Stuck in the Middle with You: How to Solve Esports’ Endless Matchmaking Problem

This is the first in a series of posts on Veebit’s work in the gaming industry. The second is here.

What is the single worst problem you have to swallow when playing Overwatch, League of Legends, World of Warcraft – or any of today’s biggest team-based esports titles?

Matchmaking.

Unless you launch a round of play with teammates you know, you are stuck with the teammates you get; an algorithm digests your game stats, your history, and a few other variables and then flings you into the next round with the four or five people it believes are best suited for battle beside you.

And off you go, with nothing more than your skills and a posse of random strangers you hope will help beat your opponents – and no guarantee whatsoever of any actual teamwork. You might win. Or you might get clumsy offense, shoddy defense, rogue teammates chasing their own score instead of a team victory, and toxic behavior – obnoxious, profane, blame-flinging personal abuse.

It doesn’t have to be this way – if you approach team-building as a science and not merely a game mechanic. More on that below.

The Developers’ Bane

Gamers are a demanding audience. We gripe about server lag, control response, character balance and a host of other problems we feel are undermining our ability to level up and win.

But matchmaking – the way millions of teams are formed every day – is the number one pain point for competitive gamers, says Bo Moon, head of the Innovation Management Group at Samsung Electronics, and a member of the Veebit advisory board.

“This type of team-format gaming suffers from matchmaking services that are not great, and that’s a huge issue when one match may take more than 30 minutes to complete.” says Moon. “A successful team has good interaction and chemistry, makes the time investment worthwhile, and likely leads to the player queuing up to play again. Bad team chemistry that ends in a loss is often a good reason to quit for the night.“

Game developers are struggling mightily to cure these problems and build better teams – but without addressing the actual human chemistry that researchers say drives team success.

Try, Fail, Try Again

For instance: Soon after Overwatch launched, players on message boards routinely slammed game maker Blizzard Entertainment’s solo-queue matchmaking system. The complaints were common: you dumped me in with the wrong teammates – too over-skilled, too under-skilled, too random and uncollaborative, or too toxic to play with at all.

The problem has kept Blizzard jumping through hoop after hoop, in ways that seem to address the symptoms of mismatching, but not its root cause: human dynamics.

In 2016, Blizzard implemented a match-rating feature. Players wound up using it mostly to criticize the matchmaking system itself.

Blizzard also launched a feature to let players flag “bad” teammates with “Avoid this Player,” then scrapped it a couple years ago when players began misusing it by tagging people who could beat them.

Meanwhile, a similar battle goes on at Riot Games, which has been wrestling for years with a litany of complaints in gamer forums about League of Legends – a veritable downpour of dissatisfaction – best summed up in yet another outcry in 2018 that “MATCHMAKING IS BROKEN!!!!!!!!!!!!!!!!!!!!!!”

Bad for Team Play, Bad for Business

Gamers who suffer bad experiences in free-to-play games like League of Legends are less likely to convert to playing (and paying) regularly, and poor matchmaking makes some avoid team games altogether:

“I play battle-royale games because I get less upset when I lose,” said a Wharton MBA student we interviewed recently. ”Each game takes less time than a team-based game, and because it’s more individual, there are no team members to get upset about.”

Time will tell whether big developers can satisfy their increasingly grumpy audiences by simply tweaking the algorithm. Until then, the problem of bad team chemistry continues to occupy developers in private and chew them up publicly in gamer communities like Reddit.

Meanwhile, third-party companies are working to fix team-building from another angle – the gamers themselves.

“Gamers want to be treated as individuals,” says industry veteran Kim Rom, an advisor for Dreamteam. Dreamteam is building a platform that lets gamers find teammates, build teams and manage their pro-career aspirations.

“Seventy percent of startups fail because of differences between the founders,” Rom says. “Amplify that [dynamic] by a factor of 700 million for a team-based 5-on-5 game where you’re communicating with other players on headsets, you can’t see anyone’s body language, you can only see the mistakes they make that cost you sweat and tears, and it’s easy to get pissed.”

Trialing by Fire

ESL Gaming Network lets players create teams online, then enter them into thousands of tournaments it presents every year.

Many players hunt for good teammates on Steam and other communities and vet their Twitch, Facebook and Twitter feeds for signs of toxicity.

Yet gamers seeking teammates still must conduct the tedious business of “trialing” – playing dozens of games alongside the candidates to learn how well they perform under pressure and mesh with the rest of the team.

Sometimes, the challenges to smooth team play are less obvious than toxic behavior or basic clashes between strong-minded teammates who don’t play collaboratively:

One gamer interviewed for this study by researchers at Clemson University and the New Jersey Institute of Technology complained: “Our mid-laner has a problem where he kind of – it’s somewhat sad to see because he is a nice guy, but if he is getting his lane camped and he’s under pressure and he’s starting to crack, he won’t communicate at all. And we kind of have to coax him out of that shell.’”

The inherent challenge of balancing team dynamics rolls on – as varied and persistent and damaging to gamer satisfaction as the pounding surf – largely unsolved by matchmaking algorithms and friend-finding platforms.

Researchers affirm conclusively that strong teamwork – which stems substantially from how well teammates’ social and psychological traits complement each other – is critical to team success.

A two-year study of 180 teams at Google boiled down successful teamwork to 5 shared traits.

A research team from the United States Sports Academy interviewed dozens of esports athletes and concluded in a paper published in July:

“It becomes clear that to get the most of the online gaming experience, social skills, including the instrumental skills to effectively cooperate with others and the emotional capacity to support team members online and offline, are necessary. For both team leaders and members, the process of social interaction requires them to communicate in an environment which is made of competition but can only be survived via cooperation [emphasis added].”

But these reports are qualitative – and reached their conclusions only by laboriously interviewing team members to learn why they succeeded.

The Science of Teamwork

Psychometrics – the science at Veebit’s core – offers a faster, more precise and quantitative route to clarity around team dynamics. And this can lead to repeatable success in designing team-building mechanics.

By measuring gamers’ built-in attitudes and social behaviors and correlating them with copious individual and team performance data generated by esports platforms, Veebit can build models to predict team-chemistry factors that are more likely to succeed.

And by modeling winning teams, we can help esports platform developers empower their players to build winning teams on their own terms – by teaming up with other players who are most likely to help them win.

We quantify teamwork by asking (and providing the technology to answer) the right questions:
What psychological traits are shared by the strongest League of Legends junglers?
Which personality types do their best Overwatch teamwork in clutch situations?
Which combinations of social attitudes and personal quirks lead to the tightest teamwork and highest kill ratios in Counter-Strike: Global Offensive?

And finally, how can Blizzard, Riot and the other MOBA and MMORPG developers powering the esports boom match players for smoother team play, better games and – ultimately – far fewer angry public complaints?

We’re working on that. Right now.

This is the first in a series of posts on Veebit’s work in the gaming industry. The second is here.

 

 

Mack Reed is Veebit’s Head of Product and oversees the information architecture, use case strategy and user-experience design for the company’s products.