Game Theory: Did Dream FAKE His Speedrun? A Final Analysis.

MatPat reconnects with a college acquaintance who is now a professor, while MatPat himself has become a professional YouTuber running Game Theory.

Game Theory breaks down complex scientific and mathematical topics in digestible ways for viewers who may not be comfortable with advanced math.

The cheating allegations against Dream in Minecraft speedrunning involved technical statistical papers that needed simplification for general audiences.

This analysis is purely educational and not meant to attack anyone; the goal is to address misinformation circulating about the controversy.

• March 2020: Set world record for Minecraft 1.14 in 32 minutes 21 seconds, beaten days later • June 2020: Claimed world record for Minecraft 1.15 in 22 minutes 4 seconds, which remains the fastest 1.15 run on record

Versions 1.14 and 1.15 are grouped together on leaderboards; 1.15 is considered harder due to removal of a villager restocking trick.

Dream's 1.15 run remains on the leaderboard without any disputes over authenticity or validity.

Dream began speedrunning Minecraft 1.16 in October, achieving a time of 19 minutes 24 seconds and placing fourth on the leaderboard, though this run generated suspicion due to exceptionally lucky drops.

• Ender pearls and blaze powder are critical for crafting eyes of ender needed to access the End • Both items require trading with or defeating NPCs, making drop rates crucial to speedrun times

• Dream achieved 16% ender pearl barter success rate vs actual 4.7% • Dream achieved 69.2% blaze rod drop rate vs actual 50%

42 of 262 piglin barters yielded ender pearls and 211 of 305 killed blazes dropped blaze rods across six live streams totaling over 20 hours.

The Minecraft speedrunning team investigated these abnormally high drop rates across multiple consecutive live streams rather than focusing on any single run.

• Flipping a coin 10 times and getting heads 7+ times has 17% probability • Flipping a coin 100 times and getting heads 60+ times has 2.84% probability

As data sets grow larger, results increasingly reflect true odds; larger samples should converge closer to the expected value, not diverge further from it.

With over 20 hours of live streams and hundreds of barter attempts, the rates should have converged toward the true 4.7% ender pearl rate, not remained at 16%.

If cheating occurred, it was flagged because Dream showed his work across multiple streams; an isolated offline run would have been unverifiable.

Winning the lottery once is lucky; winning multiple times raises fraud concerns. Dream argued he simply got lucky, like a one-time lottery winner.

• Dream's live streams were not randomly selected; reviewers specifically chose streams that appeared lucky • The sample started when luck seemed off and ended when Dream achieved his 19-minute goal • This creates selection bias that artificially inflates apparent drop rates

Both the speedrunning team and Dream's astrophysicist focused on calculating odds for any Minecraft player in existence experiencing such luck, not just Dream specifically.

The principle that extremely improbable events will eventually occur given enough time and attempts; applied here to assess whether Dream's luck falls within statistical possibility.

• Minecraft speedrunning team: 1 in 7.5 trillion (assuming 1000 speedrunners) • Dream's astrophysicist: 1 in 100 million • Commonly cited 1 in 10 million figure is actually a different measurement not directly comparable

• Different application of stopping criterion • Different estimates for total number of live streams • Different accounting for sampling bias • Different approaches to p-hacking bias

The disagreement stems from many smaller answers rather than one big answer; numbers are rough estimates based on assumptions that can feel arbitrary.

Mathemaniac's video provides detailed mathematical and statistical breakdown of all arguments point by point.

P-hacking (also called data dredging) is the practice of looking for patterns and selectively presenting data to reach a predetermined conclusion.

• Game Theory's Mario analysis selected data points from hundreds of games to argue Mario is a sociopath • This ignored decades of Mario acting heroically to save the Mushroom Kingdom • This demonstrates how selective data can lead to any conclusion

Instead of forming a hypothesis and testing it, post-dicting starts with data and creates a hypothesis afterward to explain it.

• With 10 possible factors and 2 unusual ones, cheating appears suspicious • With 1000 possible variables, cherry-picking 2 can create false correlations • Example: 99% correlation exists between declining Maine divorces and declining margarine consumption, but they are completely unrelated

• Pearl trades and blaze rod drops (already analyzed) • Blaze spawn frequency • Number of available piglins for bartering • Other unmeasured randomness sources

Minecraft speedrunning team estimated 10 different factors; Dream claimed closer to 37 factors could affect drop rates.

No clear-cut answer exists because it depends not just on how many variables exist but on how measurable those variables actually are.

Some calculations are based on assumptions that feel arbitrary, creating legitimate room for debate about the final odds.

• Minecraft 1.16 (Nether Update) created such significant changes to warrant its own leaderboard • Pre-1.16 fastest time: 19 minutes 36 seconds • 1.16 fastest time: 14 minutes 36 seconds (5-minute difference)

• The speedup does not come from tricks like bunny hopping or diagonal bridging • The speedup comes entirely from the piglin barter system for ender pearls • This is pure randomness, not a learnable skill

• Unlike other speedrun randomness that forces improvisation and 'sight reading' skills, barter RNG is like a slot machine • Bad luck can ruin even the most skilled runs • The luck swing is massive and arguably breaks the speedrunning experience

Should Minecraft 1.16 have a speedrun leaderboard if success depends primarily on patience to wait for random luck rather than player skill?

A more accurate yearly indicator might be a tournament of speedrun races where two or more players compete in the same random world seed.

• Tests who can best adapt to unexpected situations • Rewards knowledge of game mechanics and problem-solving • Removes the luck factor as the dominant variable

Piglin bartering could be restricted from competition, similar to how Super Smash Bros tournaments ban certain broken or unbalanced characters.

A potential silver lining to the controversy is the possibility of competitive Minecraft racing becoming a viable esport format.

Upholding integrity on speedrun boards is absolutely essential and non-negotiable.

• The statistical debate may be missing a more fundamental question about whether 1.16 should have a traditional speedrun leaderboard • The dependence on extreme luck changes what speedrunning represents

A major positive outcome is that people have been forced to learn statistics and probability to participate in the discussion.

Learning is fun and should always be fun; the controversy brought mathematical thinking into mainstream gaming discussion.