New Evidence Suggests Super Meat Boy Speedrunners Cheated

Illustration for article titled New Evidence Suggests Super Meat Boy Speedrunners Cheated

Super Meat Boy’s challenging platforming and hidden secrets have turned it into a popular game for speedrunning. A new method for detecting cheaters now suggests that some famous runs might not be what they appear to be.


There are two kinds of Super Meat Boy speedruns: real-time attacks, which show the entire game without interruption, and segmented runs that break the game into small chunks then aim to optimize time for each section. Occasionally, sneaky runners try to pass off segmented runs as real-time attacks with the help of video editing software. Members of the Super Meat Boy speedrunning community have developed a new way of detecting these spliced runs and are convinced that they’ve found cheaters from as far back as five years ago.

The method is outlined in a document called “Bandage Girl Autosave Animation Splice Detection” and involved watching a small loading icon and counting the number of animation frames to determine if the game is running in real time. When players clear levels, a small image of Meat Boy’s love Bandage Girl appears in the lower left side of the screen. Depending on the game’s frame rate, the animation will play for a certain amount of frames. As an example: at 60 frames per second, Bandage Girl holds her arms up for 20 frames and brings them down for 20 additional frames. When the game is first loaded, the animation always starts from where it finished last time it shows up. Looking at this, moderators can count the frames and look for differences in the animation to detected edited speedruns.

Among the runs believed to be fake is a 18:39 Any% run by ExoSDA submitted to the leaderboards in 2012. It is the second highest-viewed Super Meat Boy speedrun of all time. Examination of the run showed that eight frames of animation are missing from a portion of the run after ExoSDA enters the game’s fifth world. This tipped moderators off to discrepancies leading them to believe the run was fake.

“This anomaly doesn’t happen in live single-segment attempts,” the community moderators said in their document. “Instead, we see 20-frame arms down animations at the start of each chapter, indicating the use of saves to record each chapter individually.”

The frame-counting method was also used to identify a recent run, tied for second place in the Any% category, that appeared to be spliced and edited as well. Speedrunners are notorious for breaking down games and meticulously checking data—which is bad news for anyone who decides to cheat, whether now or half a decade ago.

UPDATE (December 1, 10:30am):

In a Pastebin statement, ExoSDA says that faulty recording and hardware were the culprit.


“Re-starting the recording mid-run wasn’t an option as that lead to the game stopping as well,” he said. “As soon as I made a mistake that I felt was reset-worthy I stopped the recording and deleted the savegame to start from the beginning. This is, of course, the critical part where you either believe me or don’t and I fully understand the people that choose the latter option.”

Former Senior Writer and Critic at Kotaku.



How simple would it be for programmers to add in some kind of authentication tool to their games? I can see it becoming common given the increasing speedrunning community.