We want to take some time to openly discuss the challenges the project is facing and how we are approaching them. Our goal is to give you a clearer picture of what’s happening behind the scenes — both the problems we’ve identified and the steps we’re taking to address them.
This collection of notes focuses on three critical areas for the health of multiplayer gameplay: the leavers problem, the ongoing fight against cheating, and the matchmaking system. Each of these issues has a direct impact on how you play, and each requires a slightly different strategy to resolve.
Part I: The Leavers Problem
Broken Arrow is a team-based game, and when a team is missing players, it is at a serious disadvantage. Player absence generally comes down to two categories:
1. Technical factors
Sometimes players fail to establish a connection to the server at the very start of the match, which means they are not registered by the server and cannot join. There are many possible reasons for this, and we are continuously working on identifying and fixing them. In other cases, a player may lose connection during the match and the reconnect button may not appear — this issue will be addressed in patch 1.0.10. Until then, restarting the game can be used as a temporary workaround to force a refresh.
2. Players leaving voluntarily
Lack of tenacity
For players that leave voluntarily and abandon their team, we hope the leavers penalty will help combat the bad habits of those players. While the surrender option allows the team as a whole to decide if they would like to leave the battle, in a team game, the situation can always change — if things are tough on one side of the map, your allies may have the upper hand elsewhere. The game is designed so that early mistakes don’t decide everything, and each phase offers opportunities to come back.
Avalanche effect
On top of that, one player leaving often triggers an “avalanche effect,” when others follow. But a disconnect doesn’t always mean someone is gone for good — it could just be a temporary issue, and they may return. There’s a 5-minute window to reconnect, and in the meantime, playing 4v5 is still viable since the smaller team receives extra income as compensation.
Unbalanced teams
Totally one sided battles are not really fun to fight and there is not much to learn when the skill difference is too big.
-> We have recently tweaked the matchmaker to make the acceptable Elo difference between teams progress slower over time and we capped the maximum discrepancy. The result is an increased probability to find a more balanced match at the cost of more wait time if teams in your Elo bracket are rare.
Note: we are aware that as a result, top teams of the leaderboard were having difficulties to find matches even when waiting for a very long time so we have recently tweaked this value again.
Development plan
Vote for surrender & leavers penalty
These features will be introduced with patch 1.0.10. We are approaching penalties with caution – we don’t want to punish players in cases where leaving is not their fault. But we need to address a problem that is currently hurting the quality of the game for our players.
So after patch 1.0.10 a new button “Surrender” and a new icon next to your name in the player list menu will be added.
Clicking this button will pop up a window for you and your teammates and if the vote is accepted the match will end in a defeat for your team and you will be able to see the scores and history of kills like at the end of any match.
Players who leave their match without using the surrender function and don’t reconnect will be considered as deserters.
Deserters will be prevented from joining a new match during a period of time increasing every time they desert but reducing over time if they don’t desert.
These durations can be modified easily so we will observe how players react to the penalty and tweak them over time.
Developing diagnostic tools
At launch, we lacked some auxiliary tools needed to fully measure the problems players encounter. Since then, we have been steadily working on building these tools and strengthening our technical foundation.
-> This is not something that players immediately perceive so while we’re working on that people have the impression that we’re doing nothing. But in the long run this will help us to react faster and better to all the issues we might face.
Part II: Fighting Cheating
A fair environment is essential for any online game. Cheating undermines trust, so we are tackling it on multiple fronts.
Shifting authority from client to server
A fundamental part of our approach is to gradually move more and more of the game’s logic from the client side to the server side. This process is not simple, but it is essential: the more authority the server has, the fewer things remain on the client side that can be modified or exploited. With every step in this direction, we reduce the opportunities for cheats to exist, and over time, low-level manipulations will become practically impossible.
Such a solution is difficult to imagine in fast-paced FPS games, where instant input recognition and hit registration are critical. However, in the RTS genre, this approach is much more realistic, and we intend to fully take advantage of this opportunity.
Combining internal and external solutions
In addition to developing our own server-based protections, we are also working on integrating a third-party anti-cheat system. This combination allows us to cover a wider range of potential exploits. External tools will complement our internal solutions, strengthening the overall system and ensuring that cheating becomes increasingly difficult and risky.
Continuous monitoring and log analysis
Another pillar of our anti-cheat strategy is the systematic collection and analysis of logs from every multiplayer match. Even if a cheater is not detected and banned instantly, the information gathered during the match helps us identify suspicious activity and take action afterward. This creates a long-term safety net that reduces the chance of persistent abuse going unnoticed.
Anti-cheat as an ongoing process
It is important to acknowledge that anti-cheat is not something that can ever be considered “finished.” It is a continuous, evolving process where we adapt to new challenges and methods of cheating as they emerge. Despite this, we are already seeing results: more than 8,000 cheaters have been banned to date, and each improvement makes the game environment more balanced and fair.
Fixing bugs that can look like cheats
Multiple bugs of synchronization can look like cheats either when they happen or when they are resynchronized by the server. For instance if a unit is located in two different places for two opponents they might both think that the other is cheating. When the server detects such desynchronization it corrects the position of the unit which is rapidly moved across the map.
Fixing most of these bugs in patch 1.0.10 will help to reduce the impression of some players that every match contains a cheater.
Part III: Matchmaking
Matchmaking is at the core of multiplayer. The goal is to find opponents quickly, of similar skill, and with stable connections.
The algorithm behind matchmaking
Our matchmaking system is based on the ELO algorithm.
It’s a simple yet well-established method of measuring players skill. And like any other rating system its objective is to pair opponents of similar skill levels to make the match more entertaining than if the teams were matched randomly.
However, the system has certain limitations that we are working to address.
Challenges with ELO rating
The first issue is the inflation : ELO points are exchanged like a currency between players depending if they win or lose.
But since every player starts with points, each new player entering the system is the equivalent of generating more currency which results in inflation like for real money.
This effect is partly balanced out by players who leave the game for good, with their rating points gradually fading from the system.
This can generate periods of inflation or deflation depending on the number of players joining or leaving the game.
Another limitation is the absence of ELO degradation. This means that players who take a long break retain the same rating, even though their performance after a long pause may differ.
Also, the rating only tracks the history of victory and defeats of the players but not their individual performance during the matches, which means that the score of a player is actually the average of his score and the scores of the other players they tend to play with.
These two issues can lead to less accurate matchmaking results. But the problems would disappear if players simply kept playing without taking rating into consideration as the discrepancies would be corrected after several matches.
Which leads to our final and trickiest problem because it’s not technical but psychological :
Players hate to see their rating reduced and treat it as if it were a mountain of gold coins that must be hoarded and slept on like a dragon.
So they try many methods to avoid losing their precious points, dodge difficult matches, refuse to play with even slightly weaker players, even sometimes stop playing at all rather than risking a defeat that would reduce their rating, which totally defeats the initial goal of the rating system which is to play interesting matches.
This is Goodhart law : the measurement tool becomes the objective and players prefer to have virtual points than to have real fun.
Beyond skill: latency and party balancing
Skill is only one part of matchmaking. To ensure a fair and smooth experience, our system also factors in additional parameters:
• Average party ELO: when players join as a group, the system considers their combined rating rather than treating each player separately.
• Average party latency: matchmaking also measures the connection quality of both parties and tries to place them on the most suitable server. This is especially important for maintaining stable gameplay across regions.
Balancing these elements ensures that players are not only evenly matched in skill but also have the best possible technical conditions for their games.
A balance between speed and fairness
Every matchmaking system is ultimately a compromise between speed and accuracy. Our priority is to keep players in the action while still maintaining fairness in matches. We are proud to share that 90% of the audience currently finds a match in less than one minute, which is an unprecedented result for the wargames genre.
Game server selection and connectivity
Our long-term solution is to implement a game server selection feature, allowing players to choose servers that provide the best possible connection (estimated development time: over 3 months). In the meantime, we are continuing to improve our server-picking algorithm so that matches are hosted on servers that minimize latency for all players involved.
Looking ahead
We will continue to refine and improve matchmaking to address the challenges mentioned above. Our goal is to make every match not only fast to find but also as fair and competitive as possible.
Post-release Retrospective
After the release, our team went into what we often call “firefighter mode” – tackling the most urgent problems as quickly as possible to stabilize the game. This phase left little room for deep analysis, but it was essential to make sure that everyone could keep playing.
Once the immediate fires were under control, we shifted our focus to longer-term improvements and deeper investigations. This work may appear slower from the outside, but it is vital for stability and for building a strong foundation that the game can grow on.
At launch, our team was very small. Thanks to the support of our community, we have since expanded and brought in new expertise. Onboarding new members takes time – sometimes up to two months before their work reaches the live game – but this investment ensures that we can deliver stronger, more sustainable improvements in the long run.
Closing
We hope this extended note gives you a better understanding of how we’re tackling three of the biggest challenges in Broken Arrow: leavers, cheating, and matchmaking. None of these problems have quick or final fixes, but with each step we are making steady progress.
We’ll continue sharing updates like this to keep you informed, not just about our victories but also about the difficulties we’re working through. We will also be sharing a roadmap that outlines our vision for the project’s growth and development in the year ahead.
Knowing that so many of you care deeply about the game makes all the difference, and we can’t thank you enough for being part of this journey with us.