If you ever want to toss a discourse grenade into a room and watch the sparks fly, bring up skill-based matchmaking. How much or little developers should factor player skill into who you're permitted to play against in multiplayer games is an evergreen debate of our hobby, but it's not everyday that an expert on the topic chimes in.
Responding to an opinion piece in PLAY Magazine about SBMM in FPSes (as shared by GamesRadar) on Twitter, former Bungie multiplayer lead Max Hoberman argues modern multiplayer games do a "disservice" to players by only serving them even matchups, citing his work on Halo 3 as how to do matchmaking right.
"Ranked filtered opponents based on level. This was for when you wanted a competitive match," said Hoberman in a tweet thread. "But even then, I intentionally allowed variability in the range of levels we matched you with."
Hoberman, who was the multiplayer and online lead on Halo 2 and Halo 3 before founding Certain Affinity in 2006, laid out the three types of matches that Halo's matchmaker was designed to serve players: an "easier" match you were likely to win, a "harder" one where you were outmatched, and an even matchup.
"The failure of modern skill-based matchmaking, imho, is that it's designed to maximize these perfect match scenarios and minimize the others," he said. "When it's working, a majority of games become super tight, super stressful. That's not fun for most players. Where's the variability?"
Hoberman's perspective on SBMM cuts against the common wisdom in competitive games that creating fair matches, and only fair matches, produces the best results. The most popular competitive games in the world—Counter-Strike 2, Dota 2, League of Legends, Apex Legends, Rainbow Six Siege—all use ELO or MMR-based ranking systems to always attempt to form balanced lobbies. This often means that the majority of players can quickly find matches, but the top 10% of players have to wait a lot longer.
"Segregating high skill players from the population at large, forcing long wait times on them, is a form of discrimination," he said. "Designers should strive to find a way that players of all skill levels can have fun together."
One way, Hoberman says, is to have designed skill variance in unranked and ranked modes. Most games already do this in unranked scenarios: casual playlists often prioritize finding matches quickly over finding players of similar skill, or even establish a separate MMR for unranked versus ranked. What almost all multiplayer games have in common is that casual MMR values are hidden.
Hoberman's argument for variance is intriguing, but it also strikes me as incomplete. To many players, a huge part of the appeal of competitive games is to definitively prove their skill against the rest of the community. There's a real sense of accomplishment in reaching a high rank in CS2 or League of Legends, and part of what makes that accomplishment feel genuine is knowing the game tries to give you a fair challenge at every step of the way. I reckon if you asked players deeply embedded in those communities if they'd prefer to be deliberately given unfair matchups part of the time, they'd probably say no. The promise of unfair matches in your favor to balance things out does sound fun, but also pandering in a way that could spoil the glory.
On the flipside, Hoberman's vision for SBMM is exactly what the most vocal corners of the Call of Duty community constantly ask for.
"The system that I designed for Ranked playlists ensured a healthy mix. Sure, it sucks watching your favorite team get their butt kicked," he said. "But it comes full circle when they're the ones kicking butt. Throw in tight, evenly matched games every so often, and that's a ton of fun."
Activision generally doesn't comment on how its SBMM works, but we know it's there in CoD, and we know that many players think its SBMM is too sweaty, or report huge fluctuations in the perceived skill of their opponents from match to match, resulting in a frustrating whiplash between getting stomped and doing the stomping. This gives the impression that there is some skill variance baked into CoD's matchmaker, but that it's not totally random, and generally favors a fair match.
Whether or not these complaints are backed up by data, they've made SBMM a hot-button issue for the series in recent years, with a vocal majority arguing against fair matchups in favor of a Hoberman-style of "fair" where everyone has an equal chance to have a bad or good time. Except, there are disproportionate winners and losers in that scenario, too: the top 1% skilled players have a 99% chance of finding an easy match, while the bottom 1% are likely to have a bad time every time.
I'm not sure how Hoberman's system accounts for these players. If the highest-skill players really do get an even match "every so often," does that mean sometimes their matchmaking times are 30 seconds and other times 5-10 minutes? Because that can be how long it takes to find a good matchup when you're among the best.
It's no surprise that developers opt to obfuscate the particulars of their matchmaking systems. There's no pleasing everybody, especially when some want fairness by way of skill ratings and others want fairness via an amorphous illusion of randomness. Hoberman's full Twitter thread, which includes more thoughts on casual matchmaking, is worth a read.