Some Numbers from Bumble Bash, the First National Killer Queen Arcade Tournament
This past weekend saw the first national tournament for the arcade game Killer Queen, called Bumble Bash. As an eSport, in relative terms, KQ is a very young game, with the competitive scene only being two years old at this point, and a very small game, Bumble Bash having just a little over 150 people spread across 33 teams. But it is a game that I love with a community that I love even more, and as such I’m very interested the trends and character of the game nationally.
I’m also someone who has been involved with eSports and local multiplayer video games for a few years now and I’m very interested in how communities of players, especially competitive ones, grow and evolve.
To that end I wanted to see how some numbers from Bumble Bash regarding individual scenes looked when put together on a graph. At the top of this post you’ll see two graphs. The first one plots the age of individual scenes (which in the KQ community are always sorted by their home city) against the median placement of all the teams that were sent by that scene. In parentheses are the number of teams sent to Bumble Bash by that scene. The second graph lays that information out, plotting the number of teams sent by a scene, labelled ‘presence’, against the scene’s median placement.
DISCLAIMER: I think all these numbers are accurate, but I compiled everything right after the tournament and so feel free to correct anything that I might have gotten wrong. Also, the numbers I’m looking at are pretty small, so keep in mind there’s a chance that the patterns are just a result of the sample size. Finally, I am a thoroughly amateur data analyst, so anyone who’s much smarter about this stuff please tell me where I’ve been foolish!
This was a fun little project and what struck me most of all after putting it together was that there was some evidence of something I had already assumed: Age, size, and strength in a scene all correlate.
In claiming this I am making two assumptions that it’s worth being up front about: that the median placement of all the teams sent to Bumble Bash is a good indicator of the general strength of a scene, and that the number of teams sent to Bumble Bash by individual scenes is a good indicator of their size. I think both of these things are true, but there is probably additional data that could be brought to bear and also alternative ways of interpreting the data available.
As you can see in the first graph there is a correlation between the age of a scene and that scene’s median placement in the tournament. Chicago, Portland, and New York City are the three oldest scenes and two out of those three are also at the higher end of the median placement distribution. The youngest scenes, Austin and Minneapolis, had on the other hand the two of the lowest median placements.
Next to the city acronyms in the first graph I’ve also put the number of teams that the corresponding scene sent in parenthesis. You can see that this number also grows as you move along both axes.
In the second graph I’ve presented these data more clearly. You can see what you might have assumed from the first graph: the greater presence a scene had at Bumble Bash (which I believe likely correlates with the size of that scene), the higher the median placement of its teams (which I believe correlate to the general strength of that scene).
Laying this all out there were two things that I found interesting:
First, Charlotte and Kansas City are over-performing relative to their age and size respectively. In other words, the cities seem like they’re actually stronger than you would assume based on how long they’ve been active and how many teams they sent to the national tournament.
Kansas City is slightly outperforming its age and its size. Looking just at its age the median placement of its teams is slightly better than you might expect. And while it’s in the middle of the age distribution it is a little lower that the mid-point on the presence distribution. Even though it was one of the sixth largest scenes in terms of presence it managed to achieve the third highest median placement, behind Chicago and New York.
Then you have Charlotte which is in the middle in terms of placement, but is a little lower than the center of the distribution of teams present and even lower in the age distribution. I interpret this as meaning that the city is outperforming for its size and definitely outperforming for its age.
Along with those observations, the second thing I was struck by is that while Portland was in the middle of the distribution on both presence and median placement, which you would expect to correlate, it was second oldest scene. I interpret this to mean that relative to its size PDX as a scene is doing fine strength-wise, but that it is not as strong or as large as you would expect from a scene of its age.
(EDIT: An earlier version of this piece made a mistake in the placement of one of the San Francisco teams. When that was corrected it turned out that SF was right on trend in terms of age and their median placement. I’ve edited this piece throughout, though most of the results remain the same. Apologies to the San Francisco team and take this as a warning about how small the sample size is we’re working with here!)
I was very curious about the case of Portland so I decided to investigate further and made a third graph to make things a little clearer for myself:
In this graph I have put the distribution of possible placements at the bottom. I have then listed, in chronological order of founding from top to bottom the scenes that attended Bumble Bash. Along the lines extending from the city names I’ve placed dots that mark the placement of the individual teams that represented that scene. Where teams had the same placement I just put dots on top of each other.
What we can see in this graph is that the teams from most scenes tend to clump in the placement distribution. Both teams from Columbus are on the same half of the distribution. The majority of New York and Chicago’s teams are pretty evenly spread across the upper half of the placement spots. Two of Kansas City’s teams are in the upper half as well. The third KC team, which was formed at the last minute is at the very bottom. However, due to the circumstances I think this still shows that KC is strong for its age.
Looking at this individual team data also turns what would be a worrisome outcome for Portland into much better news. It has teams in the upper half of the distribution, but also in the lower half. PDX has a team that placed second in the tournament (and I can tell you showed enough talent that it definitely could have taken first place) and 9th place team, with its other teams placed behind the half way mark.
My interpretation is that Portland has a lot of talent in their respective scenes, that talent wasn’t evenly distributed across the teams present at Bumble Bash.
Interestingly, this was also the case for San Francisco. SF has a huge gulf between its two best teams and its two other teams in terms of placement.
For San Francisco they have about half their talent in the upper end of the distribution. That leaves the opportunity for the talent in the two strongest teams to be widely spread among the weaker teams, potentially boosting their performance over time as the lower performing players learn, in person, from the higher performing ones. Second, remember that San Francisco is in the middle of the age distribution. This means that they have two teams that are strongly out-performing relative to expectation in regards to their time playing the game. For comparison, look at Chicago and New York, both of which are old scenes that have their teams pretty evenly distributed across the upper half of the range of placements.
Portland has a similar story.
I’m of the opinion that as you get higher in the placement distribution of a tournament, it takes more and more skill to reach each successive rung of the ladder. So the gap in skill between the teams in 1st place and 5th place is actually much greater than the gap in skill between the teams in 22nd and 30th place.
What this means for Portland is that, potentially, the gap in terms of strength between their team at the very top of the placement distribution (called Excuses) and their team in 18th place (called Stinger Scouts) is much bigger than the the gap between SF’s 9th place team (called Golden Hate Warriors) and their 22nd placing teams.
What produced this dynamic? My understanding, which is based a lot on conversations with people in both of these scenes and with players familiar with their inner workings, is the large gap between upper and lower placement that you see in both these scenes is a result of their having their most veteran players concentrated in fewer teams.
Let me put it this way, if Portland had only sent their highest placing team they would be totally in line in terms of the correlation between skill and age that we see in other teams (though, obviously, this would mean that the relationship between their age and size would be even further off). Again, if we just look at the higher placing San Francisco teams the most interesting thing we see is that SF flips from under-performing relative to their age to significantly over-performing relative to how long the scene has been around!
Far be it for me to suggest to any scene about what’s best for them. Every scene is different and has different dynamics. But looking at the relationship between the median placement of the different scenes and the actual placement of individual teams the advice that I’m going to be taking to my home scene (New York) is to spend the year before the next Bumble Bash making sure that talent is well distributed so as to skill up the entire scene.
So, how important is all this? It’s hard to say!
Looking just at these data its hard to make a judgement about the value of having skill evenly distributed across a scene’s teams, and how that might effect future performance (though my guess is that it’s a good thing to do). One thing that would be fun would be to look at similar data from Killer Queen X, a tournament that took place about a year before Bumble Bash that featured Chicago, Portland, and New York teams to see if the talent distribution there was any predictor of future performance at the top end of teams.
Furthermore, I want to point out that this correlation between the age, size, and strength of scenes in Killer Queen is just that: a correlation. It’s not clear to me in which direction the causation flows. My intuition is that the longer a scene is around the larger number of people it reaches. At a certain point a scene hopefully begins a virtuous cycle. A large scene has a better chance of finding star players and forming good team synergies that increase the level of play in their community. A high level of play and people who are capable of productive team dynamics attract more players in turn. And so on.
This is not to diminish the enormous amount of work done by members of the community to maintain and foster this virtuous cycle. At the end of the day Killer Queen is really powered by an extraordinary level of commitment and cooperation within and between scenes. The data laid out in this piece is one look at the character and dynamics at play within the competitive side of these communities, but it doesn’t do a good job of explaining the love that people have Killer Queen or the incredible caliber of people involved in all the scenes.
In closing, I’ll just say that though this snapshot in graphs seems to confirm my intuition that age, size, and strength all correlate in Killer Queen at the national level, I’m very interested to see if this correlation holds as the years pass. In future Bumble Bashes will it still be true that the older scenes hold the top spots? If the older scenes continue to grow will you eventually see a more even distribution across the entire range of placements? As the younger scenes grow which ones will look like New York and Chicago and which will follow the model of San Francisco and Portland? Will Kansas city continue to outperform relative to its age? At this point that would mean that a KC team will have a strong shot at the championship in the next couple of years!
As a scholar interested in the growth of multiplayer game communities I am excited to find out the answers to these questions. As a fan and player of Killer Queen I am simply looking forward to living through the answers with the incredible friends that I’ve made over the last two years playing this stupid bee game.
Note: If you’re interested in Killer Queen as a game then I suggest this video for an introduction https://www.youtube.com/watch?v=DyAwobPCCW0 and this one to start learning how to play https://www.youtube.com/watch?v=nRzCpzZ_CXg
