Table of Contents

Introduction

In games there are effectively 3 different levels of understanding for the player to master: The controls, the individual mechanics, the mechanical web. There are many sources and games that can be used as an example of how to do the first two levels well but generally games allow players to chart through mechanical webs with very little or no guidance. An insufficient understanding of the mechanical web can cause the players to leave early out of frustration despite potentially being skilled at many of the individual mechanics. For example the dark souls series is notorious for having extremely simple “tutorials” that pass across what the controls of the game are and what the actions do but fail to pass important information such has:

• Creating effective builds
• The methodical / back and forth approach to combat
• How to reducing risk during navigation

With players failing to understand what the game’s intention is regarding these higher level concepts it can cause them to drop the game and likely having the wrong perception of what the game was really about and not getting across the experience the game is intended to give. This effect is called Premature burnout as coined by Daniel Crook in this article.[2]

In another work by the same author they explore what might be some effective methods to identify the areas a game fails to pass its intention.[1] In this article I will explore this in the context of a game I made for the Game Design assignment, Romakni. The reason this article focuses on Romakni in particular is to have a more focused and practical approach where the theory researched can be applied.

What is Romakni?

Romakni is a 2D top-down shooter rogue-like game similar to games like Vampire Survivors and Journey to the Prairie King from Stardew Valley. In Romakni the player controls a nondescript adventurer who must endure waves of enemies that come from the sides of the screen in waves. The player can gather the loot that drops from enemies to buy upgrades for themselves and become more powerful for each subsequently more difficult wave that spawns.

An early build of the game can be found here

How to know something went wrong?

Before moving forward its important to be able to identify where and when a game is causing too much frustration to players. There are many methods to achieve this. The best and most reliable way is to directly get player feedback. Players will tell you what parts they found fun and what they found frustrating. They may not accurately determine what their source of frustration was but that is what this article will also help address. This method is very impractical at a large scale since it is qualitative data.[5]

Another method to find areas where a game might be frustrating is through statistics. Data about how often players die to a particular enemy, how long they spend in a puzzle, what quest in an RPG did players finish the least after having discovered it, etc. What statistics are useful changes game by game and how they are interpreted. For example mapping out a heat map of player deaths in a multiplayer map can show what areas might need to be redesigned or rebalanced to disperse players more evenly across the level.

The Baseline

For Romakni I had 10 people sit down and play it for 20 minutes, manually keeping track of certain statistics for each player session and also asking the players afterwards about their experience.

Although Romakni is not a particularly lengthy experience few players managed to complete runs on each character. On average the number of completed runs per player across all characters was 1.2. As the developer when playing the game about 3 in 4 runs succeed in any character, averaging around 6 or 7 completed runs depending on map generation. This is an unfair comparison as I have had more experience, knowledge and practice to master the game but the discrepancy is still worth noting. Many players knew how to move and shoot well enough at the start but consistently hit a barrier towards the middle levels and especially the boss level. The goal is to analyze why this was happening and what changes can be made to the game to better equip players to surpass that barrier.

Finding the barrier with Skill Atoms

A third method to find the shortfalls in a game where it may be causing too much frustration is to model the game and apply that model when observing players or analyzing statistics. To help identify where exactly Romakni is failing to pass along the right knowledge for players get the right experience this article will use the Skill Chain Diagram.[1] In this model the game is broken down by what actions the player can take, the feedback of those actions, and what the player can perceive from the feedback. These three steps in order form what the model calls a skill atom which when connected to other skill atoms forms a skill chain, eventually leading to an entire skill graph.

The Romakni Skill Graph

To help identify where in the process do players struggle, I have modelled a part of Romakni’s Skill Chain Diagram.

For the sake of clarity some the skill atoms after a player completes a run as they aren’t as relevant. One problem with this model is that conceptualizing more advanced skills and concepts feels like it exponentially creates more clutter in the diagram. Romakni is admittedly not all that complicated of a game as it was conceptualized and developed in 16 hours, so for a very mechanically complex game such as one might find in 4x titles or games with intricate systems such as The Legend of Zelda: Breath of the Wild, a diagram like this would be much harder to create and use.

After modeling Romakni’s gameplay experience into this diagram and combining the notes from the earlier baseline experience, players fail to master two particularly important skill atoms in their way of completing runs. These are the Kiting and the Upgrade cumulative cost skill atoms. This makes sense as the more difficult enemies get the better understanding players need to have of these more “meta” skills to improve their run’s consistency. In an article about premature burnout published in Lost Garden, it lists 4 reasons why players may struggle to master higher level skills.[2] These 4 reasons are:

• Personality
• Personal history
• Practical importance of imagined future rewards that stem from mastery.
• The ability for the mechanism to signal that there is additional depth of mastery possible.

There is little that can be done about the first 2 reasons listed, but as a designer the latter 2 reasons are within our control. Within the context of Romakni it seemed clear to players from the baseline that doing better, AKA progressing further in the map, meant that they would receive more gold and thus more upgrades. This means that the 3rd point is likely not the cause of what erected the difficulty barrier. That leaves the 4th point to potentially explain what is going wrong. In Romakni there is no direct way to tell how effectively a player is kiting enemies or how effective the build that they made is compared to another. A player is either successfully running from enemies or they have been hit and judging a theoretical alternative upgrade build is practically impossible without at least several hours of experience (more than what Romakni intends the player to put in). With these two key nodes identified, we can move on to how to address them.

Teaching Meta and Higher Level Skills

How do people learn?

Through developmental psychology it has been studied that people learn concepts more effectively when they are engaged in the process. This typically was studied through things such as classroom participation.[4] Video games are then in a prime position as they require direct player engagement to move forward and game worlds adapt to what actions the player takes. A better question to ask then is: When playing a video game, when is the player most engaged? In a GDC talk from 2006 researches talked about a study in which they measured the neuron response in a person’s brain while they played Super Monkey Ball.[3] It was found that players were most engaged when they fell off the course. This makes sense as that is a moment of high tension and also a moment where in which there is more to learn. The researchers state that this is the time a player is most excited has finds the most pleasure, which I disagree with, but I do agree that it is at least one of the moments a player is most focused on the game.

In Romakni however players were failing many times by getting hit and dying and yet it didn’t increase their mastery of kiting and creating effective builds. I believe this is because Romakni does not signal enough that a player’s failure was due to these advanced skills. Based on the findings from the Super Monkey Ball research, I have devised three approaches to better signal these advanced skills and get players to engage with them. Three different builds of Romakni were created and tested each with 10 different people. Below are the methods and their findings.

Build Planning

Before the player is thrust into a run they can find in the main menu there a button that brings up the build planner. They can also bring this menu up with the press of a button. In the build planner the player can construct a theoretical build and see its total cost. This gets the player to play around with and discover what are some efficient ways to spreading their gold. A limitation of this method is that it likely should have showed the players as well how it feels to play with the completed build but due to the nature of the initial code base and time constraints, this wasn’t done.

Results

The numbers show that players completed more runs than in the baseline however the real important piece of data is that during a run players were more likely to think about how well their build was performing. Between runs players spent less and less time on the build planner but their runs were progressing further each time. Once a player found a build that they thought was performing quite well the build planner was effectively abandoned however it had achieved its goal to propel the player forward. Most importantly players overall also reported a higher level of satisfaction than the baseline.

Free Respecing

While in the shop phase of the game the player could press a button to lose all their current upgrades and get back the gold they spent on them. The same button press brings up a menu similar to the build planner where the player can then decide to take different upgrades via this menu (not limited to one at a time like the shop is). This allows the player to quickly see some effects of buying different combination of upgrades and change their build however they like. A common problem with the baseline experience was that when a player didn’t know how effective an upgrade was and found it wasn’t as good as they hoped, they were now stuck with that choice until the run ends. Respecing allows much faster testing rates and allows the player to thinking about and iterate through builds much faster. A limitation to this design is that players a incentivized to think more about their present combination of upgrades and not as much the future combination.

Results

This method slowed down runs much more so than the build planner as many players spent considerable amount of time trying multiple upgrade combinations, especially when they reached around the mid game and had a decent amount of gold. This was the intended purpose of the respec so that the players could quickly iterate through some ideas and see if they thought it was an effective strategy. This method proved more successful in getting players to finish runs than the build planner and also boasted higher satisfaction.

Playing Tag With Wisps

Last but not least to get players to think more about kiting and generally herding enemies around a new level and enemy type was developed. In this level the player has to survive long enough as Will’O Wisps spawn from time to time. The player cannot destroy the wisps and must hold out and not be touched to successfully complete the level. If the player holds out long enough all wisps are destroyed and the player gains 5 gold. Failure however does not result in reseting the run but instead the level ends early, not awarding the 5 gold and not spawning the shopkeeper (meaning the player loses out on an upgrade). The goal of the level is to get the player to worry less about shooting and focus on mastering the movement and herding which is the real key to kiting enemies.

Results

This method was more volatile than the two previous ones as the wisp level had a random chance to occur. After players experienced the wisp level by the 2nd or 3rd time however that is when their runs started to improve. Compared to the previous two methods this was the slowest acting one which makes sense. The previous methods, especially the respecing, allowed for fast and repeated interaction with the high level skill it was tied to. Regardless there was still a noticeable improvement in both performance and in player satisfaction. In particular some players noted that they felt a sense of relief that the wisps didn’t kill them outright.

Alternative Solution

One other solution to reduce premature burnout by reducing frustration is a Dynamic Difficulty System. This was not implemented in Romakni due to time and development constraints but it is a common enough practice found in games that it should be mentioned. A DDS can help players keep progressing despite their failures so that they can still feel like they are succeeding “enough” and thus keeping them engaged. A few examples include the director in Left 4 Dead spawning more health packs if players are low health, changes to enemy aggression and spawning rate in Resident Evil 4 based on player performance, and most racing games having AI racers catch up to player artificially if the player is too far ahead (also known as rubberbanding).

In Romakni this could have been implemented by checking player run distance (how many levels they passed) and adapted the game through a few methods such as:

• Spawning more gold for the player
• Reducing the number of enemy spawns
• Lowering the tier of enemies in later levels

Overall Conclusion And Recommendations

Making players more aware of and teaching them more advanced skills can lead to overall better experiences for them. To get players to better understand and master high level skills environments that encourage rapid testing with immedaite feedback such as the respecing method proves very effective. Modeling a game to discover where premature burnout barriers lie is a useful way to identify them but may prove difficult with more complex systems than a game like Romakni has. Even without modeling out a game to find every barrier, signalling to a player about what known advanced skills exist or what effective use of them is like can still make a difference.

References

[1] D. CookBloggerJuly 19 and 2007, “The Chemistry Of Game Design,” Game Developer, Jul. 19, 2007. https://www.gamedeveloper.com/design/the-chemistry-of-game-design

[2] D. Cook, “What are game mechanics?,” Lostgarden, Oct. 24, 2006. https://lostgarden.com/2006/10/24/what-are-game-mechanics/

[3] I. Bogost, M. Consalvo, and J. McGonigal, “Game Studies,” 2006. Accessed: Apr. 07, 2025. [Online]. Available: https://www.avantgame.com/mcgonigal_GDC2006_gamestudiesdownload.pdf

[4] L. Deslauriers, L. S. McCarty, K. Miller, K. Callaghan, and G. Kestin, “Measuring Actual Learning versus Feeling of Learning in Response to Being Actively Engaged in the Classroom,” Proceedings of the National Academy of Sciences, vol. 116, no. 39, pp. 19251–19257, Sep. 2019, doi: https://doi.org/10.1073/pnas.1821936116.

[5] R. Koster, A theory of fun for game design. Sebastopol, Ca: O’reilly Media Inc, 2014.