Fairness vs. latency: what really matters for esports?

With prize pools in the millions, tournament organizers have undertaken the challenging task of creating a reliable online environment for professional players to compete. Will there be a push back from players if studios can’t guarantee a low latency, high fairness experience in these high-stakes online matches?

We have seen many esports competitions move online recently due to the current pandemic, with varying degrees of success. The 24h Le Mans Virtual comes to mind as a mixed bag of high profile, high viewership event that was unfortunately troubled by technical issues. Time will tell what to make of the recently launched PUBG Mobile World League, and the upcoming Call of Duty League Playoffs online competitions. We have also seen many cancellations, the biggest probably being the EVO Championship Series, the biggest fighting game tournament of the year, cancelling their EVO Virtual after many personalities and game studios pulled out due to allegations of sexual misconduct against the championship’s co-founder and president.

Against this backdrop of successes and failures, the state of the network connection between players and game servers has been a point of contention between tournaments organizers and players. What are the issues at play and what can be done to improve the state of these competitions?

Why low latency isn’t king

Low latency is great, but a one-sided difference in latency can create an insurmountable advantage for one of the participant. Consider a 1v1 player scenario where the average latency of the two players is 60ms. It sounds good in theory, 60ms of latency is quite manageable for most online competitive game genres. But taking in consideration only the average latency doesn’t tell the whole story; both players could have 60ms of latency, or one could have 20ms while the other one 100ms. A difference of 80ms between players makes for a match that is “unfair” by most standards, since the player with the lower latency will have a huge advantage in terms of reaction time.

This can create issues like the one experienced during the Call of Duty League back in April, where Crimsix, one of the game most accomplished player, complained that the servers were “unplayable” because his team had to play on “neutral” servers. A neutral server, in this case, means a server that would not give an unfair advantage to one team over the other. Since his team is geographically close to one of the CoD game server location, it was deemed that it would confer them an unfair advantage against the other teams, and they had to play on a different location which gave them a much higher latency. This trade-off is a typical example of the problem of latency vs. fairness, and without an understanding of how these two interact, studios and tournament organisations are bound to receive complaints from their most valuable and vocal ambassadors, professional players.

No alt text provided for this image

The fairness score

So how do we measure fairness from a network perspective? The simplest way to define the fairness of a match is to take the difference in round-trip time between 2 players in a 1v1 scenario. In the case of team-based matchmaking, a team is considered a single entity and so the average latency of the group is compared to the other group(s).

If the difference in latency is low, it means the match is fair since there is only a small gap between each player’s latency. If it is high, the match becomes unfair from a network perspective as the gap in latency between players means one of them has a sizable advantage. To take a specific example, if a player has 100ms of latency and the other player 20ms, the fairness factor is quite high at 80ms and as such the match is skewed towards the player with a lower latency. As such, a low fairness score is a sign a fair match up.

Through the looking-glass

Let’s take a look behind the scenes with a concrete example from an actual game. Through our work at Edgegap, we had the privilege to work on a case study in collaboration with a leading game studio to take a look into what can be done to create a better online experience for their players. Providing a lower latency is a great step forward, but it cannot be the only variable taken into account. Compromises must be made to make sure each match is fair from both sides, enhancing the overall experience.

Let’s take a look at a specific match that was played between a player in New York and another in Ivory Coast. Although it is an “edge case” scenario, it gives insight into what can be done to improve fairness when latency is in play:

No alt text provided for this image

We can see that Player 1 had a low latency of 39ms at first, and a higher latency of 93ms after moving the game server location. However, for Player 2 his latency went from 176 to 106, and as such the fairness factor went from a large 137 to a meager 13, creating a much fairer environment. In that sense, making sure the game server is situated about halfway between both players makes the most sense. That is why having a large global infrastructure footprint is key to provide both a lower latency, and a more fair environment for all players, as each available location has the ability to improve the quality of every match.

Taking fairness into account

There is no one-size-fits all in video games, and what might be great for one type of genre makes no sense in another one. Every game, every match and every player is unique, and requires a decision based on the unique parameters of the match. This is why policy-based decision-making is the best tool available to create a fair experience.

A policy-based system is quite simple in theory. Take all the parameters that helps create a fair experience for the players involved (level, ranking, latency, jitter, fairness, etc.) and weight them one against another so that you determine which one has priority. Is latency more important than the level of the players? Is fairness more important than the player ranking? It all depends on the game, and at first the game designers are the best to answer these important questions. In the long term however, machine learning will be able to provide a better insight into how to create a fairer, and more fun experience for all players involved.

Fairness is a fundamental principle in every competitive titles, yet it is rarely taken into account from the network perspective in a game’s matchmaker. Beyond simply comparing latency between players, a holistic view of the match is needed to provide the best conditions for the opposing players. When taking into account the meteoric rise of esports betting under the pandemic, fairness becomes especially critical to provide a healthy competitive scene.

What gets measured gets managed

Starting a match in a state of fairness does not mean it will stay fair for the whole duration. That’s why it’s important to track the network for each player during a match and react when a player’s online experience degrades. Different actions can now be taken: pausing the match, sending a QoS improvement request, or even increasing lag artificially for the low latency player to insure a fair competitive environment.

Once data has been accumulated for thousands of matches, we can see patterns emerge and use machine learning to make better decisions in the matchmaker. These can be about specific sites not being able to guarantee a good experience throughout a match, players coming from different ISPs that creates a degrading experience once a match is launched, etc. The decision making can then be adjusted to provide a better gameplay experience with the whole duration of the match in mind.

No alt text provided for this image

Conclusion

There are many things to take into account when creating an esport title, and networking has to be a priority for game studios that aspire to release such a title. The online competitive scene is taking off under the restrictions brought by the current pandemic, but many issues remain to be solved. Studios will have to find solutions to reduce latency, improve fairness and have visibility over the network issues their players are experiencing for professional and casual players so that they do not move to the next best game that manages to get things right.

The team at Edgegap is focused on providing solutions for game studios to lower latency, improve fairness and increase the reach of game titles through the help of edge computing and machine learning. They have developed solutions specifically for esports titles such as a policy-based decision maker solution, player network monitoring and control technologies, and twork visibility tools to ensure a fun and fair online experience for players. Get in touch at info@edgegap.com