We’re seeing a lot of feedback for dynamic queues, so let’s take a moment to talk about what’s going right and wrong behind the scenes, and how we’re feeling about the current state of the queue. We saw what you saw. Scarra had some massive queue times--even worse than the old Ranked system. Pro teams were outfoxing teams made up of high skill solos and we’re seeing some awful matches at the Challenger level. We also ran into matchmaking and implementation issues. Beyond this, dynamic queue raises philosophical questions about what it means to be a ranked player in League of Legends, so we’re going to talk about that as well.
First, we failed to communicate that new champ select came with a brand new matchmaking system built to handle position selection as part of matchmaking. We tested on PBE and with simulated matchmaking for weeks, but there’s nothing quite like going live. We knew matchmaking would require tuning, which is why we launched this preliminary test during preseason and not the day the ranked season starts.
Implementing this system came with a number of matchmaking tuning issues and hardware failures, some of which have already been fixed. We’ll quickly run through them and how they impacted you:
* The matchmaking feature ensuring premade teams were facing other premade teams was configured too loosely, and allowed premade teams to match against teams of solos too often.
* Our new matching making system didn’t account for hardware failures and left players stranded in matchmaking with giant queue times.
* The matchmaking feature that tried to reduce queue times for high-skill players was too aggressive, and matched pro-level players against Diamond players rather than other Master and Challenger players.
We are taking steps to address these issues. Specifically, we’ve adjusted how aggressively we match top-tier players and have started enforcing the premade matching rules more strictly. We are collecting data and will continue to adjust matchmaking every day until the season starts.
Now let’s talk about the elephant in the room: the philosophy behind ranked. Current competitive players use their ranked standing as a more pure expression of mechanical skill, game knowledge, and the ability to cooperate with solo or duo’d players. For those players, the ranked queue was doing its job by emphasizing individual playmaking over one’s ability to ‘make do’ with variable teammates. The expectation was that everyone else climbing the ranked ladder also dealt with the same variables with no exceptions. Dynamic queue changes this considerably, so why rock the boat?
Last year, we investigated why players didn’t play ranked, and a huge number said they didn’t want to abandon their friends to do it. In other words, there was a large subset of players who wanted to play competitively with friends on a ranked ladder, but lacked a great avenue to do so. We constantly talk about how League is a team game, and we want our core, competitive ladder to reflect that. The ranked queue might be synonymous with the term ‘solo queue’, but even a solo/duo queue rewards a player’s ability to work as a part or as the leader of a team. The goal was to hone in on that team-like atmosphere without hitting the tremendous commitment issues of ranked 5’s.
This lead us to the introduction of dynamic queues. We designed dynamic queues so the highest level of play relies on teamwork in addition to mechanics and clever application of encyclopedic knowledge. In other words, our vision for ranked is that it should be a more accurate reflection of teamwork, leadership, knowledge and skill.
But with all that laid out, the question is, “_Is that the right vision for ranked League of Legends?_”
We hear you on the impact dynamic queues have on the measurement of solo skill. We’re currently doing everything we can to make dynamic queues a great experience, and we [always had back up plans](https://dd.reddit.com/r/leagueoflegends/comments/3rht20/new_ranked_solo_queue_megathread/cwop63f?context=3) for the 2016 ranked season. For now, we want to share some early data on how things are going, where we’re at with matchmaking, and how players have responded so far in surveys. We understand this is a philosophical issue in the long-run, but we want to bring some data into the conversation so everyone is working with some facts.
Keep in mind that the issues we’re addressing influenced this data, and we plan on revisiting them after a weekend’s worth can be evaluated:
* **On the topic of premades vs. solo queuers**: We expected about 95% of premade 4s/5s to match up against other premades of that size, but we fell short at around 83.5%. We still expect this number to reach 95% after we tune matchmaking, and we’ll keep a close eye on this issue.
* **In terms of match quality in dynamic queues**: We measure match quality by the difference in average skill between two opposing teams. Currently we’re seeing 92.8% of games are about equal to last season’s solo/duo queue (and overall matchmaking is better than Team Builder). Because of the issues we just fixed, we don’t have good data on the skill differentials between premade 4s/5s, but they’re on our watch list of things to track and fix, if necessary.
* **When we polled a random set of players across all of League of Legends the specific question**: “I like that you are allowed to queue with any number of people to play ranked”, about 55% of players agreed, 20% were neutral, and 25% disagreed. We’re not saying this to sway opinions either way--it tells us the system is both polarizing, and broadly desired. We’ll check in on this statistic again next week.
* **For queue time (queue start to champ select)**, as measured during non-peak hours, we’re averaging three minute queues across all players, and about eight and a half minutes for Master+ players. This is also something we’ll be looking to improve over the weekend.
All that said, please keep testing dynamic queues through the weekend, and keep sharing your feedback. We’re going to be scouring the boards, and fixing issues as they come up. **On Tuesday, 1/19**, we’ll make a decision about the state of dynamic queues and share a plan for the 2016 ranked season launch.
EDIT ::: Corrected the survey results. They were actually sent to a random sample of the entire playerbase, not just players who played Dynamic Queues. Also, when you broke down the analysis to players who played Dynamic Queues or those who did not, the results were not significantly different.