Game Design Deep Dive: The digging mechanic in SteamWorld DigFebruary 24, 2020
Hi there! I’m Olle Hakansson, programmer and designer at Image & Form, a small Gothenburg-based developer in the tropical paradise called Sweden.
Image & Form is a studio with old roots, but with recent dabs into game development. Our most successful titles are Anthill (a mobile strategy game with some cool original mechanics) and the title I’d like to talk about today: SteamWorld Dig.
SteamWorld Dig is a platform mining adventure game where you play as a steam-driven robot called Rusty. The goal of the game is to dig deeper and gather resources, return to the surface and sell your loot, buying upgrades and then repeating the process. The idea was highly influenced by the Xbox Live Indie title Miner Dig Deep, which I enjoyed but felt we could improve upon. We released SteamWorld Dig in August 2013 for the Nintendo 3DS, then later to Steam, PS4, PS Vita, and most recently for the Wii U.
Strike the earth!
“From the very beginning, we had the idea of keeping the player in a constant risk of getting stuck as the main source of tension in the game.”
From the very beginning, we had the idea of keeping the player in a constant risk of getting stuck as the main source of tension in the game. If you would dig too deep without considering your route back to the surface, you would have no way to get out of the mine, and more importantly you would be unable to sell all those hard-earned minerals you had collected. The idea was that you need to consider the process of digging as a puzzle, where natural caves, falling rocks, impassable walls and other natural elements are the main obstacles in your path.
With this in mind, we started hammering out the basic setup of the game. We decided early on to go for a simple square grid of tiles that you can dig through, with tiles about as large as the player. This makes for tunnels that are comfortable to navigate around as the player, avoiding the problem of snagging on small tiles. (I’m looking at you, Terraria!) We could have gone for blocks that are larger than the player, but that would make for an ever blockier-looking game, so in the end we chose the size you can see today.
We experimented a lot with different jumping heights, as this is the main thing that determines what kind of digging patterns you can use when you are descending. For example, going from being able to jump one tile in height to two gives you a ton of extra mobility, as it allows you to build zig-zagging downward tunnels that you can then easily ascend when you need to return to the surface.
Being able to jump two tiles in height makes a huge difference for the prospective miner.
Another basic rule we introduced was that you may not swing your pickaxe (and thus dig through the ground) while in the air. Had we allowed it, and had you been able to jump two tiles upwards, you would have been able to dig zig-zagging upward tunnels which would basically have removed the entire danger of getting stuck.
Side Note: At least half of the reason for banning jumping and attacking was to reduce the animation complexity of Rusty. We would have had to make combined jump-and-attack animations for all different types of attacks if we had not forbidden it. I’d say the best design is where the game becomes better from the developers being lazy. 😀
It was a fairly sound plan. Our intent was to teach the player the importance of planning ahead and thinking of the tunnel digging as a spatial puzzle. I still believe we could have made a game out of that, albeit for a more niched audience. For anyone who has tried the final version of SteamWorld Dig, you will know that planning ahead isn’t really a requirement. You get plenty of abilities on your journey downwards that helps you from getting stuck. Most importantly, you can wall jump from the very beginning of the game. It’s certainly not impossible to get stuck in the game, but a skilled player can get out of almost all situations.
So, why did we stray from our original idea of requiring careful planning?
“Most playtesters were generally not having a good time with the game.”
Our first test session had only the bare essentials of the game: you could dig in all four directions, walk and jump two tiles. This is all you really need, and we hoped that the necessity of planning ahead would be obvious to our players. If not immediately, then at least after getting stuck once or twice.
It backfired. Most playtesters dug straight downwards, not really exploring the world. More importantly, they were unable to bring back any resources to sell, so they could not buy the upgrades they needed to meet the game’s increasing challenge level. As a result, they were generally not having a good time with the game.
So… how do I get up again?
We know of course that we had failed at communicating the rules of the game to the player. Our first approach was to extend the tutorial in order to better explain the required technique. While we did get a higher percent of players to understand how to dig “properly,” we were never close to reaching 100 percent. In addition, the tutorial became quite long, almost four times as long as the one we finally shipped. We couldn’t space out the tutorial since digging is the main thing you do in the game, so we had to rely on tedious repetition in order to really grind in the message we needed to give them. It was no big surprise that this wasn’t very appreciated by our testers. We needed another plan.
Secondly, we debated about allowing the player to dig while in mid-air. As I mentioned earlier, we had reasons for not allowing this, but it was a tempting alternative. This would solve the problem of getting stuck, as players could just dig themselves a zig-zagging path upwards if they would ever get stuck. Also, several people asked for this when playtesting, so we’d simply give people the thing they wanted, right? The problem is that would have been too effective a solution. We wanted the game to present the player with spontaneous and frequent spatial puzzles, and this feature would have given the player a single boring, tedious way of solving them all. So, this was not the way to go.
Our third attempt was to add run-jumping to the game. The idea was that if you hold in the run button and gain enough speed, you will not only jump far horizontally, but it also increased the vertical jumping height as well. You need about 3-4 tiles of continuous space to be able to build up enough speed to gain that extra tile’s height that you need to get up. While this works and feels really good, we had trouble communicating the feature to our playtesters. Some got it quickly, while some did not; we kept the feature in the final version but did not end the search here because it was not good enough on its own.
“Our playtesters loved it, and we decided to keep it.”
Finally, we arrived at the solution of adding wall-jumping to the game. We discussed a few different variations, for example one that required two opposing walls in order to work. In the end, simplicity won out and we chose the version we have today where you can get up on a single wall surface. Our playtesters loved it, and we decided to keep it.
Stuck — or perhaps not?
How is wall-jumping different from allowing mid-air attacks, you might ask? Both give you an awesome increase in mobility in a game like this. However, wall-jumping is more limited, in terms of getting yourself out of trouble. While it does allow you to get up quickly and easily from e.g. straight vertical shafts, you can still get into sticky situations if you are careless. If you fall into an open cave, you might still get stuck.
Note that we don’t want the player to get stuck, only to be wary of it. By introducing wall-jumping (and other movement mechanics), we made the state of being stuck from a binary condition into a gray zone, where players can use their wits and tools at their disposal to get out of most situations. We knew that we had hit the sweet spot when testers fell down a hole, sweated a bit, then found a clever way of getting back up again, feeling awesome.
In the end, we spent much more time than we anticipated on the digging mechanics, but I’m very pleased with the end results. My main grievance is that we had to add a lot of movement mechanics to fix our problem of communicating the game rules to the player. However, it was clearly a case of facing reality, as most players didn’t really want to play in the way we originally thought it should have been played. The end result was a game that was much more accessible and fun than what we would have ended up without compromise, and I’m very glad we made the right decision.
When we do get around to creating a sequel for Dig, I’m confident that we will be able to push even more fun out of this mechanic now that we know what we are up against.
Until next time, keep on digging!
Are you a developer interested in contributing to Gamasutra’s Game Design Deep Dive series? Email editor-in-chief Kris Graft: [email protected]