The Pros and Cons of Skill-Based Matchmaking
SSBM’s newest netplay build doesn’t have SBMM. When you arrange these Babylonian letters in that way, it means that that Super Smash Bros. Melee’s newest way to play single opponents online through an emulator doesn’t have Skill-Based MatchMaking.
Not yet, anyway, though they’re working on it. That netplay build does have rollback netcode right now, though, which I think is one of the key ingredients to saving fighting games in this age of plague. More on that from me in the future.
Anyway, skill-based matchmaking is today’s topic, though the quality of the connection is certainly a key part of the online multiplayer experience. I’ve got more of gaming’s pasts and presents for you all. Mainly, I want to talk about some SBMM pros and cons. Lots of abbreviations in this one. A few weeks ago, these various algorithms designed to pit opponents of relatively equal skill against each other became a hot button issue, to say the least. The spat, in what I must say is a rather typical fashion for him, involved Melee’s hot take artist and Swedish beast himself, William “Leffen” Hjelte, taking on what seemed to be large swaths of the Call of Duty community.
Whoever is “right” or “wrong” is a matter of personal interpretation, as far as I see it. Skill-based matchmaking has clear pros and cons, and it can vary widely from game to game. Melee, for example, isn’t all that fun for either player when there’s a huge skill disparity, but that netplay build’s Direct Connection and Unranked modes are all we’ve got for now, and it’s not making the game unplayable by any means. Newbies have a chance to play against their favorite top-level player, an even higher one than if they had entered a major tournament. That said, one-on-one games almost always come out of the box (or more often from the virtual market at this point) with skill-based matchmaking, if not necessarily rank-based matchmaking. Dragon Ball FighterZ, for example, doeshave a ranked mode, which is the primary multiplayer experience for the game, but it also allows players to duke it out in Casual matches that don’t affect their titled ranks. Casual matches do still count toward altering one’s colored square, which designates vague tiers of win percentage. There are settings within casual mode to widen or lessen the matchmaking pool based on these colored squares.
This compromise of being able to increase or decrease the number of potential matchmaking partners has appeared in various fighting games. Street Fighter IV, for instance, had settings that let its online warriors play people of relatively equal points or players with more points than themselves, but crucially not only players with less points. Searching for that last option would reduce the amount of “smurf” accounts, but it’d be a pretty ridiculous option to include. If you’re not aware, SFIV had numerical ranking systems like Elo both generally and for specific characters. These were called Player Points and Battle Points, respectively. I assume this adjustable matchmaking was tailored around the general Player Points, but either way, it was a very useful addition to the game’s online modes. That game’s successor features matchmaking features like hardware settings (Street Fighter V has crossplay between PS4 and PC) and connection status, as well as the standard ranked/casual split. Tekken 7 recently introduced indicators for ethernet and Wi-Fi connections before confirming the match, similar to Dragon Ball FighterZ’s (and some other fighters’) indications of connection bars before locking in that you want to play. Lessening these optional matchmaking restrictions can decrease queue times, while adding them allows players to tailor their online experience somewhat.
Other Games Can Find the balance
Even when it comes to team games, so long as they’re designed to be competitive, most players will opt for relatively fair matches. By “competitive,” I mean multiplayer games designed with balance and skill in mind, without stuff like Mario Kart’s items, random within a range of places in the race… though maybe people do run Mario Kart tournaments. “Competitive” can also then mean any non-cooperative multiplayer games, though I’m sure many cooperative experiences rightfully feature SBMM. Whatever your definition of “competitive” is, a search for fair matches means a necessary embrace of skill-based matchmaking. Thrashings are nice to see when your preferred team is winning, and solid strokes of the players’ egos, but are not nearly as exciting as close victories against heated rivals with big stakes. I’m pretty sure Lebron James, or pretty much anyone in the NBA, for that matter, doesn’t want to play a pickup game with high schoolers or probably even college players.
That said, challenges like this one, where one Radiant Valorant player takes on Five Irons (in the sequel, the Radiant Reyna gets paired up with another Radiant player) can be interesting exhibitions of the skill disparities that lend themselves to skill-based matchmaking. Imagine, if you will, three more Radiants with those two. That vast difference in team skill levels would technically be possible without SBMM.
Battle royales like Player Unknown’s Battlegrounds or Fortnite feature around 100 players, so it can be hard to find that many people of relatively equal skill without drastically slowing down queue times. But even Fall Guys has skill-based matchmaking, if not ranks. This is also the case for many unranked/casual modes in games that also feature ranked play, like DBFZ, SFV, and Valorant.
Compromises like making a split between ranked and unranked do seem like good ideas to me, though not for every multiplayer game on the market, and this will negatively impact queue times even more than just adding SBMM. But before splitting into distinct SBMM pros and cons, I think this is a fairly good time to mention that I think just about any game will benefit from some skill-based matchmaking, if not ranked, as a means of maximizing enjoyment for everyone involved. That’s why so many multiplayer games do feature skill-based matchmaking, even though they don’t generally tell us about it in the game itself… unless the SBMM comes with some nice shiny badges, ranks, points, player cards, or other virtual trinkets that are meant to either inflate your ego or rub in your face that you suck.
These virtual badges are still enough to get some people to “smurf,” to make new accounts in order to get matched with players of lower skill, something I covered in my article discussing the changes to Valorant’s ranked system for the game’s third Act.
Major tournaments in any esport – or, really, any form of competition arranged into tournaments – which usually happen in the double-elimination format, have a kind of natural skill-based matchmaking to them. In other words, though there is no formalized and numerical skill rating like Elo among the competitive scene of most esports, the best players or teams will often face each other up to dozens of times in any given year, depending on the game in question. Whoever wins the most games, matches, sets, or whatever your favorite esport’s players are competing over, will often end up as the player or team with the highest… let’s call it a “popular Elo.” I think of this as a kind of skill rating formed by a perception among casters, fellow players, fans, all other community members, and even outsiders as to who is the best at a particular game or skill. A popular Elo is a communal idea, a collective set of thoughts that sometimes cohere into power rankings, which can rank players at everything from a bi-monthly local tournament to world championships. “PRs” can set out to rank teams or individuals or both, depending on the type of competition involved.
Elo, if you don’t know the origin of that term, comes to us from Arpad Elo (1903–1992), a Hungarian-born American physics professor. The full history is in that link, but what I walked away with is that the man wanted to use statistics to show patterns over time in terms of players’ skill, tiering them. The Elo system is useful for creating fair matches in online chess lobbies or for finding opponents of relatively equal skill, back when human beings were allowed to see each other in the flesh. I’m not sure of the origins of the term “chess master” but for now I’m choosing to think that Professor Elo wanted to call himself one as an ulterior motive to making this statistical system.
Either way, it just so happens that it feels like you’re gaining or losing points with every match under Elo’s statistical rank tiering, making it a solid system for games that feature one-on-one competition rather than team-on-team, though it can be used like a blueprint for skill ranking in team games as well. Your Elo number represents your mean skills, and I’m using two meanings of the word “mean.” It’s a great ranking system for chess especially, and it works well enough with slight adjustments in other forms of single competition that “Elo” has become a synonym for any skill rating in any game.
Until that above-linked Twitter spat, as someone who’s spent a lifetime filled with games, chess being among the first I learned, I just sort of always took skill-based matchmaking as a good thing. I read about Professor Elo’s system early in life in one of the chess books I studied as a child (I still had lots of friends, don’t worry) and took for granted that the search for “worthy opponents” was a worthy pursuit.
Not everyone sees it that way, of course. I’ll split up our SBMM discussion into distinct pros and cons in just one more paragraph, but I have to address what follows first. A lot of Call of Duty fans and other gamers that don’t think that SBMM belongs in their favorite franchise, and a lot of people who think SBMM should be in every multiplayer game have an attitude that this is “not even a debate, bro.” Clearly, it is a debate; people are debating the issue with valid points on both sides. Not everyone will have the same opinion as you. You may disagree with anyone at any time for any reason. Some reasons are better than others, but respect should be given if its reception is expected. Also, linking to a tweet where any given pro player says “SBMM sux” or “SBMM is gud” and using that as definitive evidence is a poor substitute for forming an opinion on your own. There are plenty of pros and cons to SBMM. There’s nothing necessarily wrong with you mentally if you disagree or agree with pretty much anyone about pretty much anything. This debate has seen a ton of toxicity, so without any further ado, here’s an insult-free and as objective as I can manage assessment of SBMM’s pros and cons.
SBMM Pros and Cons: The Cons
In certain games, especially like Call of Duty, the version of the game played at tournaments varies wildly from the online matchmaking modes. Many perks, weapons, controllers, game modes, etc. have been banned in competitive Call of Duty in the past and will be in the future. As such, the online matchmaking doesn’t reflect what some pros would consider “real skills” or some such gatekeeping language, making a ranking system online somewhat irrelevant for the franchise.
But this doesn’t apply to every game, obviously. In Valorant, for example, the Unrated and Competitive modes are almost exactly the same in terms of basic rules, economy, player health, etc. Pretty much the only gameplay difference is whether or not Overtime is enabled. Now, “Unrated” is still slightly rated, still features some SBMM, but player skill has significantly less influence in the Unrated servers’ matched teams than it does in Competitive. And the Competitive format is the format that gets used in every Valorant tournament.
The biggest con with SBMM that I can see is (rather obviously, in my opinion) queue times. Especially for the players sitting atop Ranked Mountain, your Grandmasters and your Radiants and your Top 500s and your Legends and your Omega Hyper Extra Ruby Super God Demon Warlord Warlocks and whatever term your preferred game slaps on its highest rank… these players can sit through truly abominable queue times. Their waiting periods can be longer than an hour in some extreme cases, such as in the clip of Voyboy below.
Even players who are distinctly in the middle of the pack – your Silvers, Golds, or Platinums, and your 1400-Elo chess players – will find that skill-based matchmaking leads to inevitably higher queue times than if there were no rank restrictions at all. For the impatient gamer, this can be enough of a push to land squarely on the “SBMM sux” side of the debate. And that’s perfectly understandable. No one wants to spend too much time staring blankly at the main menu, rotating the skins you don’t have in the Collection while wistfully sighing, or scrolling through playercards or titles or other baubles attached to the menus, or just watching the queue clock tick and tock away. No one even wants to spend too long in Overwatch’s brilliant (and hopefully eventually the standard for shooters) queued Free-For-All matches, Practice Range access, and Custom Games. SBMM unfortunately makes longer queue times inevitable, and most games fail to make their players’ in-queue experiences anything more than a bore.
Players can also find it rather difficult to learn how good players play the game without ever playing against those good players. Part of the fun of online multiplayer is the possibility of getting matched up with anyone. SBMM allows for players to play against random players, but only within a certain range. Fans floored by the fact that you can play anyone from Timbuktu to Kalamazoo with just an internet connection will naturally be a little disappointed that SBMM culls their matchmaking pool. And when not exposed to good players in one’s own matches, much outside research is often required to get better. I can understand not even bothering to improve in those circumstances. Most esports are brutally hard, and players can deem the grind unworthy of their time at any time for any reason. And without low ranks, there could be no high ranks; like all manifestations of Taoism’s yin and yang, the harmonious opposites of scrub and pro player imply and require each other. Someone has to be the worst if someone else is going to be the best; we can’t know what’s better unless we know what’s worse. Getting stuck in the lower ranks is nothing to be ashamed of, might I add. Playing for fun is just as valid as playing to be the best. And I’m not just saying this because I’m hard stuck Silver in both of the tactical shooters that I play. Totally not. I promise.
Anyway, another con of skill-based matchmaking can be a slight downgrade in terms of entertainment, depending on the game in question or the given player’s disposition. Sometimes, you just want to chill out with an ice-cold drink and not stress too much about the game you’re playing. That can mean booting up a single-player game (at least, one not made by Derek Yu or the Kaizo Mario people) or it can mean hopping into a multiplayer lobby without SBMM. Streamers and other content creators especially benefit from “stomping noobs” without too much effort. It feels good to see your favorite player doing well, and it can be exceedingly stressful to have to play at your best day in and day out lest your rank plummet, taking viewership with it.
Higher Ranks Mean Less Fun
The higher ranks can also bring a slight downgrade to the fun-ness of matches, for lack of a better term, again depending on the game. The genre that comes to mind immediately for me is the battle royale shooter. You’re much more likely to see players running towards each other – bayonets forward, battle cries echoing in their microphones, cover and hiding spots ignored – in the lower ranks. Doses of adrenaline are sure to come more often down there, though the strategies employed are often far from optimal. And in spite of what you’re about to read, I do think the pacing offered by higher-ranked matches are undeniably better, considering that these lower-ranked brawls are rarely planned-out or particularly engaging. It’s mostly a crap-shoot aim duel between people who can’t aim that well (speaking from personal experience here). And these high-octane whiff fests only happen so long as someone’s actually within range of vision. The downtime in lower-ranked BR lobbies… I can only describe as a colossal waste of time for everyone involved. The inverse of this is that you’re much more likely to run into a lobby full of people hiding in attics, in caves, behind trees, etc. for minutes at a time in the higher ranks. And no matter how much you may like your favorite streamer, no one wants to watch them stare at a wooden floor for that long before the circle closes in, multiple times in the same match. This can make streaming even more stressful for those in the virtual spotlight, can make them feel pressured to play sub-optimally in order to create entertainment for their audience. And I doubt that even players with 0 viewers and 0 spectators want to stare for too long at those textures, whether or not the wood grain has a high resolution and whether or not hiding is the optimal way to play.
This high rank loss of enjoyment can take place in other genres as well, such as with some fighting games. The most glaring example of this is Capcom’s now-ancient treasure, a game that came out before fighting game skill-based matchmaking in any form other than who could sit at the setup the longest without having to spend another quarter – Street Fighter III: Third Strike. This game is an absolute joy to play for low- and mid-level players. And many high-level players love the good parts of the game to stick with it all these years later. This is due to its incredible sound design, fluid animations, innovative gameplay still unparalleled, great game-feel, and fun, varied characters. But the game is also infamous for how it can be annoying or outright frustrating to play at higher levels of skill, and skill-based matchmaking algorithms in its more recent re-releases pit these killers together in what can seem like the No Fun Allowed Zone.
This lack of enjoyment at the top is due to multiple factors, most especially character balance. Chun-Li dominates and has dominated Third Strike pretty much since its release.
Chun-Li’s fellow S-tier characters like Makoto, Yun, Akuma, Ken, and Dudley can also be extremely oppressive and prevalent, both online and offline. If you don’t like the way any of these characters are designed, Third Strike is probably not the game for you. I do still think that it’s still likely to grasp any fighting game enthusiast’s attention for at least a few weeks if given a try. The game’s an important piece of fighting game history and still has plenty of players in the present, both on a certain emulation platform and in its Thirtieth Anniversary Collection iteration, both of which incidentally feature rollback netcode.
Despite its depth, good design features, timeless art style, and newly added innovations in multiplayer, Third Strike features mechanics in its wake-up system (and other facets of “gitting gud”) that, if not mastered, will lead to inevitable and often frustrating losses. The game’s “yomi layers” are endless, but not everyone wants to get lost in a vortex of endless learning and adaptation. Even Umehara Daigo, world-famous for performing Evo Moment 37 under immense pressure and being a Street Fighter legend in every title after the original, is known to dislike Third Strike because of these endless mixups surrounding its knockdown game.
Just like a desire for or aversion to SBMM, the games one likes always come down to personal taste. And the higher you climb your games’ ranked ladders, it’s often the case that you’re more likely to run into “degenerate” strategies, characters, and so on. Of course, what “degenerate” means is another thing that varies wildly from gamer to gamer, genre to genre, and game to game.
It’s certainly not just Third Strike that can sacrifice enjoyment at the higher levels and ranks; optimizing any game’s metagame can sometimes lead to boring slogs for top players. During the height of the Overwatch “Goats” meta that gave us Role Queue, for example, lower-ranked players could still whip out their favorite damage-per-second hero and be relatively fine. But getting to the highest ranks with a character like Genji or McCree was… unlikely during that time of three tanks and three supports, to say the least. Without Skill Ratings on the line, it’s also unlikely that Goats would have so thoroughly taken over online Overwatch. And though I didn’t play much at the time, I doubt that Quickplay during the height of the Goats meta often featured six tanks vs. six supports.
So, to sum up this section, SBMM’s cons include longer queue times, higher stress levels, and occasional degenerations of the metagame. Fans of high variance in their multiplayer experience will also generally enjoy less SBMM, whether “less” means a removal of ranks or the total scrapping of any matchmaking algorithm beyond finding enough players to fill a lobby.
SBMM Pros and Cons: The Pros
I’ve already touched on a lot of these, so I’ll keep this brief. Skill-based matchmaking means a higher average quality of the match, and this is especially important in one-on-one games. Neither Williams sister would likely enjoy a singles match against the average tennis fan. Any pro Street Fighter player is unlikely to have fun smurfing in the Bronze ranks. And no chess master would enjoy deploying their vast knowledge of possible permutations after openings and endgame theory while playing on an lichess.org account with an Elo below 1000. Part of becoming the best is never shying away from a worthy opponent. Heated matches with relatively even or even higher-skilled opponents are the tests that can forge or break a professional competitor. Skill differentials are all the clearer when it’s just you and one other person, making SBMM not just a pro to, but something that is nearly essential in one-on-one games like fighters.
Most team games will want to have skill-based matchmaking, as well, just as different sports leagues filter players up and down from intramural leagues to the pros. Again, it’s just generally more exciting, satisfying, and unpredictable to have close games. Excitement, satisfaction, and unpredictability are some of the ingredients to the lifeblood of sports and esports. One’s blood and adrenaline really start pumping fast when the outcome is unclear until the final moments. Champions and runners-up are made in the final moments of close games. A buzzer-beater is less impressive when you’re up by seventy against a team with three wins in the season. And without some skill-based matchmaking, the average player is simultaneously more likely to be matched up with noobs playing with terrible controllers like USB steering wheels, with professional competitors, and with fellow average Joes… but probably not with this highly skilled player actually playing on a steering wheel.
In less words, one of SBMM’s pros is that it can ease the chaos of widely diverse tiers of skill clashing against one another. In making matches based on relative skill, SBMM algorithms attempt to create close, satisfying matches at a higher rate than Lady Luck can do by herself.
When skill-based matchmaking is also rank-based matchmaking, the climb from the lower rungs of the ladder to as high as you can go can be quite satisfying. I talked a bit flippantly about “virtual trinkets” based on rank earlier, but I do think that these titles, points, or other ranking systems allow players to have a tangible sense of accomplishment to go along with the grind of honing their skills. This can prevent the falloff of player interest and keep people logging back on. The flip side to this is that it can feel terrible to get stuck in “Elo Hell,” to keep spinning on the online hamster wheel while going nowhere. Still, I think that ranked climbs are overall points in favor of SBMM. An individual player can journey through various different metagames on their way to their final rank, witnessing various levels of the game, varied brackets of skill, and generally enjoy a diverse player pool without wandering into lobbies where they either destroy opponents with no effort or get destroyed with no understanding of why. The climb allows us to not only hone our individual skills, but to see the fruits of everyone else’s labor as well.
The journey of finding worthy opponents and becoming the best of the best is one often fictionalized, one that can be lived. Even without an online algorithm, the fiercest competitors will find their way to one another, ready to try and prove that they’re the best. And this drive toward finding worthy opponents is a kind of skill-based matchmaking by itself, a wonderful path to walk and to witness. Whether or not you want SBMM in your favorite game, there’s no denying that it prevents new players from getting destroyed too hard and also delivers worthy foes to competitors determined to find them, players determined to show that they’re grinding to be the best that they can be. I think that online games from chess to Valorantto even more casual experiences like Fall Guyswill continue matching players up based on their relative skill levels, and justifiably so.