HI frens, part two here for your review. (post 1/2)
After several days of immersion into the trait mappings I can say that it is a mammoth task to balance this game and requires the dedicated attention of a multidisciplinary team. Thank you everyone for your contribution on this topic, I apologise if any of my conclusions come off as dismissive of anyone’s concerns. I have tried to put an objective lens on my analysis and I am entirely open to critical discussion of everything below. Also Im not a quant guy so the statistical analysis is probably wrong.
*there are several reference to points already made by others in this forum. Just going to post this up now and if you read something that was your suggestion, I know and I’ll try getting around to crediting/answering you.
Part 2 - How to make sense of the Aavegotchi trait mappings
TLDR: The trait map is good and the curve is our fren. Aavegotchi’s unique curve of traits provides the base for exciting power ratios of abilities. From the perspective of this qualitative analysis there doesn’t appear to be critical imbalances in concept. Remapping abilities from one trait to another does not produce meaningful and holistic balance improvements. Balancing Aavegotchi’s unique statistics requires intensive quantitative analysis and might best be achieved by those working closely with the statistics in the current mapping. There are several assumptions that can be challenged in this conclusion and further research and analysis might invalidate it. There are several areas where attention may be required to ensure all gotchis and all player motivations are covered for in PVP/PVE gameplay.
Executive summary
Goal: Identify what problems might exist in the proposed trait mapping in its current state.
Outcome:
- Discovered that it is not a fair comparison to measure gotchi trait mapping against classic trinitys
- Tabled how Aavegotchi trait statistics create there own unique balancing dynamics
- Identified that quantifying the balance implications requires further statical analysis of Aavegotchi traits
- Qualitative analysis of two trait combinations determined there are no critical imbalances in principle
- No Aavegotchi trait combinations appear to be inherently over powered
- Minor number of Aavegotchi trait combinations might be underpowered
- There may not be a sufficient variety of play styles to serve all player motivations and behaviours
- Remapping of general traits does not appear to be warranted from findings at this stage of analysis
- Voting to remap traits at the current level of understanding might fail to produce a meaningful change in balance dynamics.
- Changing the traits at this stage may slow down and constrain developers, negatively impacting ability to build balanced trait statistics around the current set of features and deliver the gotchiverse on time.
- Effort might be better allocated toward researching and enaging with community to understand
- How inclusive are the trait mappings of all gotchis?
- How varied are the ability sets to the whole community’s prefered play styles?
Part 2 contents
2.1 Defining outcomes - what makes a trait map good
2.2 Critique - Modeling Aavegotchi trait map against the design principles
2.3 Approach - what to analyse
2.4 Synthesis of trait mappings - strongest pairs
2.5 Findings
2.6 Conclusion
Introduction
It is integral to do a rigorous analysis before jumping into solution-mode and changing anything. Part two of three in this series of papers is about backtesting the design principles proposed in part one. The design principles are derived from a model: and all models are wrong but some are useful. A lot of work from this analysis was left on the cutting room floor. Discovered that Aavegotchi trait mappings are extremely complex and it takes much time and attention just to discover a way to think about the trait dynamics. This report strives to begin an explanation in how they work together.
2.1 Defining outcomes - what makes a trait map good
The point of the first article was to define what it is that makes classic trait maps work. A mental model was made to describe the way classic MMORPG and MOBA trait maps enable players to experience fun and varied game play options within a satisfying balance.
The outcomes defined by these design principles can be used to measure how well balanced the Aavegotchi trait map might be.
The values in the below model are a result of variable inputs and are not to be taken for a ‘holy trinity of absolute values’. The values might be whatever suits the game in mind. What we’re interested is the relationship between value types.
Primary Attributes
Traits that enable confluent game play abilities are bound together within a Primary Attribute. Selecting to increase a Primary Attribute will increase containing traits in equal measure. Selecting for a Primary Attribute comes against the opportunity cost of other Primary Attributes.
Outcome
Balance - Primary Attributes are a core enabler of player desire for and satisfaction with trait balance.
Binary Traits
Traits that are adjacent to one another share a theme of abilities but control these abilities in different directions. A classic trinity can be defined by six binary traits. In the above example two traits live within the ability theme ‘Damage.’ ‘Inflict Damage’ increases damage abilities and is adjacent to ‘Replace damage’ which increases ability to regenerate from damage incurred. Options to optimise for binary traits promotes a variety of game play styles. As an example of damage, the classic high DPS life-steal - a high risk/reward play style that abandons all range and durability to specialise and survive on damage output alone.
Outcome
Variety - Binary traits pairs enable players to optimise for niche specialisations by playing creative hybrids of core trait abilities.
Opposing Context Opportunity Costs
Increasing one trait comes at the opportunity cost of increasing traits furtherest side away from, and in different Primary Attributes to the increased trait. For example, optimising for Mitigate Damage comes at an opportunity cost of popularly desirable opposing traits: ‘Inflict damage’ and ‘attack range.’ However, this is an attractive trade off to the player who likes to play the tank and enjoys to be the big unit on the map that the whole team is relying on.
Outcome
Fun - An interconnected relationship of opportunity costs between all traits allows players to experience satisfaction in their preferred play style, whether that be hyper-specialisation or all-rounded, within the confines of fair game balance. Constraints breed creativity.
When the internal integrity of a trait map is robust: there can be both game balance and player creativity with abilities: fun and variety.
- Too much divergence from the principles of Primary Attributes and Opposing Context Opportunity Costs will allow players to increase for trait combinations that are imbalanced against other trait combinations. Resulting in broken or over powered play styles.
- Too little option to optimise for Binary Trait pairs lowers player ability to specialise, customise and take risks on trade offs towards their preferred play style.
2.2 Critique - Modeling Aavegotchi trait map against the design principles
Attempting to understand how good the Aavegotchi trait mapping is might by begin comparing to the above model and identify: How balanced and fun might the game play outcomes be?
Current state Aavegotchi mappings
Source:
The Gotchiverse Game Bible: Chapter 2 | by Aavegotchi | Jan, 2022 | Medium
This stage of the analysis primarily focuses on the PVP values of the trait map. Energy, Aggression, Spookiness and Brain Size.
A trinity is a structure of relationships between value types and not exclusive to any absolute value. So while abilities like carrying Capacity, Movement Speed, Handling and Vision Range have not been included in the Part 1 research, they can easily be modelled to reveal how their abilities might interrelate.
Aavegotchis General Trait mappings are not comparable to the model’s stipulated Primary Attribute (Which is fine because we love the curve.) Current state Aavegotchi General Traits behave more like Opposing Context Opportunity Costs.There is not yet an obvious way to apply the design principle of Primary Attributes in the sense of ‘a master trait that binds together confluent game play abilities to balance opportunity costs.’
It follows that it may not be valid to compare Aavegotchi trait mappings to classic trinities and say things like, ‘such-and-such trait combination is always imbalanced in other games, and so it is imbalanced in Aavegotchi too.’
Balance - is the curve the holistic ratio of opportunity costs between core traits in Aavegotchi? As the strength-agility-intelligence trinity is to DoTA ect? Balance might be exclusively determined by the curve (how lucky the Aavegotchi was with the distribution of randomly generated traits) and what ability values developers attribute to traits. Meaning, statistical analysis may be required to settle the balance debate.
Above are a sample of selected trait maps arranged adjacently to look for examples of binary traits: extremities that share a theme of abilities but control these abilities in different directions.
A) Yes, Melee Damage and Health Regeneration are adjacent binary traits - the pair govern the ability to control the attrition of damage.
B) Maybe, Ethereality and Ranged Damage are somewhat adjacent binary traits (Ethereality is more related to damage mitigation than damage attrition) If we say the pair do govern the ability to control the attrition of damage then they do so in redundancy to the above pair A).
C) No, it is difficult to argue that Armour and Attack Speed make a pair of binary traits. Armour is another damage mitigation ability and attack speed controls the rate of attack per second.
Variety - traits that control for damage and durability seem to be highly saturated, maybe even redundantly (several traits effect similar outcomes). Is the carry capacity trait enough to add some variety to the heavily offensive and defensive trait combinations? Can every Aavegotchi trait dynamic be optimised towards a satisfying build?
The system of opportunity costs in Aavegotchi trait mappings is not inherently comparable to the model. Opportunity cost is intrinsic to the choice of direction in a general trait. Every time a player levels they are faced with opportunity cost in what gameplay outcomes they choose to optimise for.
Summary
Analysing Aavegtochi trait mappings through the lens of a trinity based model is not a fair assessment. They are not comparable. In order to produce a complete critique Aavegtochi trait mappings must be scrutinised for their own balancing dynamic.
Below is an attempt to analyse how Aavegotchi trait mappings might interrelate as measured by the model’s design principle outcomes: fun, variety and balance.
2.3 Comparison matrix of Aavegotchi traits
More bell curve
Finding a way to make sense of the Aavegotchi trait map requires some context setting. Load up the curve.
Source: Aavegotchi Stats
Analysis of current trait mappings must take into consideration that Aavegotchi owners have selected for some traits on the extremities of the bell curve >98 and <1. However, the majority of randomly generated traits occur within un/common rarity 10-90.
This means there are relatively few Aavegotchis likely to be fully stacked with PVP traits. But how many PVP stacked gothcis are there?
The case for looking at trait balance through trait pairs
Frequent rare-mythic NRG, AGG, SPK, BRN = 2 @ >540
Source: https://ghst.gg/market
A quick sample of the market by minimum BRS reveals that Aavegtochis with two or more rare-mythic NRG, AGG, SPK, BRN traits do not become frequent until BRS >540
If more than half of gotchis from 540 BRS upwards feature more than two rare-mythic traits, how many gotchis are we talking about?
Answering this would require analysis beyond my capability but we can proxy for total gotchis above 540 BRS. Searching for Aavegotchis with minimum 540BRS at reveals:
Aavegotchis >540 = 1393 gotchis*
https://api.aavegotchi.land/gotchi?desired_traits=x,x,x,x,x,x&brs_min=540
Source: Aavegotchi.land
So of the total 17159 Aavegotchis summoned, only something around ~8% are likely to be blessed with two or more rare-mythic traits (and this includes Eye Color and Eye Shape traits that are not player dynamic balancing concerns)
*edit: @actaeon pointed out that the API doesn’t include wearables and that the leaderboard gives a better view of how many gotchi possess >2 rare-mythic NRG, AGG, SPK, BRN traits. Down to around #3500 on the leader board it is more frequent to see >2’s. 3500 / 17159 gotchis is 20% of gotchis. So perhaps a three-trait analysis can be argued for. However, he ability values should scale on the curve and the variance might increase between strongest, second and third strongest traits as players attribute their three stat points per level. So it might be worth getting some statistical capability to weigh in on this matter
Source: Aavegotchi Rarity Farming Leaderboard
What about dreaded triple threats or even four traits rare-mythic in all four PVP traits? (NRG, AGG, SPK, BRN)
Frequent 3 rare-mythic NRG, AGG, SPK, BRN = 2 @ >560 BRS
Source: ghst.gg/market
Aavegotchis with three rare-mythic NRG, AGG, SPK, BRN traits do not become frequent until BRS >560
BRS >560 = 201 gotchis
https://api.aavegotchi.land/gotchi?desired_traits=x,x,x,x,x,x&brs_min=560
Source: Aavegotchi.land
Thats only ~1% of gotchis likely to have three or more PVP traits.
Assumption 1
If only ~8% of gotchis are on average likely to possess two rare-mythic PVP traits it might be fair to look at strong trait pair combinations to evaluate balance. If any two traits are strong, how do they balance against all other trait combinations?
The case to exclude Eye Shape and Eye Color from analysis
For starters they haven’t entered the conversation until now, but we should probably talk about them.
Source:
The Gotchiverse Game Bible: Chapter 2 | by Aavegotchi | Jan, 2022 | Medium
We’re looking for fun, variety and balance in the dynamic trait combinations a player is able to select from and attribute points to with each level. Eye shape and eye color are static values that a player cannot increase or decrease after they have summoned their chosen gotchi from a portal. It doesn’t make sense to include static traits in this analysis.
Further justification
Static shape-specific buffs
Eye Shape traits behave like attributed buffs and not at all like the other five trait pairs of dynamic trade offs between two skills.
Handling verse Vision Range
Vision Range and Handling are slightly different to all other traits in that they don’t only govern ability but arguably define player core quality of life factors. Extremely poor handling or extremely low vision range are undesirable experience outcomes. We might then assume that the values attributed to these variables would occur on a tight range so as not to punish players beyond their threshold of tolerance for poor handling and vision.
Assumption 2
Exclude Eye Shape and Eye Color traits from balance analysis.
We can more rigorously analyse how trait combinations might impact game play outcomes if we make one more set of assumptions: what are the win conditions in the gotchiverse?
At a quick glance it would seem there exist three themes in traits
- Defensive traits for farming Alchemica
- Defensive traits for tanking damage
- Aggressive traits for inflicting damage
Based on these trait themes and the alpha dropped so far, players might be drawn towards the following goals. (speculating on how users may behave in a given application is dubious at best.)
Kill stuff
Killing things is fun and more fun when Alchemica’s involved (and unconfirmed XP: White Paper, 5.5 Mini Games) Aggressive traits will be a key capability for satisfying the gamer’s urge to kill, and Aavegotchis strong in aggressive trait combinations, including damage types, attack speed and movement, may be incentivised to optimise towards killing things.
Farm all the Alchemica
Players are incentivised to collect, defend (and perhaps kill) for Alchemica. Carrying Capacity will be a key capability for winning the Alchemica game and Aavegotchis who’s strongest trait is low Energy may be incentivised to optimise towards farming.
Assumption 3
Win conditions in the gotchiverse are to kill things and collect Alchemica.
BRS up only
Also, maximum health and the sprint stamina mechanic only go in one direction. They wont make the comparison table but will be discussed in analysis.
What classes might emerge?
In a genre breaking game such as Aavegotchi, classes are likely to be a bottom up phenomenon. Classes or gameplay styles might emerge how players respond to the trait dynamics and win conditions that are presented to them. Putting a label on a group of dynamic traits doesnt mean players won’t completely ignore those classes and find their own efficiencies to optimise towards.
Its more significant to measure whether all gotchis have a fun options to optimise for than it is to give people buckets to play in. That said there are some high level patterns in the traits that are obvious to call out: farm, tank and damage per second (DPS).
*‘Kill’ used rather than DPS in this study to keep the language plain and to point to the win condition that damage output builds are supposedly optimising for, killing enemies.
Let’s test how the rarest/strongest trait in a two-rarest-trait combination may establish the win condition that makes sense to optimise for.
2.4 Synthesis of trait mappings - strongest pairs
The way to read the following tables is to compare the ‘rarest trait’ against each ‘second rarest trait’ one at a time and ask:
What what play style might make sense to optimise towards base on only these two traits? The answers (based on opinion) are coded for yellow ‘Farm’, green ‘Tank’ and black ‘Kill’
Carry Capacity
Carry Capacity exclusively incentivises farming
A gotchi who’s rarest trait effects high Carrying Capacity is arguably exclusively incentivised to optimise towards farming. Defensive secondary trait combinations extend a players capability to bank that Alchemica. Offensive trait combinations give a player some damage output as deterrence, but at an opportunity cost of any meaningfully reinforcing defensive or offensive traits to make the gotchi a viable killer.
Movement speed
Movement speed equally incentivises killing or surviving depending on the next rarest trait
Movement speed can be argued to extend the effectiveness of offensive secondary traits by either kiting* range or enabling melee range to compress attacks per second. Movement speed led pairs would then optimise towards killing win conditions, whether that kill be incentivised for Alchemica, XP or just for fun, unknown. Equally movement speed can be argued to extend the effectiveness of defensive secondary traits by adding escapability to durability. It is unknown what a fast tank would optimise for: more speed or more durability? Neither directly effect the assumed win conditions of actively killing for and collecting Alchemica.
Finding - It is not clear what the win condition might be for a fast tank. Players might not necessarily maximise kills or Alchemica by running away with a diminished carry capacity.
*Kiting = MMORPG slang for a method of killing enemies from a distance by using range and mobility to win, rather than passive durability traits
Armour
Armour is the wild card
Armour is the only trait that pairs variably with other traits for all win conditions and play styles. That is to say that high armour serves simply to extend the incentive of playing to the secondary trait. Fast gotchis survive escape. Farmers survive to cash in Alchemica. Killers survive their fight. Does this mean armour is balanced? Or a submissive trait that does not incentivise leveling any further? All we can assume is that the incentives of possessing high armour might be highly variable to what game play outcomes the next rarest trait imply. Which is a good outcome for play style variance. More traits could be like armour frankly.
Melee Damage
Melee Damage kill them all
Melee Damage exclusively incentivises offensive proficiency unless Carrying Capacity is next strongest. Your gotchi might be an axe murderer or a squishy farmer who’s better at swinging his pitchfork than shovelling Alchemica. Is Melee Damage a varied enough leading trait when it seems to incentivise almost every other trait towards killing stuff? We assume killing stuff is a win condition so we’ll move on.
Ranged Damage & Attack Speed
Attack speed kill them quick
Attack Speed exclusively incentivises offensive proficiency unless Carrying Capacity is next strongest. Farming second to high attack speed might be the least viable of the selected pairs: no durability or damage to make the attack speed an effective defence to depositing Alchemica.
Ranged Damage same as Melee
Ranged damage also incentivises offensive optimisations of all other traits except carrying capacity. It’s maybe a slightly better primary trait for a secondary farmer as it may be effective to kite at range while returning Alchemica.
Finding - Is having three primary traits that control damage output an imbalance to win condition and game play variety?
Health Regen & Ethereality
Health regen or Ethereality: does it matter?
Whether a gotchi’s rarest trait is high SPK or low SPK: the optimisations are equal across all other secondary traits. Defensive secondaries mutually reinforce a tanky build*. Offensive secondaries add durability to Killers. The only difference is that Health Regen will allow gotchis to sprint for longer due to how sprinting drains health points. Again, it is not clear what win conditions a fast tank can optimise for. This analysis is not concerned with lore however it should be mentioned that SPK was originally described as a wild card trait and it might be disappointing to some spooky gotchi holders that SPK only codes for tank ability.
*technically Ethereality ‘Your Gotchi’s chance to evade damage and reduce wall knockback’ is a more volatile durability trait than armour or health regen - however the outcome is essentially damage-mitigation-over-time like other durability traits. (reducing wall knock backs is a useful mobility perk - and very spooky)
2.5 Findings
Some traits are dominant and some traits are secondary.
Dominant traits - strongest and next strongest traits in a pair that define the win condition to optimise towards:
- Carry capacity
- Melee Damage
- Attack Speed
- Ranged damage
Submissive traits - strongest traits in a pair that defer to the next strongest trait to define the win condition to optimise towards:
- The entire SPK trait: Health Regen & Ethereality
- Movement speed
- Armour
Not a lot of variety in trait combinations.
All dominant traits are offensive traits, except carrying capacity which we say is the only trait that principally incentivises farming. All submissive traits are defensive traits or movement speed. This might imply that players may have less preference to level defensive traits in the interest of optimising for more active offensive traits or farming.
Role call
- Six farming reinforcing pairs defined by a single dominant farming trait
- Five defence reinforcing trait pairs
- Thirteen kill reinforcing pairs from nine submissive type traits and four dominant type traits.
We should define the killer classification a little more to see what it contains. Here are all the selected trait combinations in one view below: Color coded for ‘black’ high DPS builds and green Tanky Killer builds.
The way to read the following tables is to compare each ‘strongest trait’ in the top row against each ‘strongest trait’ down the column and ask:
What play style might make sense to optimise towards based on only these two traits? The answers (based on opinion) are color coded for ‘black’ high DPS builds and ‘green’ Tanky Killer builds.
4 primary traits X 2(low/high curve) X 3 secondary traits =24 combinations in this sample
This highlights the idea that there may only be five viable builds optimised towards killing things. Optimally proficient killers with offensive rarest trait pairings are obviously at a disadvantage in durability. It is encouraging to see that this is a relatively infrequently occuring build.
Tanky killers would be less proficient at damage output, disadvantaged by their next rarest trait being defensive. Tanky killers are also less mobile. In this table of strong pairs - you can only have a Fast Tank or a Tanky Killer. Movement Speed is the opportunity cost to Tanky Killers. This equally applies to Ranged Damage-Tanky Killers - they may not have the mobility to kite effectively and could be conceivably be countered by High DPS gotchis (which we say are an infrequent build.)
Ranged Damage-Movement Speed gotchis are a concern for kiting imbalances. Here Fast Tanks might find their usefulness we were looking for. The stamina mechanic should limit how long a Ranged Damage-Movement Speed gotchi can use sprint to kite before their health goes too low. A Fast Tank (less limited with stamina) should be able to run down a kiting gotchi and smaash it.
Finding - There appears to be a satisfying dynamic between trait ratios to enable developers to balance the combat dynamics for future PVP solutions.
For example, if melee damage scaled on a higher range of values than ranged damage (say melee damage = 2-10 ranged, damage = 1-9 and melee damage is +1 higher than range at every level) then balancing the game within the constraints of the current trait mapping should be feasible.
The most salient quote in the original post that is resonating in these conclusion is this:
Rather than looking to at balancing the strongest but most infrequent trait combinations, it might be more impactful to be looking at the weakest (and most frequent) trait combinations and ensuring there are
- viable builds for all gotchis
- variable abilities inclusive of all players preferences.
Below are some highlighted builds for strongest pairs that provoke some concerns.
Why tank?
It is not immediately obvious what the incentive may be to optimise purely for defensive trait combinations that only tank damage.
Attack Speed-Carrying Capacity
Is this trait pair combination potentially underserved? What can a farmer optimise for when their strongest trait happens to be Attack Speed?
Redesign the trait mapping?
This study sough to understand whether there was cause to redesign the trait mappings. The conclusion of this part of the study is that there is no observable critical issues with the current trait mappings. After several days of close analysis it is apparent that the nature of Aavegotchi statistics makes it difficult to intuitively comprehend the balance of the trait mappings. There is the concern that raising a vote towards changing the trait mapping at this stage may inhibit the development process in achieving balance.
The proposed trait mappings are a product of the full set of abilities and win conditions being built into the gotchiverse. If more about these are made public then additional analysis may be warranted.
There are the following outstanding questions:
What about hard tanks?
It is not immediately obvious what the incentive may be to optimise purely for defensive trait combinations that tank damage. If a gotchi’s two rarest traits are defensive, what rules of the game might they excel in achieving and further optimising towards?
Balance
- Balance seems to be viable across the board however, it is not clear what the win condition might be for a fast tank. Players might not necessarily be in an attractive trait pair to optimise for kills nor optimise for farming more Alchemica by running away with a diminished Carry Capacity.
Fun
- Are Spookiness traits different and interesting enough to warrant attributing points towards? Or is it really much-of-a-muchness where SPK falls between health regen and evasion? As outline above: movement speed does not appear to be congruent with high durability. So how attractive is the stamina/sprint-health-burn mechanic really?
Variety
- Is having three different traits that dominantly skew trait pairs towards optimising for more damage output too exclusive of all other play style preferences?
There’s a great paper from 2019 that classifies different player motivations and behaviours. They synthesised all the existing literature down to introduce five player motivations:
- Aesthetic orientation
- Narrative orientation
- Goal orientation
- Social orientation
- Challenge orientation
“I don’t fit into a single type”: A Trait Model and Scale of Game Playing Preferences, 2019. Tondello et al. available at https://www.gamefulbits.com/wp-content/uploads/2019/06/2019-A-Trait-Model-and-Scale-of-Game-Playing-Preferences.pdf
Broadly speaking, the Aavegotchi ecosystem has something for everyone of these motivations. Within the current trait mappings its not yet obvious what game mechanics might drive cohesive team play in battles. Social orientation might be a gap in the current play styles available. Some people might like to optimise towards support game play goals. Carry Capacity might be a support capability in guild/team contexts. Depending on how Alchemica collection works.
1/2
