(7) Dream Cheated Every World Record
https://www.youtube.com/watch?v=2lkxXfMFu2Y
无作者
You’ve heard about Dream’s cheating, a Minecraft speedrun.It was a 19-minute run performed on Minecraft 1.16.1 in October of 2020.And it was not a world record.It was fourth place,but it was Dream’s fourth and final submission, to the Minecraft speedrun leaderboards, because it would be proved to be cheated in the months following.Before that though, Dream submitted three world records in a row.And to date, no hard conclusive, evidence has been released that they are cheated.Dream still denies cheating that 1.16 run intentionally.He claims that he was sent a mod, that, added cheats to the game, and he didn’t realize that’s what it was for and that he used it accidentally.This is something that has been covered by him and by Karl Jobst, um in varying detail a couple times about five years ago,but the full details of it never really came out.So, we’re going to start there.2020 was a huge year for Dream and also Minecraft speed running.For Dream, he started his most popular series,Minecraft Manhunt,
at the end of 2019 in December, and he would continue it all the way through today.Really though, 2020 was when it gained popularity.He already had well over a million subscribers, by March of that year, and then in the latter half specifically was when he would go super viral.In terms of Minecraft itself, the Nether update released in June.And this would prove to be a controversial, but very important update for speed running.Initially, the strategies for 1.16 were not very developed.People would mine gold in the Nether manually, and try to find lone piglins to trade with.This was not a very fun strategy,but it was still more effective, than what we had for previous versions of the game.By this point,Dream had already made an entrance into the speedrun scene proper with not one,but two back-to-back world records in March of 2020.These were on an older version, 1.14.4,widely considered to be the best for speedrunning.But then he would surprise everyone.
by achieving a third world record in Minecraft 1.15.2, with a strategy that was thought not to be viable by most.Minecraft 1.16 would release shortly after though,and over the next several months,the old version records would get absolutely demolished.By the end of September, Corbinos set an under 15-minute record,a major surprise to everyone in the scene.Dream did not believe this record was real.He downloaded the world file, and looked for anything that would indicate cheating.He saw that the timestamps, on the files were from before the run was supposed to have occurred.He took this evidence to the mod team, and unfortunately,Dream and some members of the team went to Corbinos’s chat, and basically trolled him,taunted him.It was not appropriate,and it turned out that this was just a time zone difference.Corbinos was in a different time zone than the people looking at the files,and when they figured this out, the mod team apologized and verified the run.
It wasn’t really just about this specific run though.In general,Dream didn’t like how much more luck-based the new version of the game was.He didn’t really respect the leaderboards or the runners for it.At a later date,he would end up issuing an apology to an entirely different speedrunner, that he had never actually accused of cheating in the first place,just to give an indication of how he felt about the board at this time.Dream had given speedrunning 1.16 a shot at the beginning of July,but nothing super impressive happened.He didn’t like the version that much, kind of bounced off of it.Three days after the false accusation against Corbinos,he would start a second run of streams that would become infamous.It was determined that the drop rate for blaze rods were modified, to give better luck to the speedrunner,that the trading rate for ender pearls from piglins had been modified,and it would later come out that even the likelihood, of an eye of ender to crack had been modified as well.
Dream did these runs from October 4th to October 9th,when he ended up setting the 19-minute, time that he would submit to the leaderboards.A week afterwards, Mine Crusader,another world record holder who would later turn out to have cheated,tweeted out about Dream’s unusual ender pearl drop rates from trading.This prompted some members of the Minecraft, speedrunning Discord to just run the numbers,run the stats, and the answer they got was not good.We’re talking one in hundreds of billions here.At this point,Dream hops into the server to argue with people about the statistics.I happened to open the chat, and I just asked him,“Well, how many samples are we talking about here?” And in my head,I’m thinking, “Uh maybe it’s 30,maybe it’s 40, maybe the the chance of him doing whatever he did is 4%.Dream himself responds that the sample size is 240, and I do a double take.That’s not good.Two days later, the moderation team asked Dream for his mods folder,and he responded.
that he’d already deleted all of it after the arguments in Discord.He said that the only mods he would have had downloaded would have been Sodium,a popular Minecraft performance mod,and a quote chat mod that he used for recordings.The team released both a statistics paper as well as a video, going over their findings against Dream, 2 months later in December.There would end up being quite a lot of back and forth over this.Dream released a couple of different responses, and addressed it on stream a few times.This involved like a retort to the math paper that he paid a professor to write.Uh the retort had problems, the team responded.It was a whole thing.It’s been covered a bunch of times.Eventually in February, Dream ends up privatizing his response video.The team end up bringing him in for a conversation, about his 1:15 world record as well.Dream ends up apologizing, for a lot of the events that happened with regards to the 1:16 run,but he doesn’t admit that he cheated.
And when the mods tell him that they’re considering removing his 1:15, record as well over suspicion that it’s cheated,he argues that it shouldn’t be done as a punishment.Things go dormant for 3 months,but in May the team end up pulling his 1:15 world record from the boards,and Dream responds with a Pastebin.In this Pastebin,he admits that the 1:16 records were performed with a modification.He again claims that this was a chat mod, that he didn’t know affected the game’s functionality.He also uses the term recording mod,and it later comes out that his Manhunts, and other videos are starting to use a Dream Team mod, to make certain quality of life improvements,like reducing the sound of the dragon in the end,reducing particles because YouTube doesn’t like them, um things like this,things for recording.This prompts Karl Jobst to look back into the dream situation, and he opens lines of communication with both Dream and with us.Our main concern is whether the mod had any verifiable functionality.
other than cheating.Because, if there’s no proof that this mod did anything other than cheat his game, at the time of the run,we’re not really sure why we’re supposed to believe it.Karl ends up getting access not only to Dream,but also the mod developer, who Dream alleges concealed, the existence of the cheats in the early builds of the mod from him.The answers we get really don’t make a lot of sense.They tell us that the only thing that this mod did at the time, was make the game start up faster.That Dream on his high-end gaming PC with a 9900K, was having issues just getting the game to load in the first place,and so he specifically, requested, from the developer an early build of the mod just to make the game start faster.This really stinks to me in particular because, I know the mod they’re using for this.It’s called Lazy DFU.I’m the one who legalized it for use in speedruns in 2021, not 2020.In fact, this mod didn’t even exist until December of 2020,after Dream’s runs.
All that did was a precursor called Data Breaker.Data Breaker, has a 16-star, GitHub repo and it was only intended for use by mod developers with bad PCs, who were having trouble restarting the game constantly, in a developer environment like IntelliJ, IDEA, or Eclipse.Um the mod actually, doesn’t even let you run it outside of a developer environment, without passing like special flags to Java.And it’s named data breaker because if you load a world from an old version,it will just break that world.We have Karl asked Dream about this,and he ends up getting a reply like 3 hours later with screenshots, of Dream messaging the dev,and then the dev responding within a couple minutes mentioning data breaker.Now, personally to me,I feel like we’ve just blown a chance to catch them in an obvious lie.I feel like if I was communicating with them directly,we would have done this as a Discord call,and it just would have caught them then and there.
Because I don’t I don’t believe these people knew about either data breaker, or lazy DFU until lazy DFU was legalized for speed running.Honestly, Dream’s fabric mod dev,he doesn’t seem to be like that into it.All of the fabric mods that are posted on Dream’s Patreon,they depend on fabric API because the example mod depends on it.They suggest you install another mod called flamingo, because the fabric example, mod suggests you install another fake mod called flamingo.I just don’t think they knew about any of this stuff.Also,this does not, make sense with how he’s been referring to the mod up to this point.Number one,we asked him repeatedly, and Karl repeatedly, about why it was called a chat mod, all the way from back in October when he was talking with the mod team, through to May when he’s posting the public pastebin.And the answers we get are just baffling.He says, “I quote,I guess it’s just me dumbing down language so a layman could understand.Just naturally.
do it as a creator who tries to cater things to the most people possible.”, He says he saw another speedrunner, using a mod to show his Twitch chat in his game.And according to Dream,he used chat mod as a general shorthand for quality of life mod.Now, for me,when I’m hearing this,it it seems obvious that Dream is just making this up as he goes.He said chat mod originally because he’d seen another guy use a chat mod.He later realizes there’s no evidence of him using a chat mod anywhere.And also,there’s no plausible, reason, that a random chat mod he downloaded, from the internet would have cheats built in.He needs to change the story.He needs the mod to be something created in-house, by his own people to do with the Manhunt, so that there’s, some sort of way you can spin it that cheats would end up in it accidentally.And that’s exactly what he does.There’s still a critical problem for him, though.There is no observable evidence that the mod ever did anything but cheat.
There’s no reduced sounds in the end, no reduced particles for recording,no alternate coordinates for easier viewing on YouTube.There’s just nothing.And more so, there’s no proof that the mod ever had quick loading functionality.Certainly not any proof from before the day he told us about it.We made repeated inquiries to get access to, past and current versions of the mod,and every single one of them was denied.We just wanted some way to establish, that literally just before the day we started talking to them,the mod had the quick load functionality.And we were we were just brick-walled everywhere.Karl felt we were being conspiratorial, about it when we said that we wouldn’t trust a screenshot, cuz screenshots can be edited.With that said, we also don’t get any screenshot evidence,either.Um they tell us, that every single person in their team, mass deletes all of their discord messages, including the developer, of the mods.We get told that the mod was being developed for three versions simultaneously,
1.15, 1.16.1,and 1.16.3 with no version control software.Like if you’re familiar with GitHub,Git is the software that people use to upload there to track what they’re doing.We get told that this mod was being made for three versions simultaneously.None of that.No record of it anywhere.We do end up getting a direct interview with Dream himself,me and a couple of the other mods.And, in this interview, he tells us that at the time of the cheating there was no other functionality.That the the soonest anything else would show up would maybe be his, Hitman video a full month later.That it might have like improved F3 coordinates,but he couldn’t find them looking through the video.So, for reference here,we’re talking about a full month after he’s been accused of cheating.That’s the first time he’s even proposing there might be evidence, of this Dream Team mod existing.Moving away from the details of the mod itself,there are other things that just aren’t adding up.
Number one, right after Dream finishes his 19-minute run,he quits the game, he changes to a different version of the game,1.16.3, and he swaps out his mods so he can play on the Dream SMP.This means that he opens his mods folder.He sees the cheat mod in there.He moves it around.And this is before he’s submitted anything to the the leaderboard team.That wouldn’t happen until an hour later.Now, when he does talk with the leaderboard team,he offers to send them whatever and is only asked for world files.Obviously, it would have been best to ask for the mod files,the log files, all this other stuff,but he’s he’s already been tampering, with his mods folder from right after he finished the run.After the interview,I also managed to find evidence, that he was swapping out his mods to play on Dream SMP, during the week of the cheated live streams.In fact,right before he started his final live stream where he got the 19-minute run,
he went to play on Dream SMP and we still have the footage of that from Wilbur.So, to wrap up about 1.16 here,the full story we’ve got by the time Karl publishes his video, is that Dream’s high-end 9900K, gaming PC is having trouble just starting Minecraft,just getting to the main menu at all.Um that he cannot replicate this for us by the time of the interview,even though he’s on the same processor and everything,he can’t replicate it anymore.That uh he referred to this loading mod as a chat mod, all the way from October 2020, through to May of 2021, because he’s a good communicator and YouTuber,I guess.That his developer, then sends him an updated version of this loading mod that just, adds cheats.I guess it doesn’t update anything else cuz we asked them repeatedly, what other loading optimizations, there would have been other than just a copy-paste of Data Breaker,and we never got an answer.And then then after the cheating scandal blows up,the developer has already removed the cheats from the mod.
He conceals that there were cheats in the mod from Dream.From the get-go in this process,Karl was like the good cop and we were like the bad cop.Um I initially was very dismissive of like even investigating this closely,which was not the right approach.There was There was evidence for us to work with, evidence for us to look at.And through like arguing with Carl,we did obviously turn up a lot of different evidence.But at the end of this like he was he was losing sleep,he had to meet a sponsor obligation,and I I think, being kind of locked in there as the only person talking to the dev, and for the most part Dream too,other than our one interview, it affected his judgement a bit.At some point we had this argument with him where he was saying that, Dream would have to be like psycho to be lying about all this different stuff.And we’re giving pushback.We’re like “No,I mean he he’d just have to be like trying to protect his reputation.”, But But at this point Carl is like pretty convinced.
He releases his video.He leaves it open-ended, which we appreciate.Following that,I make one more message to Dream asking if he’s, gotten any of the info we asked him about.I get no response.At this point,we’re at least pretty satisfied, that Dream’s not going to try to interact, with the scene much after he like finishes, donating $50,000, to various Minecraft speedrunners.And, there’s just no appetite on the mod team, to try and do this again with no smoking gun.Because the first time around, there were multiple smoking guns.There were the pearls, there were the blaze rods,both just impossibly lucky.But regardless,there was just a ton of friction involved, from releasing something publicly about a guy this big,his responses, his fans’ responses,all the, controversy it generates, how carefully everything needed to be reviewed.There There was just no appetite, to really address it further if it looked like Dream was going to leave anyways.At some point not too long after,
a member of the team Bloons actually realizes that there’s a problem, with Dream’s old 1.14, world records,his first two speedruns ever.When you click create world on Minecraft 1.14.4,there are a number of steps for the world to generate.Number one, the screen goes blank,all the buttons disappear.Number two, the world generation visualizer shows up,and it stays stuck at 0%, with a completely black visual for at least a full second.What’s happening here is actually stronghold generation.Minecraft generates the strongholds, before it generates any of the chunks around spawn,before you get anywhere near them.And later on,a member of the community would actually create a mod called Lazy Stronghold, to make this happen in parallel with the normal world generation, to let speedrunners reset for new worlds even faster.At the time of Dream’s original 1.14 runs, this was not commonly known,but in 2021, people start to pick up on this because of the mod.
I’m pretty sure Bloons brought this up in some sort of team chat that year,but at the time, the mood was kind of we’re trying to move on.I think I think I might have mentioned that if we get a smoking gun,we would still want to put it out,but Bloons isn’t entirely sure about what he’s found,we’re not super familiar with the details on it, either.This just gets backburnered at the time.And then,the next time something happens is when it turns out that Minecraverger,the guy who actually first posted about Dream’s ender pearl trading rates,has himself cheated a ton of world records.During this time, I took another look at Dream’s 1.15 run,and I actually got visual confirmation of a splice.But, another runner,11D, proposed a possible way to generate the same effect without cheating.I managed to reproduce something that looked similar,and I just dropped it there.In retrospect, we probably could have proved it was a splice with that alone.
For the full details on how the 115 run was confirmed to be spliced,go watch ElioTex’s video.That one is dedicated solely to the 115 run,the several different ways ElioTex, and the guys behind the recent Manhunt exposés, managed to confirm the cheating there.Getting back to the timeline, at the end of 2023,Dream posts “The Truth”, and has a section in there dedicated to the Minecraft speedrunning scandals,which he ends as such.This prompts us all to take another look at things, to re-evaluate.Um, Balloons brings up the 114 cheated runs again,and this time we all take note.In April of 2026, I ping in the channel where we’re discussing this,and I’m checking if we have anything else about the 115 run,because I’m convinced that the splice evidence I found was a false positive.ElioTex actually finds his wallpaper evidence that same day.From there,we unfortunately get sidetracked, looking into like experimental splice identification methods.
They don’t end up leading anywhere at the time, and a month after,there are actually some doubts raised about the 114 world record splicing.Basically,it’s proposed that through some very precise inputs with like the Windows key, and alt-tabbing and losing focus,you could maybe get some gaps in the video recording in Dream’s run, that might explain, what appears to be the missing stronghold time, as just lost focus during that time.Fast forward to September,ElioTechs and the others independently rediscover, the splice evidence I found in 2022,but they take it a couple steps further and manage to rule out the possibility, that it’s a false positive.Um at this point we start seriously talking about putting info out, and I go back to review the 1.14, stuff.And really what this means is I just end up spreadsheeting the info.I spreadsheet all the info about, how long each stage of the world generation is taking, and then I also find a live stream, that Dream did only a few weeks after his 1.14 world records.
and I spreadsheet the world generation from all 47, runs in that live stream as well.The end result is that the shortest time to finish stronghold generation, from any of these 47 streamed runs is over 1.2 seconds, compared to half a second for his first world record, and like 2/3 of a second for his next world record.The only thing you could do with my methodology to make it more favorable, is to start measuring, from the first frame that the mouse is even on the create world button.We don’t get a very good signal of when the button is pressed.I did my data tracking, starting from when the mouse stopped moving on the button.You could instead, measure from the first frame the mouse is on the button at all.This only gets you another three frames for both world records.So, that gets you like another 50 milliseconds.Like it it doesn’t change anything in a material way.You’re still talking about his time to finish stronghold generation being, over twice as fast as any of the streamed runs I looked at.
Or at least almost twice as fast.And we actually have 11 different, streamed speedrun sessions from Dream from before his 1:16 cheated runs even.It’s about 30 hours of footage in total.Not all of that is speed running,but like if somebody wants to go through all that and charted all out,I challenge you.Find anywhere where the total time from the mouse, being over the create world button, to the end of stronghold generation, is even under a single second.Much less like .7 seconds, .5 seconds.I just don’t think you’re going to find it.It’s not physically possible for a 9900K, to get through the world generation on these seeds this fast.Unless, you know,Dream’s developer totally had lazy stronghold, like a year early and jar modded it into his 1:14 or whatever,right?Something else I’d like to point out, is that after Dream’s first couple world records,there was actually a fake world record, posted by somebody calling themselves Drem.D R E M.Dream actually calls them out.
And he points out a new anti-splicing, technique that the mod team were not previously aware of.Which is that after you’re through with stronghold generation,each world has a unique visualization, output or at least a relatively unique visualization output.Now,he claims he knows this from his like plugin development, days when he had to look at world generation.I I think the reality here is that it’s another Mine situation.He knows what to look for because he’s been looking at his own spliced, runs trying, not to get caught.And he’s noticed this.He’s made sure that at least this part was correct for all three, of his world records.I mean, by the by the 115 world record,he gets the stronghold generation included as well, right?Like he’s he’s looking at what he’s doing.He’s trying to eliminate the errors.And this just makes you paranoid.It makes you like see cheaters everywhere.Occasionally, you’ll be right about it,too.What you might miss is when after your world record,
somebody gets a second place run.We’re going to go in reverse order here.First, I’d like to congratulate Ben Ex on another world record.Your 22:12 PB was actually the fastest run ever done.And finally, on March 29th of 2020,5 days after Dream cheated 26:44, Alex Meister got a 26:53 actual world record.This has never been acknowledged anywhere.This run has actually just disappeared, completely from the internet because it wasn’t highlighted on Twitch.All we have is a clip of the world generation, that chat presumably took to help find the video to post on speedrun.com.I’d like to take a moment to congratulate, Alex Meister for getting this world record.And while we might not have footage of Alex Meister’s world record,we do have footage of him finishing a run.>> Sub sub sub now.Sub now.Sub now.Clip it.Clip it.Clip Clip it.Clip it.Fans go.When you just let it go, talking about your Lambo.