Designing Pillars of State - Scale

Game design is hard. Each medium has its nuances, but games can let players experience elements out of order or without a rich narrative to help guide them. Certain games like colony-sims, city-builders or other sandbox games can leave players free to pursue anything it supports, however, the lack of guidance might not provide the motivation the player needs to continue playing.

Introduction

Pillars of State is a game I'm making which combines colony-sim with grand strategy and therefore risks not motivating players with a story or aim to keep them playing. Games like this need to rely on alternative strategies such as Apophenia or engineering experiences. Both of these strategies have been discussed in detail by Tarn Adams (Dwarf Fortress) and Tynan Sylvester (Rimworld & Designing Games) to make colony-sim games fun. Apophenia is the tendency to perceive a connection between unrelated or random things and is heavily utilised in Dwarf Fortress and Rimworld to let the player's brain do the heavy lifting of story generation and emergent behaviour. In simple terms, to achieve Apophenia you need to engineer experiences by generating a variety of events, making information accessible to the players so they can latch onto it, and having this cocktail be enticing enough for the players to drink from.

I find the best way to design a game is by studying games that inspire you and your idea. Identify what they do to achieve their aims and adjust as you see fit. Ultimately, there is no perfect implementation, our job is to find the least-worst design.

Reviewing

Dwarf Fortress is the inspiration for many famous games including Minecraft, Factorio, Project Zomboid, Rimworld, and many more. I have played a lot of Dwarf Fortress over the years, starting with ASCII on the classic edition, before moving into the Steam version. My best experience with the game happened because 3 dwarf children climbed a ladder into an apple tree I had assigned to be harvested, the ladder was moved to the next tree and the children were stuck in the tree. I wasn't aware any of this happened until I had a frightened dwarf run to my tavern to drown his sorrows in ale after seeing a dead child fall from a tree next to him. Before I could save the 2 remaining children, 2 more dwarfs came running into the tavern to also drown their sorrows.

None of this was planned or scripted, it was emergent behaviour caused by layering simple systems on top of each other until complexity arose. My brain, however, perceived this as 3 children trying to surprise an apple picker because they were bored. They had fallen asleep when the ladder was moved and thought someone would notice them. My dwarves were so busy with all the work I had assigned them that the parents had failed to keep an eye on the children and their deaths were caused by me driving my dwarves too hard. This was a powerful gaming experience for me, and the game didn't provide me with a story, only the events in which my brain filled in the blanks.

Rimworld is similar to Dwarf Fortress but puts the focus on the colonists and turns the social dial up to 11. Whereas a fortress in Dwarf Fortress typically starts with 8 agents and scales to 200, Rimworld typically starts with 3 and often only scales to 15-20, though it can go higher. Rimworld has this habit of making you care for colonists through their impact on the colony and ability to solve the problems you need them to. I once had a colonist who seemed to solve every problem I ever had, yet they watched as the person they loved got married to another colonist they hated. I was so attached to this colonist that when they got infected from a bullet wound during a raid, I sent out most of the other colonists on a mission that rewarded me with antibiotics. The group was not prepared for the mission and they were ambushed and killed before they got home, leaving my favourite colonist to die and 1 colonist remaining to manage the fort.

Whilst this wasn't planned, Rimworld uses different AI storytellers to adjust the game based on your decisions. Each time I solved a problem, the AI increased the chances of escalation to add to the experience. This escalation led me to become attached to the colonist that helped me the most, furthered by their emotional rejection and then the random chance of being shot during a raid. To me, this was a similar experience to the weed mission in Far Cry 3 or Ghost's death in CoD MW II. This rollercoaster of emotions was achieved through emergent behaviour, AI-engineered experiences, and Apophenia.

Finally, I have been playing a bunch of Songs of Syx, the epic city-state simulator. I have really enjoyed the take on colony-sim/city-builder that Songs of Syx has taken, turning the agent scale way up to 50,000+ and adding large-scale Total War-like battles. Whereas Rimworld focuses on the colonist, Songs of Syx focuses on the state. We lose out on the unique stories that occurred in Dwarf Fortress and Rimworld but gain a feeling of grandeur and world impact that those games struggle with.

Analysis

I want Pillars of State to combine the agent scale of Songs of Syx with the depth of Rimworld. This is an easy claim to make, but far harder to back up. With each design decision in a simulation game, we are not only adding development work but also removing compute power from our cap. All decisions must be worth the performance impact.

Because of my background in Ancient History and Archaeology, I have spent years studying the relationship between the individual and the state, and low-ranking individuals are rarely covered in history, which is a shame. There is no possibility of providing agents with a Rimworld-like depth and then generating 50,000 of them. This could be optimised by creating lots of simple agents and then adding fewer but more complex agents throughout the social hierarchy. There would be a disparity between agents and there is a good chance this could be jarring or hard to communicate with the player. On the other hand, this gives a personal narrative to the state's decisions and lets the player see how their choices impact agents throughout the social hierarchy.

Having both a state-level and agent-level simulation gives us 2 areas for engineering experiences and Apophenia. Rimworld and Dwarf Fortress have succeeded with Apophenia because they are so focussed on the agents, and we as players can best associate at a personal level. I believe this means state-level Apophenia is still possible, but there must be smaller tangents between events if it is strong enough to motivate the player. Engineering experiences are a bit easier with the state-level simulation as existing grand strategies like Crusader Kings III, Hearts of Iron IV and Europa Universalis IV all push the player to make compelling decisions that excite or frustrate them.

Conclusion

Pillars of State needs to combine both state-level play and agent-level play if people are going to enjoy it. By creating simple agents for simulating most state-level decisions, players can focus on more complex agents who act as an analogue for their actions. The state-level play requires compelling events which impact all aspects of society and integrate with the complex agents, reinforcing the Apophenia within the player.

Development should iterate features and simulations as follows: add state-level decisions (e.g. process grapes into wine); add a new task of processing grapes to simple agents (e.g. workshop or building to process grapes); add new events related to celebrations and alcohol (e.g. character was stabbed in drunken argument and lost an eye); add new state event related to festivals and trade (e.g. wine is traded with a nearby state); add new social impacts of state on agents (e.g. losing an eye is a punishment in the nearby state, they perceive the character as a criminal and refuse to buy his wine).

There will be more posts in the future about specifics, but if you have any comments or suggestions, I would love to hear them.