How Don’t Starve Was Made and Why The Devs Had Many Heated Discussion

The initial idea for Don’t Starve was formed in December 2010, when indie developer Klei Entertainment held their yearly company game jam to blow off steam for the upcoming Christmas break. Developers Kevin Forbes and Ju-Lian Kwan teamed up to create a small prototype that would later be used as the basis for Don’t Starve. The prototype only featured a health meter, that also served as a hunger meter, and a day & night cycle. At night, pig creatures would show up at your camp and after a couple of days they would overcrowd you, resulting in your health going down. The only way to stop them was to pick up a nearby axe and turn the pigs into meat. Once the player eats the meat, the health bar will go up again. It took Kevin and Ju-Lian two days to make the prototype.

For about a year, their demo was just a quirky result from a company game jam and nothing more. Back then, the studio was still working hard on Shank 2, which was halfway done at that point. Development began to wrap up in December 2011, but another one of their games called Mark of the Ninja was still in production. However, most of the programmers had already finished their part of the work so the indie studio decided to run a little experiment. See, Klei Entertainment was getting tired of pursuing publishers every time they finished a game. Being a small independent studio, getting a good deal from a publisher was absolutely necessary to be able to provide salaries for all the employees and make sure there’s enough money left in case finding the next publisher takes longer than expected.

On top of that, when they do find a publisher, a lot of the game’s earnings go to that publisher, creating a less than ideal situation when you’re a studio with limited resources. Therefore, the indie studio wanted to see if they could create a game very cheaply and very quickly so that they could self-publish it, retaining full control over the IP and most of the profits. Jamie Cheng, the CEO and founder of Klei, also wanted to create a better work environment for his employees, where constantly meeting publisher deadlines and working overtime would be a thing of the past. It’s safe to say that a lot was riding on this experiment for Klei as a company.

Around the same time, Google was looking to add more gaming content to their Chrome Web Store, a platform that seemed perfect to experiment on and get a game online fast and easy. So, Klei decided to try and create a new IP and pitch it to Google. Feeling fatigued from creating three side scrolling games in a row, the devs wanted to do something completely different. Their old game jam prototype seemed like the perfect opportunity to do so. “Project Not Sidescrolling” was aptly chosen as the new project’s work title.

The plan was to assign four developers, three programmers and one artist, build the game as fast as possible, get it to a minimally viable state and let people online play it. Afterwards, iterate the game based on people’s feedback. The only design constraint the team had to account for, was that the game had to be free-to-play. This was a hard sell for some of the developers involved, but they realized that the free-to-play model was becoming more and more popular and taking up a bigger share of the market. Klei felt if they didn’t try to get in on free-to-play games now, they might get left behind.

Artist Jeff Agala kicked off development by drawing concept art to try and set the mood for the project. These early drawings depicted a much more vibrant and happier tone than what Don’t Starve ended up becoming. Kevin admitted at a GDC talk that this was done to appeal to a broad, casual audience that’s into free-to-play games. The core of the idea consisted of two separate modes. One part of the concept saw the player become the mayor of a town inhabited by pig creatures. You could expand the town so that more creatures would come to inhabit it. If the player doesn’t take care of its town, the creatures would move out. The team had ideas such as being able to buy items like truffles with real money so you can feed them to the pigs to keep them happy and full.

While this part of the experience was very much inspired by popular elements of free-to-play games, the other part saw players exploring a wilderness area to collect resources for their town. Of course, going out into the wilderness would be limited to a couple of times per day, unless you’d pay to play more. Klei pitched the concept to Google who liked what they saw. Now it was time for the programmers to start coding the project for the Chrome platform and make it playable on a browser.

They created a custom C++ engine that was mostly written in Lua and after about a month, the team had little characters running around in a world that wasn’t just a blank screen. After the basic setup, it was time to start implementing gameplay mechanics. Since developing the town aspect of the project didn’t seem that fun to the team, the developers decided to first create the adventure part that takes place in the wilderness. The general hook of this mode was the emergent gameplay that would happen from discovering all sorts of little things that the wilderness area had to offer.

One of the initial test runs of this idea was a very simple gameplay loop where the player has to craft a trap, use it as bait for a rabbit, cook it and then eat it to prevent starvation. This process is meant to teach the player the basics, while still leaving enough room to discover things for themselves. For example, some players might try to eat the meat raw and discover that it drains their health so next time they might try to cook it and see what happens. After some iteration, playtests were positive and lots of testers actually really liked the concept. This gave the devs the confidence to experiment further.

People that have played Don’t Starve probably know that its world is procedurally generated, but this wasn’t always the case. In fact, the team had built an entire level editor to create custom made environments that were tailored to certain missions for the player. As the developers unfortunately discovered, these custom made missions turned out to be very boring for the player. Furthermore, lots of different goals and therefore lots of different levels would need to be designed to keep players interested and to keep them paying to play more than a couple of missions per day. The four developers assigned to the project simply didn’t have enough time to create all the necessary assets.

Everyone went back to the drawing board and started to take more inspiration from a lot of roguelikes the team was playing at the time. For instance, it gave them the idea to procedurally generate the entire wilderness area so that manually handcrafting levels wasn’t necessary anymore. Permadeath, a popular mechanic in roguelikes, was also added. When it came to missions and the actual gameplay, the team thought it would be best to let the player figure everything out themselves, meaning no tutorials whatsoever. They decided to focus less on what they thought people would want and more on what would fit the direction of the game to make it more cohesive.

The creators began leaning harder and harder into making it an uncompromising survival game that was intrinsically rewarding. “It’s going to be fun, because it’s fun. That’s why it’s fun”. The founder of Klei compares Don’t Starve to a game of chess or basketball. You don’t necessarily get anything out of doing something like that, but you get the experience of doing that thing. That’s the core of Don’t Starve. This new direction was a double edged sword and the team knew that some players would feel left out, but it gave Don’t Starve it’s own identity and that was more important to the developers.

“I’ve read of some players bouncing off the game because it doesn’t offer a standard progression model: people will die after hours and hours of play and lose everything they have accumulated in-game. They ask: “What was the point? I have nothing to show for my time!” Well, they had the experience of playing the game, and the knowledge that they gained from it. If that wasn’t fun or worthwhile while they were playing, no amount of digital trinketry will make it so. I think that a lot of the social cruft that’s been added to games in the past console generation is a distraction that detracts from the joy of playing. Don’t Starve was built to test this hypothesis.”

As the team started to embrace the more punishing aspects of the concept, the artwork by Jeff, that was mostly made and animated in Flash, also started to shift in a darker and creepier direction, while still trying to maintain a strong sense of appeal. The main character Wilson became tougher looking and went from having vibrant colors, to a more black & white color scheme with a sketchy newspaper cartoon style that’s inspired by works from artists such as Tim Burton and Edward Gorey. This new look complemented the tone and gameplay much better.

After about two months since development began, the project started to resemble the final product. Wilson was not yet finalized but there was enough there at this point to keep the player busy and experiment with the gameplay mechanics. The big difference compared to the final version is the text on the top of the screen. Now, the developers were adamant about not implementing any sort of tutorials but at this stage they still wanted to give people some direction of what to do so they wouldn’t feel too lost or aimless. Therefore, simple tasks were presented to the player that would escalate in difficulty and complexity. It was meant to teach players the basic mechanics before letting them explore all on their own. 

After a lot of playtesting, it became clear that steering the player in any sort of direction was more of a hindrance than anything else. Playtesters would execute the assigned objectives perfectly, but after about 45 minutes, all the tasks were done and the game would simply tell players to now go off on their own and explore the vast wilderness. However, at that point, playtesters would turn to the developers and ask what to do next. Without realizing it, the team had optimized the game for boredom. Yes, players were winning the game as it was presented but they weren’t encouraged to explore in any meaningful way beyond that.

After a lot of back-and-forth between the devs, the tasks were taken out of the game. Of course, all these rather drastic changes started to add up and now Don’t Starve was nothing like the concept Klei pitched to Google. For starters, half the game just wasn’t there since no work had been done for the pig town segment, outside of some concept art. Additionally, free to play monetization just wasn’t going to work with the current structure of the game. Having people pay for a bucket of meat so they can survive a little longer for example just flew in the face of the entire design philosophy. 

The studio had some soul searching to do. Kevin explained that he and the rest of the team had many heated discussions about how to monetize the project in its current state, without subverting the central meaning of the game. Jamie was also hesitant about making the game too difficult. To him, permadeath seemed like too harsh a mechanic to include. Developers Kevin and Jeff took matters into their own hands, made a demo with the new artstyle and mechanics in place and showed it to Jamie. It was enough to convince him to go forward with the revised trajectory of the project. Luckily for Klei, Google responded positively to all the changes as well and agreed to continue their partnership.

Their new concept involved removing the town aspect and instead focus solely on making the best possible survival game; and instead of having to pay to play more, monetization would now only rely on microtransactions from cosmetic items and new biomes that the studio would introduce after launch. Though, it didn’t take long for the team to step back from microtransactions entirely and just go for a traditional one time payment model.

Going forward, the team relied on four design philosophy pillars: dark humor, constant scarcity, player discovery and mysteries. The last pillar is meant to invoke player discussion. There’s so many little things to discover that it’s more than possible a player might only see 20% of the content during a single playthrough. For example, there’s an entire adventure mode that’s difficult to unlock and the game never even hints at it. As a result, it’s really cool when you accidentally discover these surprises and you’re more likely to tell your friends about it.

In the summer of 2012, a free beta version of Don’t Starve was released on the Chrome Web Store in order to gather feedback from players and improve it as much as possible before giving players access to an expanded, paid version of the game. “Opening up the development cycle early has been an awesome experience so far. It’s great for figuring out what aspects of the game players are really responding to, and for nipping usability issues in the bud. It has also forced us to figure out our distribution, purchasing and customer service issues up-front, which should make our eventual roll-out a lot easier.”

In the winter of 2012, a paid early access version of the game was released on Steam and the developers were very transparent with the community during this phase. They used the dedicated Don’t Starve forums to communicate with players directly. Furthermore, the creators released a public roadmap and were very open to discussion when people had suggestions on how to improve the game. Every two weeks, players could expect an updated version that addressed bugs, implemented improvements and added more content in general. 

A good example of the community giving the devs ideas, is the food spoilage mechanic. The team was already planning on adding this feature, but they couldn’t find a good way without making it very grindy. Coincidentally, a thread was started on the forum by one of the players that was all about how interesting it would be if food would spoil. The thread became really popular with lots of players adding ideas of how to do it in a good and interesting way. It grabbed the attention of the developers and they started participating on the forum. This all eventually led to the current food spoilage mechanic in the game. 

Something very interesting started to happen during the beta phase. In an attempt to grow the community as fast as possible, the studio offered an extra copy to everyone that bought the early access version so they could give it to a friend. Surprisingly, many people gave their extra copy to their favorite streamer or YouTuber in the hopes they would play it and make content about it. And lots of them did. This, combined with the sense of mystery and discovery in the game, boosted sales to say the least. Before the 1.0 version was even released, Don’t Starve sold about 300,000 copies. Moving forward, word of mouth was going to be their main marketing strategy.

On April 23rd, 2013, about a year and a half after development began, the 1.0 version of Don’t Starve was released on Steam. Critics praised the uncompromising survival game but it was clear that you would either love or hate the high level of difficulty. Luckily, more than enough people loved the experience and before the end of 2013, Don’t Starve sold one million copies. The growing popularity of Klei’s first survival game motivated the devs to keep expanding it for six months after release, adding things like special abilities for all the player characters and a whole underground cave system that can be freely explored.

The Steam release sadly marked the end of the Chrome version, the platform where it all began. All the new content and features increased the scope of Don’t Starve tremendously to a point where the Chrome platform wasn’t able to support the performance and development agility the team strived for. The platform also didn’t support mods so it made sense to retire the Chrome version. Players who bought the game for Chrome received free game keys that gave them access to the Steam version.

As the playerbase got bigger and bigger, more people started to ask on the forum about potentially adding a multiplayer mode. Surviving in the dangerous open world of Don’t Starve with a group of friends seemed like a perfect fit to the community. Despite this, the studio was very much against this idea from the beginning. Klei was very focused on making Don’t Starve the best single-player experience it could possibly be and didn’t want to sacrifice that goal in favor of slapping on a multiplayer mode as well. They also weren’t confident enough that it would actually work both in concept and implementation. 

However, once the completed version was out and became extremely successful, the team now had both plenty of time and resources to see if they could make it work. In the following months, Klei hired more people to work on future projects and more Don’t Starve related content. Jamie Cheng was especially excited about multiplayer and every time a new developer was brought on board, he would pitch it to them. One of them was really into the idea as well and started prototyping it. After a couple months and an office-wide session of multiplayer Don’t Starve, the whole team was on board to develop it further.

The standalone multiplayer version titled Don’t Starve Together was released on Steam’s Early Access platform in December 2014 and the 1.0 version was later released in April 2016. To entice and support the multiplayer aspect of the game, Klei decided to give every person that buys it an extra copy, just like they did with the beta version of the original game. The plan worked and Don’t Starve Together became a big hit alongside its single player companion.

Throughout the years, the studio released multiple expansions that brought new biomes, characters, creatures, items and other features to the already rich universe of Don’t Starve. Even today, the studio is still creating new content for Don’t Starve Together as evident by their 2021 roadmap on the Don’t Starve forums. The post also mentions plans for addressing lingering issues on the many console and mobile ports of the original game and its DLC.

The team learned a lot of lessons from developing and releasing their popular survival game, namely that the market is broader than you might think. At the start they thought they had to make a free-to-play game, because lots of people in the industry were saying that traditional games are dying. This is also why the team initially set out to make a bland, watered-down game that appeals to as many people as possible to make sure the game would be profitable. However, Don’t Starve shows that a studio can make something more polarizing that appeals to a niche group of people that will happily buy it.

Furthermore, the open and transparent development process was a very enjoyable experience for the devs. It was by far the best way to optimize for quality of life improvements and gave them the opportunity early on in development to find out what worked and didn’t work for the playerbase. All the while, the team was already making money off the game so they didn’t have to worry too much about sustaining development costs until Don’t Starve was completed.

Lastly, it also taught the team that overtime can be avoided. Klei Entertainment had problems in the past of working lots of overtime to meet publisher deadlines. Don’t Starve was their experiment to see if they could survive without the help of publishers and create a company culture where employees live balanced lives. It worked even better than the devs expected. No publisher meant Klei was able to set their own deadlines and development milestones. The beta phase also kept them on a good schedule and bugs were being fixed every day, which led to a very quiet and relaxed launch day. 

Nowadays, Klei has a reputation in the industry of being a studio that treats its developers very well and is still able to release games in a timely fashion. It goes without saying that Jamie Cheng is very proud of the company he’s built. ”We wanted it all. My goal was to build a company that could survive for many, many years. For 20 years, for longer; and the only way that any company’s going to be able to do that well is by retaining staff. What I want, and I don’t know how long it’s going to take, is that people look at Klei and say ‘well, they did it and they did it without really much knowledge and experience before they started. They own everything they’ve done. Maybe we can do that too.’”