For eight years, a player who goes by BarryMode held the world record for beating the original Spelunky as fast as possible. Because most people, speedrunners included, moved onto the much more popular Spelunky HD in 2012, BarryMode's incredible run of two minutes and 30 seconds held firm for nearly a decade.
The problem, it turns out, was that BarryMode had actually cheated, and the evidence of the deception had been hiding in plain sight the entire time, undetected.
"I do apologize for that deception," BarryMode, who subsequently took down his run from YouTube, told Waypoint recently. "It was bad character and I can admit that."
BarryMode has since removed their speedrun from YouTube, but Waypoint archived it and have included a clip as proof:
"This is definitely the most major cheating situation I've heard of in Spelunky speedrunning," said Spelunky designer Derek Yu. "Actually, it's the only one I'm really aware of, off the top of my head! As a creator, it definitely means a lot to me that people spend the time learning how to speedrun Spelunky. They deserve proper credit for being the first to do something in the game. Those accomplishments are part of the game's history!"
There had been suspicions about BarryMode's run for years, because BarryMode wasn't a known speedrunner, let alone one in the dedicated Spelunky speedrunning community, and other speedrunners who'd spent thousands of hours playing the game hadn't come close. The closest was a speedrunner called Groomp, who's managed to hit two minutes and 40 seconds. A 10-second difference doesn't sound like a lot, but speedrunning is often measured in milliseconds.
As I outlined in another story about a speedrunning community dealing with a similar issue, it's hard to prove cheating. Some speedrunning happens live, some speedrunning records are established in recordings taken in the privacy of a speedrunner’s home with a copy of OBS, and no video game explicitly red flags cheating in a way that is obvious to a viewer. To unearth malfeasance, the community has to dig through video and code seeking proof of something that may not exist, while dealing with the fallout of accusing a person of lying.
Recently, Waypoint was given early access to a video by YouTube creator and Spelunky fan XanaGear, which outlined how the community caught onto BarryMode and exposed them:
In a Spelunky community Discord, someone pointed out a weird missing graphic in BarryMode's run, a single tile in the corner of the screen while the game moved between stages. Glitches happen in games, especially in speedruns, but given there were already questions about how BarryMode pulled off the run, it seemed like a thread worth pulling on.
XanaGear looked at the code and confirmed the weirdness.
"We couldn't explain why but we knew there was no way that block should ever be removed as it was hardcoded and not generated as levels are," said XanaGear.
By "generated," XanaGear means Spelunky, a game where players frequently die and start over, is always generating new, semi-randomized levels from a series of LEGO-like pieces from the developers. Many things in Spelunky are random, but many things are always the same. In this scenario, the block is part of Spelunky that's meant to be present all the time.
Even more curious, that missing block is featured in all but one transition scene. But on its own, the block is nothing more than a data point. A missing block isn't cheating on its own, but it does point in the direction of the game functioning differently than it's supposed to.
One of the things that bothered Spelunky fans like XanaGear prior to all this, as they puzzled over BarryMode's run, was how often BarryMode chose the most optimal route to the exit. "Couldn't that just mean they're good at the game?" you might ask. The problem is that even the most skilled Spelunky players cannot predict the fastest path, and instead use their knowledge of how the levels are built to make educated assumptions about what would be fastest. What BarryMode kept doing, over and over, was choosing the fastest path correctly.
One theory: BarryMode was replaying the same levels over and over again, allowing them to memorize the layout, and the uploaded video quietly spliced together these runs to seem like they're happening back-to-back. This is a common method of speedrun cheating, though to be clear, cheating does not happen all that often.
What another Spelunky fan later discovered was that a Spelunky mod that allows multiple runs of the same level, preventing the randomness that’s key to Spelunky’s challenge, placed an object in the game code in the same position seen in BarryMode's video. The missing tile.
Furthermore, the mod doesn't kick in until Spelunky's second stage, explaining why BarryMode's run only sports the missing tile in the transition to the game's second stage.
Before the video went live, I contacted BarryMode, who then published a video publicly discussing the accusations, the series of questions that Waypoint had sent BarryMode prior to this article's publication. The video was removed before Waypoint could watch it, but BarryMode also answered several emails, in which they quickly confessed to the deception.
"In 2013, I realized that being a YouTuber was a profession and many people were making a living as gamers," said BarryMode. "I had just gotten married and was having children so games were not part of my life at all. Still, I made an attempt to start a channel because I saw the potential of it helping to support my family. I posted some let's plays. Eventually, I had the idea to speedrun on my channel."
2013 is the same year BarryMode published the infamous Spelunky "world record."
The theory, BarryMode said, was that videos like this could grow the YouTube channel. It did not, and these days, BarryMode's channel specializes in conservative meme videos where BarryMode writes songs about whatever the right is upset about that day. In one video posted February 15 titled “Cannot Cancel Us,” BarryMode autotuned and put music to Ben Shaprio supporting actor Gina Carano when she was fired from The Mandalorian for implying that being a Republican today is akin to being Jewish during the Holocaust:
BarryMode admitted the video is a series of spliced runs, though claims to have used a different mod to achieve it. Strangely, BarryMode refused to fully embrace the notion that they'd cheated to achieve the record, claiming "the run is real in every way, except that I eliminated the 'starting over' factor in order." That's a big part of why speedrunning is hard, because you have to do everything in a single run and not screw it up before it's all over!
"The numbers are all accurate as if it was a real run and there's no cheating during gameplay otherwise," said BarryMode. "Any time I went the wrong way, was too slow, or died, I would just reset the hearts, bombs, ropes, money, and time to maintain the continuity between levels and try again. I bypassed luck so that I didn't waste valuable time away from my family and my life."
That said, BarryMode has apologized both in private and in public, and folks like XanaGear have accepted the apology with sincerity. It's also why BarryMode removed the original run from their YouTube channel.
"After I apologized and had some conversations with different Spelunky speedrunners on Reddit," said BarryMode, "I gathered that they really wanted to see the video removed so I removed it. I don’t want continued hard feelings to be there. Even though it seems like something stupid I did a long time ago in my past, to them it’s fresh so I wanted to be respectful of that and honor their wishes."
The deception also reveals a quiet tragedy: for the past eight years, BarryMode's run has been considered the best in the world. All along, however, someone else, a speedrunner named ExplodingCabbage and one of the earliest Spelunky speedrunners, was that person.
ExplodingCabbage's legitimate run of two minutes and 53 seconds was uploaded in July 2010, long before BarryMode.
"It's a nice discovery," ExplodingCabbage told me in a Twitter DM. "The [BarryMode run] was always kind of underwhelming to me; nothing in the run looked particularly impressive and it just seemed like he got very lucky level layouts. By contrast my run was full of tense moments and close shaves with death. I never suspected at the time that the boring nature of the 2:30 run indicated cheating, though; I recognise that a big part of getting a fast Spelunky run is just having good luck with the level generation, so I figured the boring nature of the 2:30 was just how I'd expect a very fast non-TAS [tool-assisted speedrun] Spelunky run to look."
"I'm glad it got resolved, especially since ExplodingCabbage's run is what made me realize that Spelunky had any speedrunning potential," said Yu. "When I first released [Spelunky] Classic, I thought nobody would speedrun it because of the randomness. It was very, very cool for me to read [their] thoughts about the run, and the strategies involved. Definitely changed my understanding of speedrunning and I got a big kick out of it at the time."
ExplodingCabbage's run has since been legitimately beaten by the aforementioned Groomp.