Developers know that gamers are happiest when they’re playing, and the longer they have to wait to get into multiplayer matches, the faster you’re likely to lose them. Players don’t enjoy waiting around in lobbies; they want to get in on the action straight away. That’s why optimizing matchmaking is key.
Systems such as Google and Unity’s Open Match are accelerating the ability to build matchmakers with custom logic, giving developers access to tools that negate the need to build a system from the ground up. Without that labor-intensive work, developers are freed up to focus on optimization, tweaking inputs, and selection logic to best suit their games.
The more matchmaking evolves beyond skills-based pairings and further into complex decision trees, the more AI and machine learning are coming to the rescue to keep it quick.
Fast matchmaking algorithms and tools can improve the quality of a gamer’s experience, but when you’re optimizing for quick matching and working to marry latencies, regions, and servers, how can you also ensure your matches are full? This is just as important to a game’s success as a speedy start, so it’s imperative to ensure matchmaking is both fast and full.
The Importance of Ensuring a Good Match
The beauty of gaming is that players can engage with anyone, anywhere, but they need to be well-matched in order to have a fun and fair gaming experience. When done successfully, matchmaking plays an integral part in lowering player dissatisfaction and churn.
**Lowering Player Churn **
New players consistently matched with experienced competitors are likely to leave, as are highly skilled veterans who aren’t challenged. If the balance is right, players will continue to be engaged.
Furthermore, healthy matchmaking can lead to connections that, in time, may develop into clans, guilds, and long-term friendships. This increases player enjoyment, engagement, and retention.
Simply put, highly engaged player pools lead to greater game success and increased ROI for publishers.
Leaps in AI and Machine Learning Are Improving Matchmaking
Great matchmaking requires fine-tuning a significant number of parameters, including optimization for players’ specific game and play style. This can take substantial time and resources.
The burden on the developer is now being lightened by tools that provide flexible off-the-shelf matchmaking solutions, enabling them to focus on defining the value of the metrics and automatically optimizing them over time.
Both Open Match and Amazon’s GameLift FlexMatch open-source matchmaking frameworks use AI and machine learning (ML) to increase the speed at which a matchmaking engine can be developed and the algorithm computes.
Microsoft’s TrueMatch goes even further into reinforcement learning, adjusting its algorithm in real-time to find the best combination of matchmaking allowances for the current situation.
The benefits are clear. During his GDC 2020 presentation, Dr. Josh Menke showed how TrueMatch was able to drop wait times for Halo 5 six-player free-for-all matches by 72% (10 minutes) during lower population hours by tuning the matchmaking rules in real-time.
Sony’s also getting in on the action. Last year it published a patent that uses neural networks and complex ML capabilities to learn from gameplay and make predictive skills-matching decisions. These advancements can only improve optimization.
What Happens to Matchmaking in Latency-Sensitive Games?
When a game is latency-sensitive, matching players by region is often prioritized. For many games, this means an immediate reduction in the available pool of players.
When this concept is taken a step further and latencies are matched, the pool becomes smaller still. This continues throughout the optimization decision flow, as players are grouped by skill, mode, region, server, and latency.
Let’s take a pool of 10,000 gamers. First, we split the pool by available players. If an average game lasts, say, seven minutes, that takes the pool down to roughly 1,429 players available per minute.
The pool is then split by gameplay mode. For this example, let’s say there are five options, and so we’d divide that gamer pool by five, taking it down to 286. Next, the players are matched by skill. If we assume that a third are close enough in skill level to provide a challenging but fun game, this takes our pool down to approximately 95 players.
As you now see, if we keep breaking this down by region, latency, etc., our suitable group of players shrinks down to 32 and then 11 players a minute. This highlights the need to expand reach wherever possible in order to ensure a suitable pool of players.
Keeping Your Matches Full Requires Expanding Your Reach
Both the speed of decision-making and the network’s ability to deliver filled games must be optimized to provide the best possible matchmaking experience.
As we’ve seen, successful matchmaking requires huge numbers of players. An optimal network supports this by expanding regions. Heavily reduced and optimized latency immediately allows for larger pools thanks to a wider geographic reach. With or without skills matchmaking, and no matter the game style, this leads to a more competitive game.
Inter-region connectivity also creates larger communities and positively affects the gaming world by increasing diversity and inclusivity. For example, improved network solutions have allowed gamers in an increasing number of countries and regions to join the competitive gaming community, including Jordan, India, and the Middle East.
Ensuring Fast and Full Matchmaking
In the end, a variety of elements contribute to a solid player pool, but none of that matters without speed. While developers look to optimize matchmaking for skills, gamers are crying out for a focus on connection — none more so than the vocal Call of Duty (CoD) community.
When Activision launched CoD Warzone in March 2020, there was an uproar over lag issues that ruined the gaming experience for many long-term fans. This is the last thing publishers want, but those issues are minimized with a network optimized for gaming.
Competing interests aside, the importance of both decision-making speed and a network’s ability to deliver filled matches is unlikely to lessen, and one without the other still does not provide fast and full matchmaking.
Better in-Game Matchmaking
Full matchmaking pools are critical to the success of and engagement with a game. Dutch game designer Joost van Dongen highlighted this in a breakdown showing that once limiting criteria such as latency, region, and skills are accounted for, a 1,000-player pool would yield availability of 0.15 players per minute, resulting in a 30-minute wait to fill a match.
Simply put, bigger player pools allow for quicker matching of criteria and, therefore, faster starts.
The strongest games have both.
Prioritizing the quality of experience (QoE) and quality of service (QoS), computer networking company Subspace aids in matchmaking by expanding the reach of player pools through its high-speed overlay network. Optimized for online gaming, this sits on top of the existing internet, finding and securing the most reliable, highest quality, and the fastest path between any two points in the world.
By finding the quickest routes for data to travel, latency, jitter, and pack loss are reduced, leading to gains of more than 100 milliseconds for some players. This plays an important role in increasing player enjoyment, as Subspace data shows that for every ten milliseconds latency is improved, players will stay 4% longer.
Low-ping radiuses are also expanded thanks to Subspace’s network, increasing player pool sizes. This improves the matchmaking experience for multiplayer games, ensuring a happy gaming community and, in turn, improving revenue generation opportunities.