the elevator pitch
Auto Fighter is a unique auto-battle/fighting game hybrid. The builds, luck, and chaos of an auto battler meets fighting game reads, presentation, and action!
what i did as game designer
-
Developed 6+ core "Action Card" mechanics
-
Ideated 400+ Action Cards
-
Utilized XML to script 50+ Action Cards
-
Designed 3 CPU difficulty behaviors
dev specs
Platform | PC
Dev Time | 6 Months
Engine | Unity
Team Size | 7
fun fact:
I was responsible for editing the Auto Fighter EVO trailer shown on main stage!
my action card XML scripting
a snippet of my action card ideation work
design case
"the random checkmate"
key design problem
The game's dominant strategy in testing was to simply randomize your hand every round. This created a degenerate gameplay state that was further exacerbated by a lack of a way to meaningfully break minor leads/stalemates from creating guaranteed "checkmate" situations that players had no ability to meaningfully impact to begin with.
why was this REALLY bad?
In testing, the dominant effective strategy among testers was to simply be random. Players ultimately had no incentive to try to predict what the opponent's action order/hand was going to be since they could rearrange cards at will before the round started and you had no way to preview what your opponent's current hand is - only what they had the previous round. Not being able to view your opponent's current hand live or even "live actions" (networked non-turn based actions) was a technical limitation that we designers had to work around.
This lead to a situation where since little "anchoring" was occurring with your hand to disincentive randomizing your order. Players would ultimately reap the rewards of being unpredictable (in a game that heavily values it) with almost no downsides. The strategy would further degenerate into broader macro gameplay of buying cards with little synergy but with high stats. The idea was statistically neither player was more likely to abuse an element of predictability in the opponent's hand. Therefore, statistically - filling up your board with high damage centric options albeit a lack of synergy made more sense. A player just need to hit the opponent enough times to win, with little care or consideration for what the opponent may or may not do. This was further exacerbated by the "checkmate" issue.
The current card set at the time had little variance in terms of order. Once a card's order was set by a player, nothing would impact it. This meant that if one player got a health lead by the end of the loop, there was absolutely nothing to stop it - resulting in a "checkmate" where both players watched the same cycle of actions for 10-30 seconds. Combined with the little agency players had to begin with since being random was the most effective strategy, it created a degenerate gameplay state.
I was then tasked with addressing this key issue by my boss, Tim Fowers!
how i tackle a design problem
1) test + validate
A key part of my design process is testing and validating the core issues of the problem to find the root causes. In this case, I actually wanted to grab some more playtesting data. This involved playing a ton of VS CPU games and playing with peers to gather key data on.
Overall, the tests resulted in the validation of the problem. Being "random" was a dominant strategy across a range of skill levels and "checkmate" situations further negatively impacted the experience. A core piece of feedback was how the strategy felt like it nullified a lot of player agency and the lack of variance in the loop created a checkmate situation that fed into that strategy as well.
The data and player sentiments pointed to the root causes being the lack of "anchoring" in a player's hand order as well as a lack of a way to impact an opponent or your own hand order once a round started.
3) ideation + bounce
.jpg)
In this step I take the learnings from understanding your root causes and inspirations + references to formulate a plan of attack using my own design knowledge, tastes, and intuition.
In this case, this birthed the new keyword mechanics of "Shift" and "Position".
Shift was directly inspired by the literal shift effect in Darkest Dungeon, when your party or the enemy's party can be moved along their party's slots as part of an attack or ability's effect. In this case, your hand order mid-round. The objective was to create hand order variance in the middle of the round while accounting for the aforementioned technical limitations of "live" player actions. This was to largely address the "checkmate" issue.
As a designer, I'm very "flavor" focused - I design for player's logical heuristics. In this case, it was easy to justify the effect as "your fighter got dazed and jumbled up the order of actions they should take." Similar to how being dazed in a fight in real life impacts what actions a fighter uses.
As for Position, this was to address the issue of the lack of "anchoring". Position granted powerful bonus effects if played in specific positions or after specific card types. This encouraged players to risk predictability for reward that greatly outweighed the reward the random strategy granted.
Upon developing these key ideation points, I would bounce these ideas off of other designers on the team and Tim. We'd dive into the impacts, tech weight, and M.V.P (Minimum Viable Product) for such plans. If an idea from the ideation phase fails here, I return to step 2. The bouncing between step 2 and 3 is a common occurrence to really develop strong concepts from ideation.
2) reference + inspiration
Once the root causes are investigated + validated, I begin seeking reference. I firmly believe given the sheer amount of games in existence there's bound to be a game that touches on, experiences, or addressed my root causes.
Usually, this is me either remembering a specific game I played, actively looking online, or a combination of both.
In this case, I was reminded of the game Darkest Dungeon and how that game tackled position shifting. Both games were turned based with positioning playing a key role. Some parallels could be drawn from there I could gain insight from.
Gaining this newfound inspiration from the reference of Darkest Dungeon, I moved into my ideation step.
4) implement
A critical element of my process. Actually implementing the ideation to view its impacts on live play is absolutely critical to the process. Its here theory meets reality and where a lot of concepts make or break.
In the case of Shift and Position, given these were keyword effects I worked closely with our gameplay programmer on the team to tightly outline function.
I then adapted and developed a couple of new cards into the card set to utilize these two new keywords. I would script these into the game using our XML-based scripting pipeline to directly plug and play these new/adjusted cards with the mechanics into game.
5) test + iterate
The final step of my process is simple: test + iterate. Get the implementation in front of players and gather data, responses, and sentiment. Utilize A/B testing to gather comparative data, do analyze player surveys for keywords, etc.
Very rarely does an initial implementation ever solve the root causes, so ultimately I usually repeat this whole process multiple times until the data broadly signifies that the root cause has been successfully addressed.
In the case of Shift and Position, some adjustments had to be made on how those effects were distributed with "self shifting" effects being a key adjustment. It took a couple of iterations not only from a keyword perspective but also the cards that had them to really sand out the root issues found!
case closed!
The root causes of a lack of agency due to the strength of the random strategy and lack of anchoring were largely addressed via my work with Shift and Position.
The inclusion and development of Shift and Position also ultimately developed a stronger design foundation for cards and how they were classified. Higher level design concepts such as more tangible hand builds, hand counters, eco-rounds, etc. Emerged as part of my work and largely influenced the rest of Auto Fighter's design.
I was incredibly happy to have successfully tackled and been given a chance to tackle in the first place such a foundational design challenge!
design takeaways
less stick, more carrot
-
Reward players into performing specific behaviors vs outright disincentivising it. Realistically, the random strategy has a place - it just needs to not be dominant nor particularly effective.
little work, exponential returns
-
In a game like Auto Fighter, if a player engages with a system (in our case Position) that requires extra steps, reward them exponentially - not slightly. I had originally planned position effects to be minor effect buffs but it turns out players will go the extra mile if it really is that much more rewarding.
a little more for rich flavor
-
Players tended to memorize and enjoy cards that were extremely flavorful or "flashy" - even if the card was significantly more complex. For instance, we had a card called "Dragon Uppercut" - if positioned last, it basically was an invincible attack. Similar to an actual Dragon Uppercut in a fighting game. A huge downside was that if you lost the exchange, you were stunned for 1 whole action. In comparison to the simple jab card, it's a lot more complex. I personally expected only experienced players to appreciate or even use the card but during feedback many noted the card as being their favorite with a high usage rate overall.