Mario Kart 8 has consumed most of my entertainment hours since it came out a couple of months ago. Its online play is great. When you queue, the game randomly gives you three (of thirty-two possible) tracks to pick from, or you can select random if none are to your liking. Social scientist that I am, I saw an obvious data collecting opportunity. So the last few weeks, I have painstakingly charted every single choice I have observed. This allowed me to create a rough ranking system of all the tracks in the game. Which track do people like the most? The least? Check below:
The numbers reflect the percentage of the time I observed players picking any given track, not the track the game randomly selected from those ballots. For example, over the many, many times Sunshine Airport randomly popped up in the queue, players selected it 48.3% of the time. The tiers simply cut the data into a top bucket of four and four other buckets of seven.
There are a number of important caveats to the image, so please read what follows before boldly declaring that Bone-Dry Dunes is the worst thing Nintendo has ever created.
- I don’t claim that this is the be-all, end-all to Mario Kart track popularity. Rather, without any other metrics to rank the courses, I think that this is a useful first-cut at the question.
- While I gathered a lot of data to do this, I am only one man. The number of potential picks ranges from 113 for Water Park to 258 for Bone Dry Dunes. We should expect such randomness from the queue selection system. However, it also means that some of these percentages are secure than others. I plan on continuing to collect data over time.
- Be careful about making pairwise comparisons. Based on what I have, it is reasonable to conclude that players prefer GBA Mario Circuit (41.8%) to Electrodome (27.7%), but it is not reasonable to conclude that players prefer Electrodome to Mario Kart Stadium (27.5%).
- With people duo queuing, I included both votes. I can see why people might think this should only count as one, but the choice from a duo queue (in theory) reflects the preferences of two people. So I count it twice. It would be very difficult to count them as one vote anyway; I would have to keep tabs on who is submitting at the same time, which difficult when I am trying to count so many things at once.
- I collected the data as I rose from 2000 to 3100. So if you believe that preferences are different for this group than a different one, you are not looking at the image you may wish to see.
- I did not count my votes. We want a measure of what people like the most, not what I like the most.
- I excluded “forced” votes that occur if players take more than the allotted time to make a selection. These votes are pure noise anyway.
- An active vote for random counts as a vote against everything else. For example, suppose the choices were Yoshi Valley, Royal Racewway, and Music Park. Three players select Yoshi Valley and one picks random. Then Yoshi has received three of four votes and the other two tracks have received none of the four. In other words, the “random” doesn’t magically disappear from the denominator in the data tabulation.
- I only played worldwide games.
- These were all races. No battles.
And now for a little bit of analysis:
- I did some fancy statistical tests to see if a variety of track qualities (length, difficulty, newness) determines player preferences. All of the results were null. So whatever is driving these votes is highly idiosyncratic.
- The new Rainbow Road was very disappointing. It was the last track I played when I went through the game for the first time. I was very excited until all I found was boring turn after boring turn.
- Some might also describe the original N64 Rainbow Road as boring turn after turn, but it seems that Nintendo made a smart decision to turn the course into a straight-shot and not a five lap race.
- I question Nintendo’s wisdom in putting Music Park, Grumble Volcano, Sherbet Land, and Dry Dry Desert in the game. What’s the point of having classic tracks if no one wants to play them?
- To be fair, perhaps players actually wanted to see these tracks and just failed in the execution. But that still doesn’t explain why you would put Grumble Volcano back in the game. Its main course feature is that lava randomly shoots up and kills you for no good reason. I understand Mario Kart is full of randomness, but let that come from interactive item blocks and not from the computer.
- I feel really bad for whoever designed Bone-Dry Dunes.
See you in the queues.
Update: With eight new tracks coming out this week, I decided to update the data one last time. Here’s where we are today:
I’ll probably run the data once again after the new tracks have been out for a couple months.