Skip to content
Behind the screens

How to cheat at Super Mario Maker and get away with it for years

Creator says he "was just at the right place at the right time" to abuse TAS techniques.

Kyle Orland
The Frankenstein's monster of a GamePad hack responsible for the creation of Trimming the Herbs. Credit: Ahoyo
The Frankenstein's monster of a GamePad hack responsible for the creation of Trimming the Herbs.
A cardboard box provides the inelegant case for Ahoyo's Arduino-and-breadboard TAS solution.
Close-up of the Arduino connections necessary to control the Wii U GamePad.
Ribbon cables feed into holes cut through the back of the gamepad to connect directly to the mainboard.

Last month, the Super Mario Maker community was rocked by the shocking admission that the game's last uncleared level—an ultra-hard reflex test named "Trimming the Herbs" (TTH)—had been secretly created and uploaded using the assistance of automated, tool-assisted speedrun (TAS) techniques back in 2017. That admission didn't stop Super Mario Maker streamer Sanyx from finally pulling off a confirmed human-powered clear of the level last Friday, just days before Nintendo's final shutdown of the Wii U's online servers Sunday would have made that an impossibility.

But while "Trimming the Herbs" itself was solved in the nick of time, the mystery of the level's creation remained at least partially unsolved. Before TTH creator Ahoyo admitted to his TAS exploit last month, the player community at large didn't think it was even possible to precisely automate such pre-recorded inputs on the Wii U.

The first confirmed clear of Trimming the Herbs by a human.

Now, speaking to Ars, Ahoyo has finally explained the console hacking that went into his clandestine TAS so many years ago and opened up about the physical and psychological motivations for the level's creation. He also discussed the remorse he feels over what ended up being a years-long fraud on the community, which is still struggling with frame-perfect input timing issues that seem inherent to the Wii U hardware.

"I see discussing it as sort of a reputation damage control," Ahoyo told Ars. "I saw value in the 'ruckus' that TTH would cause, in that it would bring in outside eyes to look at SMM with uncertainty and excitement... but it was a betrayal of my competitive values."

Hardware hacking

Ahoyo recalled first hearing of the possibility of a Super Mario Maker TAS in "late 2015 or early 2016" when a viewer of his Twitch streams messaged him about a Wii U TAS project they had been tinkering with. In response to a follow-up in June 2016, the viewer sent a video "showing a controller attached to a Raspberry Pi and showed how it was controlling Mario on the screen," Ahoyo told Ars.

While the viewer told Ahoyo they had since "abandoned" their efforts to get reliable TAS recording on the Wii U, Ahoyo said he "showed the video of their hardware to my friend, and he told me he thought it would be easy to replicate, so after an Amazon order, he built it." This kind of level of automation was "not something I [had] thought about before," Ahoyo said. "In the video demo I was sent about the TAS, it appeared to me like it was working fine. I thought it was a waste to abandon the project, so I pursued it spontaneously. I was just at the right place at the right time."

The inner workings of the Wii U GamePad mainboard.
The inner workings of the Wii U GamePad mainboard. Credit: Koushirowolf / Reddit

As you can see from the pictures atop this article, setting up a Wii U to play back pre-recorded inputs isn't exactly a plug-and-play affair. Ahoyo's setup involves an Arduino with multiple external connections to a breadboard, each one representing a button on the Wii U GamePad. Those outputs are first fed into a pair of 10-pin converter boards, then directly onto the Wii U gamepad mainboard via ribbon cables (fed through holes cut into the existing rear mounting holes on the GamePad). A separate set of colored physical buttons connects to the breadboard to control the playback of the TAS scripts through the Arduino, Ahoyo said.

"It turns out my hunch was correct; this was a bespoke, hand-crafted solution specifically for Super Mario Maker," tool-assisted speedrun expert and TASbot keeper Allan "dwangoAC" Cecil told Ars after seeing the setup. The solution Ahoyo and his friend rigged up "isn't general purpose in any way that likely wouldn't have worked for anything else," Cecil added.

Getting in sync

Even with the somewhat elaborate hardware hack (which Ahoyo said took his friend a couple of days to set up), recording and playing back a series of Wii U inputs reliably wasn't a simple process. TAS recordings for older systems usually make heavy use of frame-perfect emulation tools that can precisely sync inputs to actual hardware, down to a window of just a few milliseconds. The emulators available for the Wii U when Ahoyo worked on his TAS in 2017 didn't have the software features or the accuracy needed for this kind of work (which is still difficult using the best Wii U emulators available today).

Instead, Ahoyo used a guess-and-check TAS creation method, hand-coding inputs into specially formatted text files that laid out which buttons to press and for how many frames. Ahoyo said his process involved writing up a few inputs and then "observ[ing] playback on the screen repeatedly."

Desync issues can happen on any TAS run, not just on the Wii U.

But just figuring out the correct inputs for a tough in-level sequence wasn't always enough. That's because Ahoyo's hacked-together TAS solution would frequently send some inputs to the hardware slightly earlier or later than expected, causing the playback to "desync" from the intended deterministic result.

To work around the problem, Ahoyo said he would record the playback via Open Broadcaster Software and watch frame by frame "to get a better idea of the state of the inputs... the idea is to find the inputs that are played back [on the video] most often and build from that." These hardware desync issues help explain why TTH is so short. "Because of the desync issues, longer playback durations can take exponentially longer to actually appear on the screen," Ahoyo said. "Most playbacks will result in a failure, so you have to start over often."

But Ahoyo said he also developed some level design workarounds that allowed for TAS techniques to be used on longer levels. "To deal with longer playbacks, you can build level sequences in pieces by making use of doors and pipes and then combine the parts after," he said. "They likely won't work perfectly after combining different parts, but they will still speed up the process."

Trimming herbs with damaged hands

Well before Ahoyo stumbled onto the first known TAS method for the Wii U, he had already built up a reputation as what he calls "the bomb guy" in the Super Mario Maker community. By late 2016, Ahoyo said he had already put "a year or more" into the creation of "bombs5," a sort of culmination of all the intricate Bob-omb manipulation tricks he had spent years developing.

Around November 2016, though, Ahoyo said he started developing intense pain in his hand that prevented him from manually testing the level any further. Frequent repetition of the complex motions needed for intricate techniques like the "meow dive" had slowly worn down the tendons in his thumb and wrist, he said.

A detailed breakdown of just what makes "Trimming the Herbs" so difficult.

"I've played way too many games over the years, but Super Mario Maker was the first one I played with a d-pad and the first one that gave me this kind of issue," he said. "I think it's due to me pressing the d-pad hard, whereas a game that uses a joystick doesn't involve any 'pressing' for that thumb at all."

Even after a "multi-month" break from Super Mario Maker and regular hot water treatments on his wrist, Ahoyo said that "it was pretty clear that... the tendons weren't healing properly." Despite a long history in the Super Mario Maker community, Ahoyo said, he "began to accept the reality that I had to move on," a decision that "felt like weight off the shoulders."

With bombs5 testing set aside, Ahoyo said he finally had time to set up the Super Mario Maker level design contest that he'd always wanted to organize. That contest, called PogChamp, would also end up being the world premiere of Trimming the Herbs.

Ahoyo and other contest judges discuss Trimming the Herbs in 2017.

Ahoyo said TTH's core trick— which involves dropping a Bob-omb on some piranha plants and then re-grabbing it in mid-air—"was something I had in mind long before I got the TAS." But as the PogChamp contest got underway, Ahoyo said he "took the trick, dressed it up with skull coaster/climbing/ending sections, and made TTH in probably a day." While Ahoyo said his hand issues prevented him from manually testing and uploading TTH for that contest, his working TAS tools let him craft and test a level that he felt would let "the bomb guy... go out with a bang."

Unearned clout and praise

Ahoyo said his intention was to upload TTH and then "announce TAS existed for SMM" after the level had been "played, dissected, and scrutinized." But Ahoyo told Ars that most early players ended up giving up on the ultra-hard level after an hour or less, "so it had no way of being uncovered." With the level making less impact than he expected, Ahoyo said, "the sunk cost fallacy of wanting to follow through with the failed troll, combined with new [real-life] responsibilities, allowed the nonsense to continue."

(While Ahoyo said he used TAS to finally finish and verify bombs5 in late 2017, he wouldn't upload that level until 2021, just before Nintendo halted level uploads completely. A human would manage to beat that level in 2022, long before its TAS origins were revealed last month.)

Ahoyo's "bombs5," which was created in sections to help ameliorate TAS syncing issues.

Ahoyo said he heard about Team 0%'s effort to beat every uncleared SMM level in 2018. At that point, he said, he "was holding on to hope that they would find the level one day" and realize its TAS-based origins. As the legend of Trimming the Herbs grew with Team 0%'s efforts, though, Ahoyo said the "clout and praise" he received for the level "felt awful."

Though he had long since moved on from the Super Mario Maker community, Ahoyo came forward on March 22 with a remorseful social media message finally revealing the existence of his TAS setup. "I'm sorry for the drama [the level] caused within the community," Ahoyo wrote. "In the end, the truth matters most... I deserve no credit [for the TAS levels] and will not deny Team 0% their victory."

Speaking to Ars Technica, Ahoyo said he saw what happened with TTH as "a situation which involves competitive integrity, but it's also a creative thing. I thought I could separate the two, but clearly that's impossible."

It’s about ethics in TASing

Following Ahoyo's admission, discussions of "competitive integrity" reared their head in the tool-assisted speedrun community, which "has strict ethical standards to label all TAS content as such," as TASVideos Ambassador TiKevin83 told Ars. The idea that previously unknown TAS tools had been used to actively troll the active Super Mario Maker community rubbed many the wrong way. "We never want TASes to compete with real-time runs," TASVideos publisher EZGames69 said.

Morimoto's famous SMB3 TAS spread widely on the early Internet without any indication of how it was created.

TASbot keeper Cecil linked Ahoyo's deception to Morimoto's famous early '00s video showing off tool-assisted tricks in Super Mario Bros. 3. When that video spread as a WMV (Windows Media Video) in a pre-YouTube Internet era, it usually did so without any indication of how it was created.

"The [Morimoto] video caused widespread fascination as well as significant consternation from the speedrunning community that felt the video was a fraud," Cecil said. That controversy led directly to the creation of TASVideos to "ensure Tool-Assisted Speedrun content is properly labeled and to provide a safe space for TAS runs to thrive without interfering with human speedrunning efforts," as Cecil put it.

While Ahoyo's deception went against that longstanding ethos, Cecil said he was still "relieved" when he heard Ahoyo's apology and suggested the honest admission was better late than never. "He fully shared all of the information about what he had done, and his reaction was the best possible outcome from the tricky situation he found himself in," Cecil said.

The ghost in the clock

Ethical issues aside, Ahoyo's admission set off something of a race for the TAS community to re-create Ahoyo's results with their own tools before Super Mario Maker's April 8 server shutdown. Those efforts ended up attracting dougg3, a first-time TAS creator who said he was "intrigued" when he stumbled across videos of people trying to beat TTH.

"I work on microcontrollers at my job, dealing with time-sensitive things, so [crafting a TAS] was right up my alley," he told Ars. "I thought it would be a fun little soldering/programming project to try to beat it."

Dougg3's TAS method involved an autopsy on a Wii Classic Controller hooked to a Wii remote.
An off-the-shelf HDMI processing board helped provide frame timing data for dougg3's TAS solution.
Close-up of the soldering needed to connect a Raspberry Pi to a Wii Classic Controller.
More soldering shots.

That "fun little project" involved soldering a whole bunch of connections between a Raspberry Pi Pico and the main circuit board of a Wii Classic Controller. Figuring out the specific pins to connect to wasn't that difficult, dougg3 said, nor was using an off-the-shelf breakout board to decode specific frame timing signals from an HDMI board.

But despite that timing signal, dougg3 said his setup only successfully completed the level 1 to 5 percent of the time, an oddly low success rate for a system that's supposed to play with robotic precision. In a video shared with Ars, dougg3 showed his TAS setup completing TTH just three times in ten full minutes of attempts.

Despite the same electrical inputs from the Wii U controller, dougg3 says the first two spikes on this graph result in Mario jumping while the third does not.
Despite the same electrical inputs from the Wii U controller, dougg3 says the first two spikes on this graph result in Mario jumping while the third does not. Credit: Dougg3

When dougg3 put together a simple test program to figure out the specific source of those timing issues, the results were frustratingly inconsistent. "What I found is although the actual electrical pulses I'm sending to the controller are exactly the same length each time, verified on an oscilloscope, sometimes Mario doesn't jump," dougg3 said.

Further testing within the TAS community has determined that the Bluetooth-based Classic Controller gets less consistent when it's far from the console. But even the Wii U's tablet GamePad—which runs on a 5 Ghz wireless standard that's less susceptible to distance-related issues—shows some frustrating timing inconsistencies across TAS runs.

"Maybe the Wiimote doesn't read the buttons quickly enough and sometimes never sees that A was pressed," dougg3 theorized. "Maybe a Bluetooth packet is lost or delayed. Maybe the operating system running on the Wii U adds to the delay and the game sometimes doesn't see it in time."

Cecil said further testing will determine if the Wii U's Gamecube controller adapter (which only works natively with Super Smash Bros.) can eliminate any wireless-related timing issues. Regardless, Cecil said that "our tests so far appear to indicate it's impossible to achieve 100% frame-perfect accuracy on the Wii U." Which makes it all the more incredible that, seven years ago, one Super Mario Maker creator was able to overcome those timing issues to hack together one of the most impactful gaming trolls of all time.

Photo of Kyle Orland
Kyle Orland Senior Gaming Editor
Kyle Orland has been the Senior Gaming Editor at Ars Technica since 2012, writing primarily about the business, tech, and culture behind video games. He has journalism and computer science degrees from University of Maryland. He once wrote a whole book about Minesweeper.
Staff Picks
marsilies
Although I know every move in TTH must be frame- and pixel-perfect, the clear video looked oddly effortless. Only sanyx91smm2 breathing at the end like they'd just seen the face of God really gives any indication of the difficulty.
I think that's part of what caused Ahoyo's troll to stay hidden for so long: it's much harder than it looks. I think that's also why it became the last uncleared level in the Team 0% project before it was disqualified: it's so short and deceptively simply that when players couldn't clear it right away, they thought "Oh well, time to move on and work on another level, that'll take more time/effort to beat. I'll either then come back to this level at some point, or someone else will have cleared it." The fiendish difficulty of it was hiding in plain sight.

I think sanyx91smm2 should get some sort of "John Henry" award for beating a level thought only beatable by machine by the creator.
dwangoAC
It would be interesting to see if they got more consistent results with a wired connection. Does the Wii U even support that?
I'm hopeful we can find a path forward to perform some kind of tests on unmodified hardware but the Wii U GameCube controller adapter only works with Super Smash Bros and we don't currently have a way to determine the frame-perfect accuracy of that. However, I've worked with Ownasaurus and other folks in my TASBot community to play back perfect inputs on a GameCube so we know that part can be reliable.
Prev story
Next story