Introduction

The game was originally inspired by Ant Colony Optimisation (ACO), introduced by Marco Dorigo.
It is truly fascinating how a system could create such dynamic phenomena, and I saw another potential that dynamic phenomena could create dynamic decision-making for players. Sid Meier, the creator of Civilization, said that “A game is a series of interesting decisions”, which fits with algorithms with dynamic systems. The limitation of conventional simulation games is that, for some players, they feel too complicated to learn every game feature, even though the fun stems from that freedom of controlling a game as a system. That is why I started this project to create an easy-to-understand but dynamic and endless decision-making game.

Development Process

I started the project with MICROVERSE, where I gained the fundamental knowledge of low-level physics and optimisation and the way an emergent system works. However, from the playtesting, players did not find it fun to play because the execution of decisions players made did not quite influence the game's outcome. According to HCI theory, feedback is important to increase player motivation, but because the player in MICROVERSE had to play as one of thousands of automated agents, there was no space for players to make meaningful decisions. At the same time, I could not make the player's influence greater than other automated agents, as it was directly against the philosophy of emergence. Moreover, with a conventional particle life system, agents can only detect nearby agents because of the optimisation issue, which further limited the player's influence.

To resolve this game design issue, I had to apply a distinctive algorithm on top of the particle life I had built, which was an ant colony algorithm. Each agent leaves the so-called ‘pheromone’ on the grid-based field, and the other wandering agents can detect that pheromone and steer their direction based on that. That way, agents can now detect agents in a greater field, unlike conventional particle life.

Pheromone-Based Control

To control thousands of agents, I invented the distinctive system where the player can spray pheromone on the field, which enables other agents to follow it. This way, I could make thousands of agents really responsive, which could increase player motivation based on good feedback according to Self-Determination Theory (SDT) and HCI theory, while resolving the optimisation issue thanks to array-based grid calculation.

Behavioural Studies and Game Design Intention

The application of conventional behavioural psychologies in games alone cannot fully address why a game is fun, as a game is a medium of multiple layers. Moreover, considering all psychology studies before having a rigid system does not often work well because it could limit creativity, and most psychology implementation and analysis work mainly during and after development. However, I still believe applying and analysing behavioural psychology studies in games is good practice, as I personally believe behaviour could be controlled and measured.

Skinner's Operant Conditioning

Based on Skinner's operant conditioning, the player's behaviour is often reinforced when a reward is given (Positive Reinforcement) and punishment is taken away (Negative Reinforcement). The player will likely be motivated when they reproduce themselves by eating food (Positive Reinforcement), and when enemies disappear by starving to death (Negative Reinforcement). Even though there is no extra UI that pops up every time the player gets rewards, the reward and frustration system is directly embedded inside the system I have made. I think Tetris is well designed for this, as people play it again and again not because they like the fancy UI effects, but mainly because of the fundamental reward and punishment structure directly embedded inside the block-stacking process.

Dynamic Difficulty Adjustment (DDA) and Flow Channel

Dynamic Difficulty Adjustment (DDA) is systematically implemented in my game. If there are many blue agents in the field (the player is good at playing the game), there is more chance that red agents eat blues and reproduce themselves, but if there are only a few blue agents (which means the player is bad at playing the game or could not thrive because the red enemies are surrounding them), a fewer number of red enemies will appear because they could not eat blues in time and disappear. This made the tension consistent no matter how skilful the player is, and the game difficulty dynamically changes, which keeps them in flow, which happens when the player's skill and game challenges correspond.

Self-Determination Theory (SDT) – Competence, Autonomy, Relatedness

I considered following all three factors that can possibly motivate players with Self-Determination Theory (SDT) – Competence, Autonomy and Relatedness.

Competence – When players feel they are getting better every time they play the game, they are likely to play it again. For this, I intentionally designed the system to be as responsive as possible to the player's input, so that the player can feel that failure stems from the player's direct execution. Moreover, the player must set a tactic every time to thrive longer.

Autonomy – When players genuinely think they can control the system with no limitations, they will likely play again. With a drawing gesture, the player can guide blue agents dynamically however the player wants, which promises players freedom. And its control is not rigid like pressing a keyboard button with scripted behaviour, but requires a higher dimension of decision-making.

Relatedness – Relatedness means feeling socially connected, valued or part of a group. Sometimes Relatedness is ignored in non-multiplayer games, but I put the score panel so that it pops up every time the game is over, so that players can compare their scores with others, and this worked surprisingly well during the many playtests at showcases and festivals. They took photos of the best scores they had achieved and showed them off to other players.

Outcome Uncertainty, High Stakes, and the Near-Miss Effect

When there is only one blue agent left, the value of that agent increases severely, and miscontrol could lead to a game over. It is uncertain whether the player can reach the food or not, and when they fail to reach the food, the near-miss effect works and creates positive frustration (negative frustration prevents players from continuing to play the game, mainly because of a falsely designed system, like latency or a bug), where the player is likely to play the game again.

Feedback and Limitations

As the developer of HYDRA, I think the biggest design mistake is its contradictory reward system. The best tactic is to keep the number of blue agents as low as possible to create fewer enemies. If the player gets greedy and creates too many blue agents, the game difficulty skyrockets, and they soon die. In other words, the act of reproducing is rewarding, but the outcome is punishing. This could create frustration because of the mismatch between expectation and outcome, especially when players die after believing that having more blue agents is good. However, it ironically created Dynamic Difficulty Adjustment (DDA), which motivated players at the same time.

Laurence Oldham, former Design Lead at Frontier Developments, pointed out that I should add more “layers” to the game. Because the agents connect to each other, he expected players to make certain shapes or surround enemies strategically, instead of simply avoiding them. Moreover, I should consider designing an achievement system rather than relying only on the score panel. He also advised me to add a more progressive tutorial system before moving players into the main gameplay.

After receiving this feedback, I played Jurassic World Evolution 3 by Frontier Developments, and I could understand how the game applied those ideas: small consecutive missions with visible progression, followed by allowing players to build whatever they want. I personally entered a state of “flow”, playing it for 30 hours in a row with great excitement.

Ricky Haggett, founder of Hollow Ponds, advised me to improve the feedback system. For example, when the player eats the green food, it could make a “pop-pop” sound. I could also consider adjusting the pitch of the painting sound based on the cursor’s position on the screen. He also pointed out that I should explain more clearly what the enemies do.

After receiving his feedback, I played one of his games, Hohokum, which was a perfect example of excellent feedback and communication. I could immediately understand what to do without any textual explanation, and I felt deeply satisfied when I realised that I could interact with every object in the game because of its precisely designed visual and sound feedback.