Synthesizing the Emergent Gameplay Discussion

Investigating how fighting games are developed- by designers and players

I watched all of the YouTube discourse so you don’t have to.

Before we begin, I must point out that this is going to be unraveling a long thread of discourse, and while it seems like it may be meandering or retreading old ground, there’s value in understanding these arguments. If we understand what has changed on a fundamental level about fighting game design, we can figure out when and why it began changing. Once we understand that, we can predict what games will look like once developers reconcile their attempts to bring new players into the genre with their attempts at wild experimentation. We can also recognize that the development of a game is also tied just as much to players who are passionate about it as it is to the actual balance team. Once these principles are taught and understood, ideally players overall will have a more positive outlook on current design trends while understanding the reality of the perceptions of older games in context.

Before GGST

Over a year ago, my stream and other players in some Guilty Gear-centric communities I was a part of asked me some questions about how I felt regarding the GGST betas (the ArcRevo 2019 build, the CBT, and the two OBTs.) I was specifically asked to define player expression and whether I felt Strive encouraged that. I defined player expression very literally- the game’s capacity to allow you as a player to express yourself. If fighting games are a conversation, then the games should allow you to ‘communicate’ with your opponent back and forth, and games that have high degrees of player expression allow for more in-depth ‘conversations.’ I focused on three main areas where this level of expression would be found:

1. The most obvious is the roster. A large roster typically has a variety of superficially different playstyles that appeal to all types of players. Characters become part of your identity as well- if your main doesn’t come back in the sequel to a legacy game, it’s highly likely you won’t be playing the new version. Conversely, if there’s too few characters to choose from, and thus very few different playstyles, it may be harder to pick one that speaks to you.
2. The kit design for each individual character is the second area. Even changing the typical ‘shotoclone’ archetype slightly results in completely different characters and is why those clones continue to be added to Street Fighter, for instance. Much like with the above, if your character is nerfed or reworked between a significant version of the game, you may not find yourself able to express yourself in the same way as before. Of course, if your character gets some buffs, you’ll be over the moon about it.
3. The third and final main area is the system mechanics. Some games will be deliberately tuned in a limiting way so that the focus becomes primarily on neutral exchanges, horizontal combat, and more methodical gameplay, while others will be a lot faster, with longer combos, and more verticality. There are also other engine quirks that can be used to extend offense, defense, or movement, and whether they’re intended or not, they can end up becoming critical parts of the metagame- and of the level of player expression.

Them’s Fightin’ Herds is an example of a game I like that has a limited Area 1 (only seven cast members) but wide Areas 2 and 3 (lots of character tools, and combo systems that only have rules against infinites.)

My conclusion was that GGST would overall have a lower level of player expression than previous games, but I didn’t know how low it would be by the time the game’s development cycle was concluded. The game as it existed on launch had fewer characters, slimmer character kits, and several existing system mechanics had been unified into single tools. However, I didn’t know if that would totally limit discovery and exploration or if that level would be raised by subsequent patches, so I didn’t want to make any judgements at launch. For what it’s worth, I’ve been having a ton of fun playing GGST I-No and I think the balance team has taken lots of steps to make sure that the game has more options rather than less. Things like providing net buffs to nearly every cast member and adding new Gatling routes makes me feel like the game is being polished towards something where player expression and open-endedness is encouraged.

The definition of “Emergent Gameplay”

This is the part where everything gets murky, because so many people have different definitions of what “emergent gameplay” is to the point where, much like “player expression” above, it has become something that bends to what whoever is talking at the time wants it to mean. Let’s start with a clerical definition and work our way from there. Eventually, we should be able to define emergent gameplay, list good examples of it in specific titles, and determine whether titles developed in the current generation still exhibit it.

According to Wikipedia, emergent gameplay “refers to complex situations in [games] that emerge from the interaction of relatively simple game mechanics.” (emphasis mine) The article uses many examples, only one of which involved fighting games- the development of combos in Street Fighter II. Emergent gameplay can range from the sheer amount of creative possibility found in Dwarf Fortress’s lack of a win condition or Minecraft’s nominal gameplay loop, to glitches like combos in SFII or the Halo 2 BXR glitch which lets you fire faster. Essentially, emergent gameplay can result from developer intent- deliberately creating an extremely open-ended system, to the point where such elements in other games are often affectionately called ‘open-world’ design- or it can result as an unintended consequence of something that was missed in initial design. We can see that the definition of emergent gameplay mostly follows how we previously defined player expression, especially with focus on the second and third areas.

Minecraft can be literally vast just as often as it can be mechanically vast.

BrianF’s video about emergent gameplay in fighting games is clinical and mostly provides examples of the above definition in various titles across the genre. His main examples, in addition to the Halo 2 BXR glitch and SFII combos, focus primarily on fighting games developed around the turn of the decade: roll cancelling in select Grooves in CvS2, the entirety of the MvC2 meta, and wavedashing and L-cancelling in Melee. However, he also provides modern examples of emergent gameplay that ended up being patched out rather than encouraged and left in: specifically, Oro’s fastfall and Dan’s infinite in SFVCE Season 5. While I would argue that Dan’s infinite was patched out in a fun, in-character way (the fireball link required for the infinite occasionally, accidentally, being a ‘good’ version of the fireball that ends the combo,) the Oro fastfall is an example of a technique that was unintended but made the character fun to play and granted an additional layer of options to his kit that was ultimately patched out.

While Brian laments the removal of fun, unintentional discoveries such as those that were removed in later versions of the game, he also points out that some emergent elements should rightfully be patched for the health of the game. He references obvious examples such as Grappler’s throw infinite in the DNF Duel beta as something that is unintentional but would plainly be not enjoyable for most players to experience. (We’ll talk about other design elements that get trimmed, and why, later.)

BrianF’s video was in turn inspired by Kayin’s tweet about older games being more emergent than their modern counterparts.

In his “What Happened?” video on the Street Fighter III series, after recapping the game’s troubled production and initially poor reception with the FGC and casual arcade crowd, Matt McMuscles ends with this quote on the game from Shinichiro Obata, the battle planner for Second Impact:

“You know, there are many different approaches you can take to game design. One approach, which we took in Street Fighter III, is to design your game around unanswerables. I think with any game, players will search for the best tactic, the best strategy, like, if X happens, then you should always do Y. If you do this here, you’ll always win. There’s competitive games like that where the match is essentially a confrontation of theoretical knowledge that each player has built up. But Street Fighter III is a game that, by design, doesn’t have a fixed answer to those questions. There is no best tactic. You could spend your whole life trying to find the perfect theoretical approach to a situation in SFIII but it will never be quite right. You always have to be reading your opponent in the moment. You can’t just fall back on your theories. It’s a game that lets you search for answers forever.”

This quote forms the basis of what Maximillian Dood’s definition of emergent gameplay is- not defining it directly, but citing BrianF’s video as an example and pointing at battle royale shooters like Fortnite as modern games that encourage it in their design. Maximillian seems to have several examples of pain points- things he doesn’t like in modern games- and contrasts them with what he believes is emergent gameplay design in other fighting games. For instance, he laments that cancel windows in some games are only available at the beginning or end of certain moves, and that you couldn’t cancel into V-Trigger in SFV as a free cancel off every button. On the other hand, he praises active tag mechanics in Power Rangers BFTG, custom assists in Skullgirls, and the Conversion mechanic in DNFD as examples of what he describes as “loose” gameplay, or mechanics that otherwise open the possibility space for the player. While Max cedes the point that eventually even the most expressive games will be ‘solved’ and have a well-defined meta of optimization, he argues that those games will take longer to solve than modern, more rigidly developed games that allegedly have less opportunities for emergent gameplay.

What emergent gameplay design can foster

Max’s definition of emergent gameplay, as well as the examples he provides, is paradoxically vaguely-defined and limiting in what it implies. It mainly seems as though Max is fixated on being able to get away with basically anything on offense, which makes sense in principle. Offense is where most players have the most fun- it’s where you do your big damage and flashy combos, it’s where you get the tactile satisfaction from your buttons, and it’s probably what you practiced the most in training mode. On the other hand, Max’s specific examples fall apart after enough scrutiny. For instance, VT meter in SFV has been expanded in uses greatly since the game’s initial season, with customizable VTs, V-Reversal guard cancels, and V-Shift parries to start offense again, in addition to how strong many VT activations already are. The lamentation of tight special cancels only at the beginning or the end of a certain move is also odd since, not only is that how most fighting games function, it would be horribly imbalanced if you could do a meterless cancel at any point in the move (and in fact, a main criticism of YRC in Xrd was that it was so cheap while allowing you to cancel neutral!) Even the DNFD example is very strange since, even if you could do Conversion into anything at any time during your offense, movement was very limited, the Gatlings were short and stunted, and the addition of a guard button meant that basically all offense was easy to defend against, reducing the most common mixup to strike/throw.

Aerial Rave and Guilty Gear-style Dust combos will always be one of my favorite things to do in any game.

If a mechanic has multiple uses and lets you carve out your own playstyle, even if it doesn’t let you do ‘open-world’ combos, it still counts as an example of emergent gameplay. My VS2 VT1 Akira will play differently than other Akira players who use different loadouts, and even if they’re using the same loadout they may spend their V-Trigger meter on other tools like V-Reversal or V-Shift. Even if the possibility space only has a handful of branches, and even if only one of those branches is good or optimal, it still leads to diverse playstyles at most levels. Of course, even if you have a truly open-ended mechanic like Conversion, it doesn’t matter if the cool offense you create with it can’t actually go anywhere. Oftentimes, mechanics that create the feeling of open-endedness may either not be very strong, or can be overtuned in the other direction.

Romolla’s perspective on emergent gameplay focuses on this primarily by examining characters in Xrd and what was lost in the transition between Rev2 and GGST. Romolla argues that even though it’s true that modern iterations of games like Guilty Gear tend to have very linearly-developed characters with a smaller set of tools, it’s that same set of tools that makes their gameplans and win conditions better defined. She points at tools like Riot Stomp that went missing from Sol’s kit as moves that were only good at low levels of play and became useless once you faced players who knew the counterplay. From her perspective, several characters were designed this way wholesale in Rev2, with Bedman and Jack-O’ being extreme examples of characters who are totally bad once you figure out the gimmick, and characters like Raven and Elphelt who are incredibly strong even without playing to their unique mechanic. Romolla’s point overall is that not only does it hurt the metagame to have entire toolkits that are functionally useless at certain levels of play, but casual players will not have the patience to learn how to pass the abundance of knowledge checks and be turned off from playing entirely.

Eddventure manages to use Countdown here to great effect, but very few Jack-O’ players could probably get this setup and there may even have been a more efficient option.

Of course, that’s not to say that tools with incredibly niche use cases don’t still end up in modern games. Sol’s Heavy Mob Cemetery super, for instance, can find use as a full-screen whiff punish due to being an unblockable guard point, but in basically every other scenario it’s useless. You can’t combo into it and Sol has an abundance of better defense options, meaning there’s basically never a reason to use HMC over Tyrant Rave. Similarly, GGST Jack-O’s Countdown command can often provide some extended mixup situations and even give her an unblockable. Unfortunately, it’s very expensive for her Servant Gauge and is thus almost never used unless you’ve explored those scenarios before. At the very least, in newer games that are still going through their development cycle, there’s opportunities for them to be patched or buffed later. In a subsequent version of the game, HMC might be a much less situational super, or the Countdown unblockables may be more viable. Still, to Romolla’s point, trimming the fat across the roster means that it’s easier for the attacker to put their gameplan together, and for the defender to learn how to use their own tools in response to offense.

Synthesis: The history of Guilty Gear

When I wrote about Guilty Gear’s level of player expression, I mentioned that the series has a history of taking mechanics from previous games and ‘simplifying them’ by unifying them with other principles or tools and making them much easier to use. In other words, even when tools or mechanics disappear, they are often replaced with something that fulfills the same or similar purpose. This is a principle outlined thoroughly in AzeDev’s video on “The Simplification of Guilty Gear,” but it’s also easy to explain succinctly if you’re familiar with the games. The difference between Force Roman Cancels (FRC) in +R and Yellow Roman Cancels (YRC) in Rev2 is almost night and day in their execution and use-case. FRC is a cheaper cancel with a tight execution window (it can be wider, but most of the time it’s only 2f) that can only be performed on a handful of moves in a character’s kit, while YRC can be done at any time or on almost any move.

Jam’s most useful FRCs are on her 236P (for instant overheads) and her throw (for better throw combos.) Learning these are essential for her mixup game and throw knockdown/damage respectively, but the nature of FRC makes them difficult to execute, or for her opponent to remember the counterplay.

While FRC is an example of emergent design, as it opens the possibility space for players after certain moves, its tight execution and the need to know exactly which moves have FRC points makes it very constricted and limiting. On the flipside, while the changes to the system made when FRC functionally became YRC may have made the mechanic exceedingly expressive, it also became overcentralizing to the game. You will often hear Xrd players describe YRC as being balanced by itself- in other words, the strength of YRC was mitigated by the opponent having the ability to YRC back. In Strive, this was rectified by changing YRC to a Blue Roman Cancel (BRC) that cost twice as much and could only slow the screen if the opponent was in a particular area of effect. This let BRC remain a strong pseudo-parry while making it costly and preventing it from defending against all full-screen actions. Of course, the RC system itself expands in GGST with the Drift and Fast RCs, allowing for even more expressive combo routes in addition to the rebalanced use cases.

Strive has continued this development philosophy in other ways, such as the unified Gatling system. While it may seem stunted at first glance, with the inability to chain P into K into S into H and skip around, it can also provide more combo utility in other ways. For instance, in previous Guilty Gears, not everyone followed the same set of Gatling rules, meaning it was hard to switch between characters or learn new ones. Slayer and Xrd’s version of Ramlethal, for instance, are examples of characters who don’t play by the typical Gatling cancel windows, instead relying mostly on links and baked-in strings respectively. Even among the more lenient Gatling sets, not every character plays the same way, sometimes being unable to cancel normals into command normals. However, in GGST, everyone has the same Gatling sets, meaning everyone can go from P buttons into command normals into special moves and so on. While the efficacy of each Gatling table is still situational based on the actual toolkits present, it can make picking up the fundamentals of a handful of characters easier while also giving you a better sense of how to create combo routes. Both the old and the new Gatling systems limit and foster combo creativity in different ways. The emergent design of the new system is also slowly being expanded, since every GGST character can do a full range of Gatlings in the air as of patch 1.10.

Conclusion

As Romolla pointed out, it’s often the case that the emergent gameplay involved in the interaction of certain mechanics may not be appealing for players to interact with at all. One of my favorite games currently, Melty Blood Type Lumina, has been criticized repeatedly for how strong the Shield mechanic is when defending and parrying offense, especially on reversal situations or against zoning. I’ve written a primer to MBTL defense, and in it, it’s established that Shield has a variety of counterplay options. While it stuffs mids and the B followup can be used to close the gap against zoners, it’s a committal option whose recovery frames are worse the more Moon Gauge is expended, and punishing it with the correct overhead or low option (or a throw) will lead to a Fatal Counter (an untechable combo starter.) Beyond that, interaction with Shield encourages mindgames like delayed Shield options or anti-Shield tech such as Shield/jab/throw option selects. There are even characters like Aoko or DA Noel who have built-in anti-Shield tools using delayed okizeme options. However, even if there is a ton of counterplay to this option, at a low- or mid-level of gameplay, the perception is there for players new to the game that Shield is overpowered and meta-defining. Thusly, it doesn’t matter if the mechanic is expressive or fosters emergent gameplay, because if people don’t think it’s fun to interact with then they aren’t going to want to learn how to interact with it.

Most people will probably see this and just assume this is the game, like Smash 4 and run-up shield.

I probably wouldn’t agree with anyone who says that the fighting game genre has been simplified or has had its tendency towards emergent gameplay reduced. For every mechanic that has been toned down in functionality, there’s always something that exists in the same game that provides more creative options. Kits in BBTAG have been reduced overall, but active tag mechanics let you play fast and loose with new combos and team compositions. GGST I-No might not have YRC VCL anymore, but she has a frame 0 command grab super and immediate IAD out of hoverdash now. I’m sure in the final version of DNFD, all the open-ended offense that Max was heaping praise on will actually go somewhere, whether it’s because of developer intent (changing how defense works when the block button is held) or because of truly emergent gameplay (players figuring out how to optimize mixups.)

The true challenge for developers who want to encourage emergent gameplay is balancing what would be fun for everyone to use versus what would be balanced in the final product versus what would be useful during gameplay. It’s a dilemma that many games over the past three decades have attempted to solve, and games in the future will all come up with different solutions. Some games will shuffle up their box of options, others will double down on what they think works, and still others will get wildly experimental with their totally new systems. From there, it’s up to the players to learn how the game works and how its systems interact together- even the most ‘rigid’ of systems have something expressive about them. The key to finding happiness in the genre is to pick the game with the solutions you like the most, because then those games will let you express yourself how you like.