It’s become quite clear in the last few updates and blog posts that Dota’s matchmaking system is receiving heavy attention from Valve.

Better late than never - this is great news for all Dota players and in this article, we’ll try to read between the lines to see in what direction the devs are trying to take the game and the matchmaking system. 

The first big and quite surprising change was that Valve got rid of Solo queue in its original form. This was unexpected because Solo was the standard “tryhard” mode for Dota. That said, Dota is a team game, so pushing players towards playing in pre-made parties to fully enjoy the game makes a lot of sense.

That said, the main reason Solo was more popular than Party was that most players don’t have friends of a similar skill level that are available to play when they are. This might currently be a problem because the new matchmaking update makes Solo a bit less enjoyable (especially on the higher MMR brackets) while finding a dedicated same-skill-level party is hard.

Maybe the next logical step would be to integrate a friend-finder in the client (finding people close to your skill level, play-time preference, role preference, language, etc.) so that you can form long-term 5-stacks much more easily.

Bad Behavior Score

“Users that reach this low level of behavior in the game are too big of a tax on the rest of the community and are not wanted.” (Source)

Although it seems unrelated, this change goes hand in hand with the Pary/Solo change. One of the major reasons that playing with a party is more enjoyable is because you don’t encounter the toxicity of the Solo queue. So, to make the overall experience better for everyone it makes sense to incentivize players to behave more heavily. They are currently employing the stick method with the low behavior score ban waves, but a carrot could also be implemented:

For example, maintaining a behavior score above 9k can give you periodic rewards like a free Rylai Wheel spin every month or something along those lines.

Something like disabling chat/voice for people below 5k behavior score could also be a good idea.

Cores and Supports

The partial separation of the Core and Support MMR is also linked to another trend we’ve been seeing in Dota – empowering supports: cheaper support consumables, more shared net worth and levels, tomes, etc. All of this is aimed at making the support role more enjoyable (to make it possible to own and visibly decide games from a support position).

It’s not a secret that pos. 5 is by far the least desirable position in pubs and that this is a big problem in the current MM system in terms of queue times. Because of this, we definitely expect this trend to continue.

We wouldn’t be surprised if the balance team slowly tries to move the game into a direction where we have what feels like two 4^th positions rather than a 4^th and a 5^th (don’t forget that years ago the supports were, in reality, two 5^th positions, so we are already moving in this direction).

Smurfs and Boosted Accounts

Smurfs and Boosted Accounts have been a problem in many competitive games since matchmaking exists. They are problematic for the matchmaking system for sure, but it might be argued that it’s not the game or the matchmaking system’s fault that they exist, but rather human psychology.

• Smurfs: People like playing on smurfs for the same reason that people play competitive games in the first place. It feels amazing to own. When you are playing on a smurf, you are just owning a lot more because you are better than everyone in the lobby. The problem is that this is a positive experience for the smurf and a negative for everyone else involved. Behind the anonymity of the internet, however, this behavior is unlikely to stop. Bans are unlikely to be effective because Smurfs are throwaway accounts by definition.
• Boosted Accounts: Boosted accounts work on a different psychological note. Dota is a team game. First, it’s very natural to place the blame on someone else for losing when you’ve put your heart and soul in a game rather than to focus on your own mistakes. Second, it’s a fact that playing with better, competent teammates is more fun. Because of this, even if it’s not allowed, using a Dota 2 boost service isn’t likely to go away anytime soon.

Valve is working on an algorithm to detect and More accounts that are exploiting the system, but it seems that they are acknowledging this behavior is impossible to stop and are focusing their main efforts on faster calibration. The biggest emphasis is on getting people to their “real” skill bracket as fast as possible (i.e. calibration based on performance).

In theory, in the future, a machine-learning algorithm could learn to detect player skills in a single game thanks to the huge amount of Data it has (e.g. mouse and camera movements, optimal and suboptimal spell usage, etc.). Valve is likely excited to strive towards something like this because it has applications outside of the Dota 2 matchmaking system – it could be used for matchmaking in any other game