by Hollie Wegman ・ Office Hacks

Office Hack #36 — Joyent’s Mario Kart Hack

 

Envoy is all about making things easier and more fun in the office. In that spirit, we are proud to bring you our new Envoy Office Hacks podcast series. Every week, we deliver the coolest, most ingenious, and just plain fun fixes people have invented to improve efficiency and productivity in their workplace.

Today’s Envoy Office Hack was born out of love for Mario Kart 64. This version of the very popular Nintendo game released in North America around 1997. Who could have predicted that almost two decades later, one talented fan would find a way to hack it to make it cough up its secrets?

Dave Pacheco works at Joyent.

Joyent, recently acquired by Samsung, is a software company that specializes in cloud infrastructure and big data. A number of years ago, a small group of employees at Joyent would spend a lot of off-hours battling each other at Mario Kart 64.

There were times when emotions ran high. Trash talk ensued. Arguments. Sore fingers. Things would get really heated when a lead racer was knocked out seconds before crossing the finish line to claim his victory.

Most of the time, the banter was innocent enough, but eventually, suspicions grew.

What if the game manipulates outcomes? What if certain characters are doomed from the start? What if there are specific conditions that could trigger one weapon over another?

Inquiring minds wanted to know. So Dave built a web application, which he soon dubbed “Kartlytics” that would generate data about each race that was played.

Each game would be recorded so still frames of the video could be analyzed. Kartlytics would facilitate the browsing of records and statistics from each recorded race.

“I modeled it on a website for baseball called Baseball Reference. …One of the things about Baseball Reference is that everything on the website is clickable. If you click on any player, you go to their team. You see their season stats. Then you can see the batting stats for someone else on that team or similar. For Kartlytics, you can also click almost anything. You can click on one of the tracks and see all the races on that track. You can click any of those races and see all the people who raced it. …I wanted to build something that was interactive in that way … that you could spend a lot of time playing with and picking apart, even though it’s only Mario Kart.” Dave Pacheco, Joyent

And pick apart, he did. After some 200 races, Dave could start to identify triggers, advantages, and disadvantages.

“Some people would say, “Well, you got really lucky in that race. You got really good weapons.” Could we quantify that? That’s the kind of question that we could go answer now.” Dave Pacheco, Joyent

But more data only led to more questions.

“…as you might expect, having more data to answer these questions about the banter only leads to more banter because then you say things like, “Well, there’s a selection bias in the data,” for example. I am the one that recorded most of the races, so I show up in most of the statistics, and that distorts a lot of the statistics.” Dave Pacheco, Joyent

So, Kartlytics couldn’t provide conclusive data, just yet, but it did have one surprising outcome. It served as an accessible (and fun) case study for one of Joyent’s products, a cloud storage system called Manta.

“One of the things we talk about with Manta is doing log analysis with something like this. What I found was if I told people you could do log analysis with Manta, they’d be like, “Okay, yeah, that’s kind of cool. I can sort of see that.” If I presented Kartlytics to people, they would say, “Oh! I could do log analysis with that,” and they’d be really excited about it. It was a really good way of getting people interested in it. Then, they could see what else they could do that was kind of like that, which was fun.” Dave Pacheco, Joyent

So what has Dave learned about the best races and most favored characters? Nothing definitive yet, but he has been able to determine:

First-place racers stand a:

  • 33% chance of obtaining a banana peel (mediocre weapon)
  • 28% chance of obtaining a single green shell (mediocre weapon)

Fourth-place racers stand a:

  • 25% chance of receiving a triple mushroom (3 speed boosts)
  • 18% chance of obtaining three red shells (superior weapon)
  • 18% chance of receiving a star (protection + speed)

Conclusion: People who are doing poorly get better weapons.

 

How to hack it

Want to build your own Kartlytics, or take a look under the hood of this one? Dave posted all the magic over on GitHub.

Here’s a link to the Kartlytics website: http://kartlytics.com/

And finally, a blog post on Joyent talking about this project.

Listen & Subscribe

You can listen to the full story in the Envoy Office Hacks podcast. Please subscribe to the podcast in iTunes — just click here and hit ‘subscribe’.