Game Name: Oxygen Not Included

Creator: Klei Entertainment

Platform: PC

Target Audience: Players who like complex systems, base building, resource management, and hardcore survival games.

Play time: I played for 2 hours on Steam, experiencing 5 cycles.

Game Link: https://store.steampowered.com/app/457140/Oxygen_Not_Included/

Experience Overview

First, I know this game is hard. I bought it in April, opened it, saw a screen full of various properties, and felt like I was back in a chemistry lab (I was a chemistry major in my freshman year).

Actually, this is exactly what I came for: I love games with complex systems. But at the time, faced with this complex system, my first reaction was to look for tutorial videos—then I discovered these videos were often an hour long, and there was even a 40-hour tutorial collection. I watched a few short tutorial videos but felt they were complicated and hard to understand, which made me doubt if I could ever figure the game out. So, I gave up. My playtime was stuck at 8 minutes. However, I felt I would come back one day, so I didn’t refund it.

This was the right decision. This time, I finally had the chance to challenge this game again. I gave up on watching tutorial videos and didn’t look for any “beginner-friendly” text guides. I decided to just open the game and rely entirely on the game’s internal guidance.

Overall, this experience was quite good. I found that the problem with many tutorial videos is that they try to explain every element, every attribute, and every rule right at the beginning—in short, trying to make you find the “optimal solution” from the start. However, actual gameplay doesn’t have to be like that. When I focused only on the immediate tasks and the prompts given by the game itself, the game flowed smoothly. In the process, to achieve my goals, I naturally consulted the in-game tutorials and learned a lot. Although I knew I was making many mistakes, I overcame my “perfectionist mindset” and played until Cycle 5.

Game System

Oxygen Not Included attempts to create a sufficiently complex scientific system. The physics, chemistry, and biology it covers can serve as a “practice field” for STEM knowledge. Back in April, I found this complexity intimidating. But this time, after grasping some of the rules, I began to love this complexity—when I saw the huge technology tree, I felt excited because this game still had so much rich content.

Analyzing the “elements of the system,” the game’s complexity is reflected in its “properties” and “behaviors.” First, the “objects” include Duplicants, facilities, tiles, etc., which doesn’t seem too complicated. But what sets it apart from other games is that each “object” has a considerable number of “properties”: for example, a tile has physical properties like mass, temperature, specific heat capacity, thermal conductivity, melting point, and even biological properties like “surface germs.”

Furthermore, the “behaviors” a player can command a Duplicant to perform are also quite numerous. The game presents all the actions at the start: a total of 12, which made me feel overwhelmed. However, the game’s guidance system was quite restrained. At first, it only told me I could use Dig, and only gradually informed me of the usage of actions like Priority and Harvest. There are still some actions I’ve never used, such as Wrangle (capturing critters).

Figure: 12 types of behaviors

As for “relationships,” I believe that with such complex “properties” interacting, complex “relationships” will inevitably arise. For example, the “heat dissipation” system I imagined—facilities generate heat while working, causing changes in the surrounding temperature based on the specific heat capacity, thermal conductivity, etc., of nearby tiles. However, since I only played for 5 Cycles, I didn’t experience overly complex “relationships.”

Loops

I recorded my entire gameplay session and uploaded a part of it here. When reviewing the video, I found that the operation I performed most frequently was “Dig.” Through “interaction loops,” my “mental model” regarding “Dig” was constantly updated:

At first, through “instruction,” my “mental model” was: I can use the Dig command to mark locations that need digging, and have the Duplicants dig them.

Digging (Initial Understanding)

• [00:00] Decision: By observing the nearby area, I decide to dig the land to the left, right, and bottom-right to obtain plants and mineral resources.
• [01:18] Action: I start marking the tiles to be dug.
• [01:18] Feedback: Some tiles are highlighted in white, while others are yellowish-brown.
• [01:41] Feedback: The Duplicants start working, digging the tiles.
• [01:49] Instruction: A Duplicant is only 2 tiles high and can only climb 2 tiles high. (I didn’t understand this at the time.)
• [02:03] Feedback: After a Duplicant finishes digging one layer, the tiles below it turn from yellowish-brown to white, and the Duplicant continues digging.
• [02:03] Mental Model update: I learned that white highlight means “can be dug now,” and yellowish-brown means “cannot be dug yet.” Only after the outer layer is dug can the inner layer be dug.
• [03:11] Decision: The Duplicants have finished digging, and I want them to continue.
• [03:11] Action: I mark tiles within one block of the just-dug area (based on my new mental model).
• [03:11] Feedback: These tiles turn white, meaning they are diggable. This validates my mental model.
• [04:03] Feedback: The Duplicants start digging the new tiles.
• [04:04] Decision: I decide to continue digging inward.
• [04:04] Action: I mark the new “outermost layer” of tiles.
• [04:06] Feedback: This tile does not turn white, but remains yellowish-brown.
• [04:09] Mental Model update: I’m surprised; this doesn’t fit my mental model. I understand that the “outermost layer” isn’t always diggable.
• [04:09] Action & Feedback: I try repeatedly, but they all stay yellowish-brown. So, I temporarily give up on digging further and do other things (build a latrine).

Figure: White and yellowish-brown tiles (with shovel icon)

Building a Ladder

• [06:24] Decision: I discover the ladder-building function. Although I don’t know what it’s for, I try building one. (Skipping action, feedback… in short, it was built successfully).
• [07:15] Mental Model: I guess that by using ladders, Duplicants can reach new places.
• [07:15] Action: I use the Dig tool to mark an outermost tile that was previously unreachable but is now connected by the ladder.
• [07:17] Feedback: Success! After being marked, these tiles turn white.
• [07:17] Feedback: A Duplicant climbs the ladder and digs the corresponding tile.
• [07:17] Mental Model Update: Only after connecting with a ladder can some tiles be dug. This must be related to the “Duplicant’s height limit” mentioned earlier.

Figure: Building a ladder

Skill chains: Ladder + Dig

• [18:19] Decision: I try to dig an area in the bottom-right, but only one tile is white; the others are yellowish-brown.
• [18:19] Decision: Use a ladder to allow the Duplicants to reach the area below.
• [18:50] Action: Build a ladder.
• [19:22] Feedback: After the ladder is built, all the marked tiles turn white. A Great Combo!
• [19:22] Mental Model Update: By using a ladder, Duplicants can reach the corresponding location, and then they can dig the outermost tiles within 2 blocks of themselves.

Figure: With the help of ladder, yellowish-brown tiles turn white

In summary, my “mental model” was constantly updated and refined through these “loops.” From initially discovering “can only dig the outer layer,” to understanding “movement range limits,” then “using ladders to expand movement range,” and finally learning the “Skill chain” (Ladder + Dig). This learning process is based on trial and error. I frequently make “decisions” and “actions.” These operations don’t take much time and provide immediate “feedback” (like digging colors, Duplicant movement), helping me build a correct and in-depth “mental model.”

Arcs

In this game, a very obvious “arc” is the “night phase.” As shown in the video from 23:22-26:16, upon entering this phase, the player can hardly make any “actions” for the Duplicants, and can only watch them automatically complete the following events: eat, chat, use the latrine, and sleep. (They also disinfect after waking up.)

This phase is very important for the “mental model”: it clearly shows “what the player needs to do,” conveying the most important “game objective” to the player. This phase told me—Duplicants need to eat every Cycle, so starting from the second Cycle, I built a Microbe Musher. Duplicants need to use the latrine every Cycle—I was glad I built one earlier, otherwise, horrific consequences might have occurred. Duplicants need to sleep every Cycle—good thing I prepared cots for them in advance. And I also discovered the conditions needed for sleep: it can’t be too bright (25:43), and there can’t be a lack of oxygen, etc.

Figure: The night phase

In the first Cycle, I wasn’t clear on the game’s objective; I was just exploring the digging system. But from the second Cycle on, I clearly knew what I had to do: latrines, food, cots, and oxygen. These goals weren’t presented at the very beginning of the game, but were hidden in the “arc” phase after the first Cycle ended. They didn’t appear through “instruction” text, but used animations to show “what they need to do” and “the consequences if their needs aren’t met.” I think this perfectly fits the definition of an “arc”: efficiently conveying a large amount of information through “feedback,” updating the player’s “mental model.” Furthermore, when the night of the second and third Cycles arrived, my excitement and the amount of information I gained were less than the first time, which also fits the characteristic of an “arc” (diminishing returns on repeated presentation).

 

Game Architecture

This game’s “Game Architecture” is basically a “sandwich structure,” meaning “Loops” and “Arcs” appear alternately. The daytime of each Cycle is for “Loops,” and the nighttime is for “Arcs.” From the perspective of information acquisition, this structure provides information progressively, gradually unfolding the multiple facets of this complex system, rather than causing “information overload.”

Additionally, I want to analyze this design from the dimension of sensory experience: during the night phase (the “Arc”), the whole screen turns dark, the music becomes more psychedelic, and the player’s hands are freed from the keyboard (unable to operate)—the game characters’ rest also constitutes the player’s rest. This reminds me of the experience in Hades: after an intense battle in the lava of Asphodel, you occasionally enter a special room where you hear Eurydice’s song: “Farewell, To all the earthly remains…” This scene effectively soothes the previous tension and fatigue from dense operations. This shows that the alternation of “Loops” and “Arcs” also implies an alternation of “high-frequency operation” and “low-frequency operation,” building a better gaming experience for the player not only from an informational dimension but also from a sensory one.

Figure: Eurydice’s song in Hades

Value

At first, this game seems to present a “colonialist” value: humans use science and labor to transform nature and survive in a harsh natural environment. As a player, I reshape the terrain, harvest plants, and kill animals according to my personal will. But as the game progresses, I gradually realize that “colonialism” is just the shell. It also seems to contain a spirit of “environmentalism”: because this natural system is closed, if the player causes too much damage or pollution to the environment, it may ultimately backfire on them. For example, using a latrine causes pollution. If the pollution isn’t cleaned up in time, it will cause the Duplicants to get sick. I may need to play further to verify this.

Takeaway

Oxygen Not Included indeed has an intimidating exterior. Its complex system, on one hand, brings a rich gameplay experience (many people are addicted to it), but it also makes beginners shy away. But after actually experiencing it, I found that the game’s information presentation (using “loops” and “arcs”) is not hard to understand. With its restrained presentation, relying more on in-game “feedback” rather than text, and the “sandwich structure,” I experienced the process of updating my “mental model” (although there’s still a “long way to go”). This method of information system design will be present in my future games: for example, in the first round, not telling the player all the consequences of their actions beforehand, but letting them observe the consequences during the “settlement phase” after the first round to grasp the game’s objectives.

 

Mindmap:Working With System Dynamics