2020 has been a tremendous year for Dream, a Minecraft personality who gained millions of followers so quickly that YouTube named him the number one breakout star of the year. Dream is also number two on the list of overall top YouTube creators of the year. All of this attention and acclaim has sprouted from his wildly popular speedrun videos, where he competes against other Minecraft players to complete the game as quickly as possible, sometimes setting records while doing it. But now, the very people who keep those records are contesting them.
Video game records are officiated by speedrun.com, which keeps tracks of the various categories and runners who vie for spots. In mid-December, the moderation team overseeing Minecraft records came together and published a 14-minute video that summarizes a two-month investigation involving a 5th place run submitted by Dream earlier in the year. It is meant as a primer on a much longer document, which is formatted as a research paper that breaks down all the high-level math the team did to verify Dream’s speedrun.
The paper is a whopping 29 pages long and includes a variety of graphs as well as concessions that take into account potential bias. Geosquared, one of the moderators on the team, tells Polygon that the group is composed of volunteers who are studying mathematics and computer science, which might explain the thoroughness and format of the report. Geosquared is a significant enough authority on the subject that he’s been featured by Mojang on the official Minecraft website.
Before we get into the specifics of what is being disputed, it’s important to note that Dream himself categorically denies all accusations. On Twitter, Dream has said he’s taking the time to make a “quality” response to the video above, while also noting that specific smaller claims within the video are incorrect.
“Sad to see people jumping on the hate wagon before hearing any opposing view point,” he wrote on Twitter. “That’s just how the internet works though!” Dream did not respond to a request for comment.
So, what’s so controversial here? The team took a series of six livestreams where Dream gets incredibly lucky in a way that many deem if not impossible, then at least extremely unlikely. Without getting bogged down in the details, the gist is that in order to get to the end of Minecraft, you need two specific items. The quickest way to get one of the items is to trade with Piglins, in-game creatures who will give you something random when you barter gold ingots. There is only a 5% chance that a Piglin might give you the specific thing you need to craft the item necessary to trigger Minecraft’s ending. The second item has slightly better odds, with a specific mob having a 50% chance of dropping said object after being killed.
In the handful of livestreams, Dream is shown successfully bartering for the key item 42 out of 262 times, whereas 211 of his overall mob kills dropped the second necessary item. In the video report of the livestreams, the team concedes that a small data set may not bear out the actual chances of the results — just because you flip a coin 10 times, for example, does not mean you’ll get exactly 5 heads and 5 tails. But then the team went ahead and actually accounted for any potential bias, and even giving Dream the benefit of the doubt statistically speaking, the odds are, in their opinion, incredible. They are so lucky that even compared to other lucky runs — which all top runs are, in some way — Dream’s odds are well above those of his contemporaries.
“If nothing else, the drop rates from Dream’s streams are so exceptional that they ought to be analyzed for the sake of it, regardless of whether or not any one individual believes they happened legitimately,” the paper reads.
And what, exactly, are the odds? For the trades, there is only a 1 in 177 billion chance of getting as many successful trades as Dream did. Giving him some leeway and attempting to account for some bias, the team still ended up determining the chances are 1 in 82 billion for the barters shown during the run. The mob drop rates that Dream had during his livestreams, meanwhile, come in at only 1 in 113 billion chances of occurring. Basically, the moderation getting one set of odds like the above is already unlikely, so having two extremely lucky outcomes like this puts the subsequent record-setting run even more into question. While they cannot prove it, the moderation team believes it’s possible that Dream is running a modified game of some sort.
Dream, for his part, has tried fighting the accusations. Ten days after he completed the livestream, the moderation team asked for the files that could show what was active in his folders when the run was completed. Dream provided the files as asked. Even so, the moderation team alleges that the files could have been changed during that 10-day time gap, which Dream himself says is likely the case as he changes his game up depending on what he’s streaming. The moderation team also alleges that the original files containing the settings for the run were deleted by Dream, but Polygon cannot verify what was given at the time. Dream has made a file public and available to download for anyone curious. But critics allege there are other means of altering Minecraft drop rates that do not involve mods.
I was asked if I had modified it, because they wanted to check modification dates. I obviously said that I had, because it had been 10 days, and I change what mods I'm using depending on the version I'm on. (dream SMP is 1.16.3, speedrunning is 1.16.1, servers 1.8, etc)— dream (@dreamwastaken2) December 14, 2020
While the video that caused all this ruckus has only been live for a few days as of this writing, the fight over the legitimacy of the run has been going on for weeks now. During that time, Dream has posted a variety of responses, calling the investigation nothing more than clickbait meant to rack up views, especially given Dream’s visibility and popularity. Dream further poses that the investigation was apparently flawed enough that some of the moderation team was threatening to quit over it, but Geosquared tells Polygon that this isn’t true.
“All moderators voted unanimously in our decision and no one is threatening to leave in protest,” he wrote in a Twitter message. “From everything we know that is unsubstantiated or complete hyperbole.”
Thank you for the totally unbiased, 2 month, 29 page "investigation" into whether a 16th place run had "too good luck", that was then made into a clickbait Youtube video by a head moderator (what a shocker)— dream (@dreamwastaken2) December 11, 2020
While it’s hard to say without speaking to every single person on the team, there must have been some consensus among the moderators — after all, the run is no longer listed on the world record page. Other runs that Dream has submitted that could be verified in full are still up. When asked why the team would put so much time and effort into disputing something that was never heralded as the number one run, Geosquared told Polygon that the placement didn’t matter to the team.
“Any run can be subject to scrutiny of course, in this case after members in our community brought the exceptional numbers to light, it would have been disingenuous not to look into it just because it was not the world record,” he wrote.
The veracity of the speedrun brings into question the engine that made Dream big in the first place. If viewers — many of them children who love Minecraft and may not be particularly critical or discerning about who is providing the entertainment — are tuning in because they are dazzled by Dream’s prowess, what does it mean if potentially some of it was faked?
Dream, meanwhile, maintains that all of this is a personal grudge of some sort — especially since he does indeed have legitimate runs that the team has verified. Why, then, fake a fifth-place video of all things? Onlookers, meanwhile, are attempting to make sense of what’s out there, though consensus seems to be skeptical too — one viewer, for example, tried to simulate billions of runs without running into the same odds as Dream apparently did.
but on top of that, the speedrun verifiers and mods have trash talked me in public in the most unprofessional way, and allowed server members to spread bs figures while responding to new people saying that "Dream is under investigation". This goes for multiple mods and verifiers— dream (@dreamwastaken2) November 28, 2020
There’s been fallout on social media, too, with one of the mods having to tell viewers to stop sending Dream negativity over the research paper. “Dream does not deserve hate,” he wrote. “Criticism, for sure, but not hate. Negativity is not justice, and being vindictive gets us nowhere.”
Dream has also apologized for the way he’s responded to the allegations, writing, “Although I have reason to be upset, I have no reason to act like a baby. I tend to act before I think when I receive intense criticism.”
While there seem to be attempts to smooth things over between both parties, the moderation team is steadfast in its findings.
“The events that were observed on Dream’s stream cannot be modeled by any sensible, conventional probability distribution,” the paper concludes at the end of its long investigation. “After accounting for any contributors of bias, the likelihood of this occurring is still unfathomably small.”
Correction (Dec. 17): A previous version of this article indicated that the Minecraft speedrun moderation team analyzed the odds shown in a single speedrun by Dream. We’ve edited the article to reflect that the mods took data from multiple livestreams.