game the screenshots show a guy exploring an ancient city probably about to jump down and assassinate some people for rocket League the player fantasy is you're up a rocket-powered car playing soccer and that's very obvious from this very clear from the screenshot and even the title and papers please Europe be a border control agent in a dystopian country and you can see from the screenshot you're gonna approve or deny this guy is passport like it's clear that that you are kind of a customs agent here kind of person so going back to cannon brawl the player fantasy here is not super clear and I think the the best I could come up with was you fly an airship which can build attack buildings and defense structures and you use these to defeat your opponent manage your economy and it all takes place on a destructible battlefield but that's really just a description of the game's mechanics and

it doesn't spark player imagination very well before they've tried it and at the time we were coming up with cannon brawl it came from this very technical design place which was we're gonna take RTS game and blend it with an artillery game and that that results in a really fun and interesting game but if if I'm a player looking at this on the steam store it's not it's not very clear what I'm gonna get and I'm gonna have a hard time imagining how compelling this game is going to be especially if the next game in my queue is something that's very clear I probably might go with that one so if kingdoms and castles we wanted to make the player fantasy simple and very clear and so that is bill to manage a medieval city and the title is a little generic but it sums up the game very explicitly the trailer we made sure shows you building a town at different sizes over time and the screenshots show very clearly a medieval city and also specifically from camera angles and distances that you're familiar with in other city building games so when the

player frame is used is clear it's easier to imagine all the fun stories and good times you might have when you play this game and that makes the game's appeal easier to communicate in the game easier to share so at this point we've got a game that we think resonates and we can scope it appropriately and the player fantasy is properly communicated and now we need to make it as efficiently as possible so we lean very heavily on existing metaphors to save time going back cannonball again this game instead invents and repurposes a lot of existing metaphors and we had to spend a lot of time getting that to communicate properly to the player so for example in cannon brawl we have this magical shield tower and there's not a lot of real-world examples of a shield tower or there's not a lot of magical shield Tower in fiction except maybe in the context of sci-fi with a force field but our shield tower is directional comes out in an arc it can be moved but only before it's activated it needs to be reactivated but only after it cool down it can be upgraded shots breaks through relative to their strengths and the

shield tower strength there's a lot to it and we needed to make a tutorial level specific to teach this game and we also had to spend a lot of time and all the different UI pieces to communicate all of it and cannon ball is made up of a lot of these for example you have a headquarters where you dock your airship with to pull a building out of and people you don't expect a building to make new separate buildings that's not a real-world thing that happens the airship carries the buildings across the map and places them that's also not something you expect airships to be able to do so every time we were doing this we had to teach that and that that took time in the building's themselves like this little bomb thing here on the right it shoots out a robot bomb that when it lands walks to another building and explodes it's just a lot of stuff we had to teach because it was newish or we were repurposing something that already existed then that required a lot of effort so kingdoms and castles relies heavily on existing cultural knowledge like everyone knows what a farm in a

road and a house do and what to expect from those and so people were able to figure out how to play the game without a tutorial just by poking around and we were able to get away with shipping with that one which saved a lot of time and a lot of games use stuff like zombies or giant spiders which are kind of tired but that game is saving a lot of time by by using something it doesn't need to teach you how it works and the trade-off is that if you if you lean super heavily on existing design then you're losing a chance at novelty or innovation and counter brawl is for sure more innovative Xand kingdoms and castles so the question is like when should I innovate or invent design and for me now that's if it's absolutely core to the game experience so another game I worked on 7 billion humans with tomorrow corporation in this one we invented a programming language a multi agent runtime and in all the programming puzzles and this took a lot of time to invent and iterate and get this properly communicating and working but without it

the game would not exist at all so like that invented design was the game so for me like if it is absolutely a core pillar to the game or if if I have time and then I invent otherwise I lean heavily on existing design so another way we save time was with the art style and it's a very simple low poly style and I could go from zero to in game finished asset in a usually about two hours and that's because everything was made out of simple primitives so like if I have a building and I want to revise it I just duplicate and rotate primitives it takes it's super fast and if I want to make new buildings I can just take old ones and kit bash them together we we kind of cut some corners with the UV where we just take the UVs for a particular shape and then drag them on to a pallet texture and every building in the game uses the same pallet texture so I don't have to spend time properly on you be wrapping anything I actually don't know how to properly on anything

and if I want to change a color for something I can change it once in the texture and it changes for everything in the game so that the simple shapes and flat texturing gets us to the left and then unity stock shadows ssao plug-in color correction and some slightly customized fog carry us the rest of the way and the other benefit of this kind of simple abstract art style is you can get away with some goofy behavior like these farm farmer guys down there they look like little pegs and they are very clearly clipping through the wheat and there the farmer guys will actually clip through everything in the game but if they had heads and arms and legs and the wheat looked like actual real world wheat you would have more of an expectation that they would walk properly down the rows and bend over to pick it up and all that would have added time and gameplay code a design like the more realistic your assets look the more realistic the higher expectations players have for their behavior and and of course the

trade-off is that you end up with a game that looks very abstract so if you're making something like inside that gets a lot of value out of the really high fidelity animation and interaction you want to lean into that instead of what we did which was the opposite so beyond the concrete work on the game our studio culture was another big part of making the most of our time and the biggest thing for us is that we don't crunch specifically to get more done because we found that when we crunch we get sleep-deprived we burnout then fights more easily escalate we introduce more bugs we make worse decisions which leaves us less time to get work done and then we feel like we have to crunch more and that spirals downward so instead we want to stay well rested and that helps us resolve arguments constructively introduce less bugs make better decisions which gives us more time to get work done and more time to rest and the big way we found that crunch hurts is in evaluating decisions and so we can frame decisions in two ways one is a tactical decision which is like maybe our skill tree needs three more skills and a strategic decision which is maybe

we don't need skill trees at all maybe we should cut them and so we'll take a we'll do a little thought experiment in the context of skiing in the context of kingdoms and castles so let's in this thought experiment let's imagine we're crunching and we see another game and their UI looks really good this is north guard on the left there and our UI on the right and ours looks a little basic and we might think that well they're basically a peer of ours let's bring our UI up to par we'll just put it up an hour to extra a day to to really make this game great but there are some big issues with this decision that we're not able to properly evaluate and what we're gonna pay for that layer so when we're crunching we have the mental space to to properly zoom out and consider things from a high level we're only taking this next little tactical step first I'll redo this UI piece then I'll redo this piece of art and change this piece of code around and

when when you're crunching it's much easier to think this way because these decisions are easier to make than evaluating things from a very high level but the problem is it can turn out that 20 steps down this tactical decision tree there's some big pitfalls like maybe now disagreements with your team get very heated and very unproductive and now maybe we realize that all the you all the new UI is going to take a lot longer to do to execute it at this new high quality bar and and also because we're spending so much time doing this maybe we're putting off reworking core design problems or we're not even going to be able to see problems that we don't know exists yet because we just don't have any mental space to think about it and the final result can be you know there's three months left till ship and you've got really polished UI but big pieces of the game are not coming together right and we've got to make huge course corrections and end up throwing out work and so the like the only thing to do is is make these big changes and work ends

up being wasted and go back to where you started on the decision tree and do a different path and we think maybe well this just how game development is like how do we know this pit is here until we've fallen into it but if we go back to the same scenario and this time we're not crunching we do have a lot of mental space to think strategically we might come to a different conclusion which is we're not going to redo our UI yeah it it looks not great but it's functional it gets the everything it needs to get across and players are going to get more value if we solve like this other different problems you know are maybe we have a mid game decision low we need to fix and in this case we do have the mental energy to climb all the way up and foresee all the pitfalls that we just talked about and we can make better better big decisions more often and there won't be any more of these big come to Jesus moments where a big course correction will save or maybe not save the game we'll have the mental space to discover the problems and fix them before they become

a big issue and then the question is what we we just have so much to do and only so much budget in time we have to crunch to get get it all done but I think that itself is a tactical decision and the strategic decision in our case was to just cut early and all the time so this is a screenshot of our to-do list and most of these are big features that we did not ship the game with and just everyday we reprioritized this list to make sure that we were working on only the most absolutely necessary thing to ship the game in the time that we had and we could have crunched to try and get as much done as possible but then we would have run into all the the problems we just talked about so and I think the other thing with crunch is that there's this conscious and sometimes subconscious attitude that we have to suffer for art and that ends up meaning work ourselves to the bone and then the crunch also is sometimes romanticized there's like a battlefield bonding experience but this is kind of crazy because battlefields are a bad place to be so I'd like to reframe the suffer for

art to instead mean just overcome interesting art design and code problems because we can definitely do that without the downward spiral of crunch and there's a lot of really great and successful games made now without crunch so let's make all of them that way so we'll move on to how we marketed the game that our marketing was mostly not separate from making the game itself we didn't send tons of emails to press our YouTube we didn't go to any trade shows I'm we just didn't have time to do any of that because we both had other jobs and I taking care of my daughter so really it was mostly about going through the market fit testing process maximizing the game's appeal and and each time we did a new market fit test that was a way to show it to more and more eyeballs and and have a higher and higher chance of some kind of serendipitous event Ivo mentioned that

fig did help with some marketing especially around the crowdfunding campaign they were able to convert some of their newsletter subscribers to backers they ran some Facebook ads and they helped with some press releases around the campaign and lunch but by far the biggest thing for us was that the game became organically popular on YouTube so the way that happened was we gave out one free alpha key to a youtuber after the crowdfunding campaign ended and he made a video and then other youtubers saw it and wanted to try it too and it bubbled up and down the youtuber hierarchy in each time we did alpha update we've got another bump in videos and the key was that when viewers are watching the video they need a way to act on their interest and in our case we had all three of these they could sign up for our newsletter which listed on Steam or buy into the Alpha as part of like a late back crowdfunding backer option when we were working on cannon brawl we had totalbiscuit do a video early on but we didn't have a really good way to for people to follow or

track the game at that point and that was kind of a missed opportunity so one question is why was this how did this become popular organically popular on YouTube and one answer is luck there are some other factors that the game is very clear what it looked it is from a thumbnail like it looks like a city building game from a thumbnail so if you're browsing and you see your youtuber has made a video like this and you like city building games you're probably going to watch it when you do watch it there's a lot in there that you want to try for themselves like oh why are they building their tavern here I would have built it over here in my residential district and it's easy to make into a long series of videos because growing a kingdom and kind of a sandbox building environment can take as long or as short of time as you want and we also found that there is a specific indeed in D strategy simulation set of youtubers that that play only those games and this this our game fit really well in there into like what they were looking for which we did

not know ahead of time and really this slide is just taking from the slime Rancher G C 2017 talk about how to make a game appealing for you too so we tried what they said and found that it worked for us so last we'll talk about the games community and that was mostly discord which we started from day zero and now we have about a thousand act members on any given time and we mainly just link to it everywhere in our scheme description in steam announcements crowdfunding updates on Twitter and the main benefit is that you're able to drill down into issues in real time chat with people so like it helps you get past the feedback problem where people say oh this game's broken and then that's it instead if if they say that you can start asking questions right away like okay what specifically was broken like how does that work and in in this example this guy had a city of five thousand people and we had just introduced a graveyard mechanic and it

was causing this problem where he had to build graveyards constantly and it was taking over the city so like in chat we were able to figure that out and then he sent us his save game through discard and we were able to fix it that day same thing for bug and crash reports like a lot of people don't have formal QA training so trying to get repro steps out is really difficult but if someone says oh the game crashed you can ask them Oh like well what happened before it crashed like right then and there and then spin that off into a private chat and really dig in and like on this one the Reaper steps are something crazy like pick 10 Maps and on the 11th map like then the game will hang and like we're only able to figure this out by chatting with this guy for like 30 minutes and in a private private chat it's also worth thinking about when you do if you are going to do a discard the design of your channels so we made a hands-off Ideas channel because everyone has a lot of ideas about where the game

should go and it's nice to have an unbiased kind of view on where people think it should go are our main policy with the ideas channel is that we don't comment on any ideas and this is this is also because if say if someone wants to add the killer bunny from Monty Python they have a place where they can get that off their chest and feel heard but then we'd also don't have to have an argument about argument about that with them about whether that makes sense to add to the game or not we also used to have a random channel and all kinds of weird scary random stuff ended up in there and I was moderating that every day so I changes instead to something called cool fines and stuff and it's not perfect I still have to moderate it but a lot less and then we also get a lot more cool fines and stuff like some somebody posted this Lego scene they were making somebody else was making something in Maya that they're just posting for fun and also somebody is a big baked goods fan on our channel so

I'll close out the talk with a disclaimer which is none of this is any kind of recipe for success this is kind of how we robbed the bank in 2017 but now it's all changed for example steam greenlight doesn't exist anymore so we would have to do that part differently you can't count out luck in any aspect of our success just a quick example is that steam changed their visibility algorithm in October 2018 which hurt our daily sales we had launched well before that if we had launched after that would that have helped or hurt our visibility it's impossible to say and totally outside our control and then you can't count out privilege we're very fortunate to have the opportunities and support network over the years that helped us get to where we are so that's it for the talk thanks everyone for coming [Applause]

yeah okay I get we have time for two questions oh great uh at the beginning you were talking about how you were using Twitter to evaluate your very first ideas and visuals on the game how many sort of iterations did you have and how many were you planning to try before just sort of tossing the idea entirely and thinking of some other game um let's see I mean we we iterated I mean that that very first screenshot looked quite a bit different from the final result we did iterate a bunch because it was so early we were we were just kind of trying to get a feel of like if we do a voxel or low poly city building game does anyone care at all so like if say we had posted and been trying for maybe a month and it just wasn't growing then we would say okay time to try something else I understand the need to make money but when your

thing on Twitter wasn't good enough hits or whatever how happy were you at the final product like oh I really wanted that feature versus you know I obviously need to make money that put food on the table but how happy were you with that final release when you're like oh these five features been really cool I grew as people would have liked him more something like that uh what we released with I was happy with like it was basically the vision we had for the game not like if we had worked on it 10 more months which we didn't post launch it would have been everything we wanted but we had a really solid core in the in when we did launch that I think that I was I was happy with thank you all right Oh so you talked a lot about or not a lot but at the end about having to manage the community and talk with them how much did that kind of detract from your

production process because that's one of the issues I've been running into is just the more I have to interface with the community the less mental space I have to focus on production so I'm just kind of curious how you manage that on discord I tend to check in once a day and it's it's mostly settled down and it's kind of more of like like when we when we launch an update a beta version of our update then I'm very involved because I need that to to fix whatever issues are we have with the update but otherwise we just kind of let it sit and just chime in here and there the main like the main community issue that I've run into is how to deal with the steam forums because like that feels like a garden that you need to be weeding a lot and if you're not then when you come back it feels overwhelming so I like with the steam forums I tend to check in kind of once a week that way I have a mental space to to work on the game and then for that one day during the week I

go in there and pull weeds and ten flowers and stuff okay cool thank you okay you opened your talk talking about how you have you can simulate up to like 5,000 actors how are you able to measure the importance of what your audience are being the weight of importance of your audience places on that kind of simulation it's I mean there's there's no clear answer but when we i i'd say it's important because because each one of those people is simulated kind of everything in the game depends on that so like if i want to build a house I need five simulated builders to physically go out and get the resources one by one and bring them back so depend on how far those resources are away like where did you put your stockpiles where do you have your wood cutter guys that directly impacts the time it takes to build that house and so kind of the

whole game is built around that so I don't know if we have a good measure for how specifically important that is to people but without it kind of the core of the game goes away a little bit thank you so my question is also also about the optimization process so basically would it because you you you you son lay today each character independently would that be too slow when you have too many characters yeah and what kind of optimization techniques you use to tackle that first so for the villagers there there's the main optimization is that they're very simple like they're not animated meshes in any way they're just little static cube peg guys that hop around and then then for other things we've like we had a military system that also has thousands of little people and for that we used instance

rendering and distort giant arrays of positions of all the guys and then send that off in a single instance draw call to to make that work okay so the main cost computational cost is from rendering Ming well a big part of it is from rendering but a bigger part is from pathfinding for like all those four bits like a path finding that part what kind of technique you used so we have we have a deferred pathfinding system where each frame you send pathfinding requests to a like a six thread pool and then they burn through those as fast as they can okay okay we got a wrap-up thanks for coming [Applause]