Michael's Appel Speedrun Controversy

1. Suspicious Variable in 15.4 Level 5 Run

• Michael’s 15.4 Level 5 run includes a variable not present in the vanilla game.
• This variable appears to be a TAS (Tool-Assisted Speedrun) input display.

• In later runs (15.3 and 15.2), he removes the variable from the recording but maintains the same perfect, repetitive side jumps seen in Sir Farcelot’s TAS run.
• Michael rarely shares replay codes, making verification difficult.
• However, in his 15.2 run, he did share a replay code in he video’s description: 12346370jlp-141304LJLJ6LJ79LJ59LJ1LJ68LJ9LJ71LJ1LJ87LJ9LJ100LJ1LJ123LJ2LJ127LJ10LJ132LJ9LJ133LJ1LJ139LJ0LJ140LJ2LJ141LJ10LJ142LJ9LJ147LJ1LJ148LJ3LJ149LJ9LJ150LJ1LJ152LJ3LJ153LJ9LJ154LJ1LJ156LJ3LJ157LJ9LJ158LJ1LJ160LJ3LJ161LJ9LJ162LJ1LJ164LJ3LJ165LJ9LJ166LJ1LJ168LJ3LJ169LJ9LJ170LJ1LJ172LJ3LJ173LJ9LJ174LJ1LJ176LJ3LJ177LJ9LJ178LJ1LJ180LJ3LJ181LJ9LJ182LJ1LJ184LJ3LJ185LJ9LJ186LJ1LJ188LJ3LJ189LJ9LJ190LJ1LJ192LJ3LJ193LJ9LJ194LJ1LJ196LJ3LJ197LJ9LJ198LJ1LJ200LJ3LJ201LJ9LJ202LJ1LJ204LJ3LJ205LJ9LJ206LJ1LJ208LJ3LJ209LJ9LJ210LJ1LJ212LJ3LJ213LJ9LJ214LJ1LJ216LJ3LJ217LJ9LJ218LJ1LJ220LJ3LJ221LJ9LJ222LJ1LJ224LJ3LJ225LJ9LJ226LJ1LJ228LJ3LJ229LJ9LJ230LJ1LJ232LJ3LJ233LJ9LJ234LJ1LJ236LJ3LJ237LJ9LJ238LJ1LJ240LJ3LJ241LJ9LJ242LJ1LJ244LJ3LJ245LJ9LJ246LJ1LJ248LJ3LJ249LJ9LJ250LJ1LJ252LJ3LJ253LJ9LJ254LJ1LJ256LJ3LJ257LJ9LJ258LJ1LJ260LJ3LJ261LJ9LJ262LJ1LJ293LJ9LJ294LJ1LJ357LJ9LJ384LJ1LJ402LJ9LJ420LJ1LJ442LJ9LJ444LJ1LJ458LJ9LJ461LJ1LJ481LJ9LJ496LJ1

https://www.youtube.com/watch?v=xbRtFE_wugw&list=PLl54lUDD_62NCLsu5GN_eTTNxbG36xRAl&index=29

• Analyzing the input graph for this run reveals a level of precision that only a TAS could have as it is unnecessary to be that precise for a 15.26.

Tool I made to graph runs: https://turbowarp.org/1147978412

When asked about this, michael said “'characteristic' of tas - nowhere a proof, scratch is a 30fps game, perfecting it is possible. Plus, never is it like i get it perfectly everytime, but rather just selected the best one out of the hundreds.”

This does not make any sense because you do not have to have perfect inputs in order to get a 15.26 or even for a 15.23, you just need to have good inputs. Also, it took me way more than “hundreds” of attempts for my first 15.2x, but maybe it was just an estimate.

2. Questionable Route Choice in The Beginning of Level 7

• In his WR run, Michael goes under the crusher at the beginning Level 7. This is an extremely difficult route that does not save time.
• Even a slightly seasoned player would know this route does not save any time.
• It appears he copied a TAS without understanding why the route was used.
• When asked about this, Michael said “That route is slightly faster for any%, in Level 7 you need to wait a bit, which is unoptimal for full game runs.” This is just straight up false, as you can see in these two images, one from my pb and one from Michael’s 3:50, you get on the blue platform at the same time due to the cycle.
• Michael’s 3:50:
• 
• My pb as of writing this:
• 
• You get there at the same time. Michael’s argument is invalid.

3. Inconsistent Statements About Automation

• Initially claimed to use an autoclicker to press "Continue."

• Later admitted to modifying the code to bypass the need to press it at all.

• Both are acceptable, but this contradictory story makes me wonder what else he might be hiding.
• When asked about this, Michael said “Yes, autoclicker first, then modified. Published the project, asked speedrun dot com leaderboard moderators for permissions, accepted everything.” This would make sense, but if the mods gave permission, why did Michael say “autoclicker is not against the rules, I assume” if the mods did give permission, why would he have to assume?

4. Unrealistic Speed of Improvement

• The skill gap between a 3:59 and a 3:50 run is massive.
• Achieving such an improvement requires extreme dedication and practice.
• As a student, Michael likely lacks the time to grind this much.
• Just improving from a 3:59 to a 3:55 took me over three months.
• When asked about this, Michael said “In videos i have revealed, it's not a direct jump from 3:59 to 3:50, but a lot of inbetween unposted. See fail montage.” If when I first got a 3:55, I was not thinking, “this run SUCKS, I just shouldn’t post it until I get a 3:50”, I was thinking “This run has a few parts that could be improved, but if I submit it I will move up a place and pass Swskrei!” for Michael, getting an 3:55 or faster would mean taking back the world record, so it would be more logical for him to just submit it instead of waiting.
• Also, Michel said “fail montage.” The fail montage he is talking about is clearly not at the skill level for a sub 4. The fail montage is going for the strats that he went for in his 4:18, not strats used in a sub 4. The fail montage does not include anything from his sub 4 runs.

5. Lack of Supporting Evidence for His Runs

• Michael only posted his WRs and never shared full grinding sessions.
• Legitimate record-holders usually stream or upload full sessions for transparency.
• If his runs were legitimate, he should have no issue proving it.
• When Michael was asked about this, he said “See fail montage.” As said in #4, the fail montage is from his grind from his 4:18, not from his sub 4.

6. Defensive and Illogical Justifications

• His primary defense is: “I’ve contributed so much to this community.”
• Contribution does not equate to legitimacy.
• Instead of offering concrete proof, he plays the victim to gain sympathy.

7. Timing of WR Submissions

• Michael submits WRs shortly after losing them, often claiming they were achieved earlier.
• Example: After losing the WR on October 5, 2023, he submitted a run four days later, claiming it was set a month prior.
• This pattern repeats multiple times.
• Michael said “Four days. Four days is almost 100 hours. Yes, I get angry when my record is stolen, but that just means i get motivation to improve.”
• 1) 4 days isn’t enough to go from 3:59 to 3:50, just ask me or bas. Also, -1 month is definitely not enough time.

8. History of Cheating

• Previously banned from speedrun.com for cheating.
• His 4:07 run was proven to be sped up (x1.06 speed).
• Unbanned due to hype around his sub-4 run, despite his past offenses.
• Michael said “Sped up was due to music.” You don’t speed up a run so that the music lines up. You just find a new song, cut the song off, or add text at the end. Wait! He did add text at the end. Why not just make the text there a little shorter, or don't play the song during the fail at the start? YOU CANNOT SPEED UP A SPEEDRUN! There is NO excuse for speeding up a speedrun. If you have to for a real reason, like the video being too long, speed it up by a clean number, and tell everyone that you did and why you did it. You cannot submit it with the sped up time, submit it with the non sped up time.

9. Performance Discrepancies in Other Content

• In his Viewer Levels series, his gameplay is good but not WR-level.
• Even when grinding fun levels, his runs are ok but not spectacular.
• If you showed me Michael’s gameplay in his Viewer Levels series and told me that the same person was capable of getting a 3:50 in a full game run, I would tell you to stop pulling my leg. This level of gameplay is nowhere near the level displayed in his world record.
• Michael said “9.I don't get 3:50 speedrun performances in viewer levels. I never practiced them, they are simply 'first glance'. Appel's levels are ones i played hundreds if not thousands. Impossible alignment and false mirroring.”
• If you can get a 15.4 level 5 in a full game run, you are able to sidejump quickly and consistently, not like this: This Viewer Level Got Me Angry(built in timestamp) I didn’t cherry pick this, it was just the first sidejumping I found.

10. Use of Interpolation Mode

• Interpolation makes frame-perfect tricks less consistent.
• No other top player uses it due to its drawbacks.

• Michael said “read the turbowarp description or javascripting - does not affect screen due to imput, only VISUAL DISCREPANCIES. (original wordings i forgot)”
• Yes, and the “VISUAL DISCREPANCIES” mess me up when I play as it delays everything by 1/30th of a second so I cannot react as quickly.

11. Contradictions About Coding Knowledge

• Claims to know nothing about coding.
• Yet, when accused, provides complex technical justifications for differences in his game version.
• It seems like Michael is just throwing around big words to seem like he knows what he’s talking about, even though he doesn't.
• Michael said “Simply put - my absurd appel ascertainings were done by random scratchers that commented. My coding is good, but doesn't mean anything.”
• I do not understand how this has anything with my point. Also, nobody has ever come out saying that they played a level for Michael, and each level has a completion video come out the same day with michael implying that he did these runs.

12. Failure to Record Grinding Sessions

• Despite being accused for over a year, Michael has never recorded himself grinding.
• He could have streamed or uploaded full attempts but chose not to.
• If his runs were TAS or spliced, recording a grind session would be impossible.
• If he was legitimate, he would have no trouble proving himself by posting a grind session, and would even be happy to clear his name.
• Michael said “FAIL. MONTAGES.” As said in #4, the fail montage is from his grind from his 4:18, not from his sub 4.

13. Unbelievable Instant Level Performance

• Claims to have beaten The Instant, yet never found the obvious cheese strategy.
• Any skilled Kaizo player would have found this trick quickly.
• To this day, even the widely agreed on best kaizo player, Krackerz, has not beaten The Instant cheeseless.
• Michael said “Yeah, the instant has been beaten by others~ Plus, some have succeeded in doing it in segments. Just because you subjectively believe it to be impossible doesn't mean anything.”
• I do not believe it to be impossible, I just think that nobody is good enough to beat it cheeseless yet. Also, The Instant has not been beaten cheeseless by anyone yet.

15. Evidence of Video Speed Manipulation

• Comparing Michael’s 4:06 to my 4:08 side by side shows his video is sped up, but his timer isn’t.
• Fastest possible Level 1 time is 18.93, yet Michael finishes at 18.0.
• This suggests a speed increase of 1.06x. (18.93/18.0)
• Correcting for this, his actual time would be 4:20.76.
• More detailed explanation on how I did this here: https://www.speedrun.com/appel/runs/yl62v0ny

Side-by-Side Video Comparison

• Micheal said “Again, audio issues. Disputed run as well.”
• See #8

16. Unrealistic Level 5 Performance

• I tried matching Michael’s 15.4 pace in Level 5 without waiting for the optimal cycle and it took nearly 20 minutes.
• Somehow he managed to get this in a full game run?
• I hold the IL WR for level 5 and tied the TAS, so I would consider myself good at level 5.
• Maybe Michael is just better at level 5 than I am, but if he is, why did he need to cheat his 15.2?
• Micheal said “15.2 is legitimate, and is even on the leaderboard before scratch removed it - that's also another proof, somehow, to me not TASing runs. Plus, your perspective is just simply harrowingly subjective and you don't even realise it.”
• The 15.2 is not. Proof is in #1.
• The history of the cloud scores are in the remixes, but there are no remixes with a 15.2. This is a lie.
• My perspective is harrowingly subjective because I do not like it when a person who can’t even get a run faster than 4:18 pretends that he can get a 3:50.
  

17. Changing his level plans after The Instant. (Written by OrangeFruit)

• After beating The Instant Michael said he won’t make levels harder than 1 bomb, likely due to fake proving him not cheating since he can beat bomb-levels (presumably).
• Both programminguy and I have tried to get replay codes from Michael for KTS and The Instant, but have been turned down.

18. Absurd Appel speedruns. (Written by OrangeFruit)

• Michael states that he doesn’t submit his Absurd Appel speedruns because they’re “too good” or in reality it might be that he doesn’t want to share his replays.
• In one of his videos he says that Centyme beat his Absurd Appel level 4 speedrun, and says “this is going to be hard to beat” then proceeds to beat it in a few attempts.

19. The Instant Cheese. (Written by OrangeFruit)

• When I asked him about how he did not find cheese in the instant he stated that he has “creator blindness”.
  

I made over 70 levels in Appel and over half of them are difficult, but somehow, they weren’t cheesed, even if the cheese wasn’t minor. I always playtest my level and find cheese, and I’m a worse player (presumably), after I finish a level I check it for cheese again before posting it. Somehow Michael doesn’t. He also stated that “makes everything flying and it looks cool” before that in his video on The Instant, he says that he “didn’t just wake up one day and decide to make a top 1” he mentions that he took a long time to make the level. And somehow he didn’t notice huge cheese in his level.

• He also says that he won't change anything, and I have no idea why not. Does he not want his level to be top 1? It may have even been harder than the new top 1, Among the Stars, and he’d keep his level at number 1, but for some reason he chose not to, also stating that he doesn't decheese levels, even though he did it twice.

Final Thoughts

The controversy surrounding Michael's Appel speedruns raises numerous red flags pointing toward potential TAS usage and dishonesty. From the suspicious variable in his 15.4 Level 5 run resembling TAS input displays, to identical movement patterns with known TAS runs and questionable route choices that mimic tool-assisted strategies, Michael’s explanations consistently fall short under scrutiny. His inconsistent statements about automation, rapid improvement speed, and contradictory technical claims further damage his credibility, especially when paired with a lack of full-session uploads and evasive responses. Despite prior cheating history and a noticeable gap in skill between his WR runs and casual gameplay content, Michael continues to defend himself with emotional appeals rather than concrete evidence. All of this casts serious doubt on the legitimacy of his records and calls for a more thorough verification of his runs.