For months now, popular Minecraft streamer Dream has insisted there was nothing fishy about six “Any% Random Seed” speedruns he streamed last October, despite evidence to the contrary presented by the moderators of clearinghouse Speedrun.com. Over the weekend, though, Dream said in a message posted to Pastebin that he had “actually been using a disallowed modification during ~6 of my live streams on Twitch” while maintaining that he “didn’t have any intention of cheating.”
The admission seems to finally put to rest months of drama and dueling accusations between Dream and the mods, settling an argument that relied on complex mathematics to prove that Dream’s runs were vanishingly unlikely to be the result of random chance alone.
The math
To understand the accusations at play here, first you have to understand just how much of a role luck plays in a top-level Any% speedrun of Minecraft and how ridiculously lucky Dream appeared to be in the streamed runs in question.
Quickly beating Minecraft is a process with a lot of steps. One of the most crucial is bartering your gold with Piglins in the hopes of getting Ender Pearls, which are needed to eventually construct a portal that will take you to an area known as The End, where you can fight the Ender Dragon that serves as the final boss of the run.
In an unmodified game of Minecraft, Piglin randomly drop Ender Pearls in about 4.7 percent of trades. In the six speedruns Dream livestreamed starting in October 2020, though, a full 16 percent of the Piglin drops were Ender Pearls (42 out of 262).
Dream’s luck didn’t end there. He also had to battle several Blazes to obtain Blaze Rods for the portal. Defeated Blazes drop Blaze Rods 50 percent of the time in an unmodified game. In the livestreams in question, Dream got a Blaze Rod from just over 69 percent of defeated Blazes (211 out of 305).
Being lucky isn’t proof of cheating, of course. Unlikely things do happen sometimes, and speedrunners often power through thousands of runs in the hopes of getting that one random seed that will let them shave just a few seconds off their best time.
When you sit down and do the math, though, the chances of Dream being this consistently lucky, across this many consecutive livestreamed runs, become so small as to be practically zero. In a 29-page report issued last December, the Minecraft Speedrunning Team went into excruciating detail to calculate the chances of those Ender Pearl and Blaze Rod drop rates (or better) occurring in a truly random set of speedrun attempts. The final odds: 1 in 20 sextillion, or 2 x 10^22.
The Stand-Up Maths channel adds some context for just how lucky Dream would have to be to do his streamed runs unassisted.
The Stand-Up Maths YouTube channel has an entertaining breakdown of that math that puts that level of luck into context. There, author and host Matt Parker calculates that a theoretical task that was performed by each of 10 billion humans once every second for 100 years straight would result in 3 x 10^19 random trials. Yet even with that many trials, you wouldn’t expect to get a specific, pre-sought outcome that was as unlikely as what was shown in Dream’s speedruns (which, again, would happen once in 2 x 10^22 trials, assuming they were legitimate).
“If anything is plausibly going to happen as being done by a human, it has to have odds below [3 x 10^19],” Parker summarized.
The controversy
Math or no, Dream didn’t take these accusations lying down, and he went so far as to commission a rebuttal report from an astrophysicist through science-consulting company Photoexcitation. That report made use of a variety of simulation assumptions and “bias correction” algorithms to determine a more reasonable 1 in 10 million chance that some Minecraft speedrunner would encounter Dream’s level of luck at some point in the last year.
In response, the Minecraft Speedrunning Team issued a rebuttal to the rebuttal, pointing out problems with the initial rebuttal’s methods and assumptions. In any case, even taking the first rebuttal at face value, the second rebuttal points out that “the 1 in 10 million posterior probability… still impl[ies] a 99.99999% chance of Dream cheating.”
In the meantime, Dream loudly and publicly decried the whole process. “Luck is not a reason to accuse a runner (who has gotten multiple world records, and done thousands of speedruns) of cheating,” he tweeted last November. “The level of disrespect I’ve been shown is just ridiculous,” he continued. “The mod team has shown inability to remain unbias [sic], and has been unresponsive and unprofessional.”
That kind of talk led to plenty of back-and-forth arguments and online harassment between Dream’s many fans (he has 4.7 million Twitch followers and 22.8 million YouTube subscribers) and those backing the Speedrun mods’ conclusions. “I’m sure this drama has been stressful for most of the Minecraft community, and a lot of that was probably due to my original response to the drama, so I take full responsibility for that,” Dream said in a January tweet that he claimed would be his last one regarding the issue.
An “official moderator analysis” lays out the mathematical case against Dream’s runs being the result of luck alone.
This weekend, though, Dream offered a lengthy explanation of what happened, writing that after “exploring the possibilities” involved with the runs, he “ended up finding out that I HAD actually been using a disallowed modification during ~6 of my live streams on Twitch.”
Dream chalks this up to the use of an “overall recording mod” that he used after Minecraft was updated to version 1.16 last June. He wrote that he thought this mod was “basically just a chat mod” without any gameplay effects and that he “was 99% sure” it wasn’t on during the runs. In actuality, though, that mod was broadly similar to separate mods Dream had used previously (and openly) in videos prior to version 1.16 to “increase the enderman spawn rates and pearl drop rates out of convenience. It makes the videos better because we don’t spend hours looking for pearls or spend so much time farming blaze rods.”
“I felt really terrible for the mods because I dragged them through the mud even though they were mostly right.”
Dream
When Dream realized this a few months ago, he wrote, he deleted a scathing video response he had posted regarding the Speedrun.com mods. He didn’t announce his finding publicly then, though, because he said he “felt like the community had been through enough drama and that it was pointless. I didn’t want to be the center of controversy for the hundredth time.” He said he figured “it would be a story I would tell in a few years when no one really cared.”
Looking back on the whole thing now, though, Dream wrote that he was too quick to assume “that I was being targeted due to the fact that I was a YouTuber” and that the speedrun mods “were out to get me.” Dream wrote he became “paranoid and didn’t think straight,” leading him to become “scared and stressed and [say and do] shitty things.”
“When the drama first started, I cared more about defending myself and being right, then [sic] about figuring out what was actually going on and I shot myself in the foot by doing it,” he wrote. “I felt really terrible for the mods because I dragged them through the mud even though they were mostly right.”
Now that Dream has owned up to his mistakes, at least some in the Minecraft speedrunning community are ready to move on. “Dream and I are cool again,” tweeted Geosquare, who worked on a video explaining the Speedrun mods’ decision. “Glad we’re seeing some closure today.”