I'd love to hear people's honest feedback about this video, how it could be improved, and inspiration/ideas about future videos. discord.gg/pcYs2tHTCE I also want to clear up a few things about this video: 1. Some people might interpret the intro as an attack against the mods or that the mods are covering things up. That is not what I meant to imply and the mods do a great job as volunteers to verify runs. I would also note that they are the best people to determine what the proof standards should be, and increasing the proof standards to demand things like handcam might just not be viable for something as niche as speedrunning. That's perfectly understandable. 2. "Just use a hash/checksum to verify the ROM". The problem with this is that it just shifts the problem to having to verify the hash. How do you stop someone from just providing a fake hash? Any program that outputs the hash could be modded to spoof it. 3. "This is old, it's just the same as ROM hacking". This is mostly true, and it's also true to say that these cheats could've been done many years ago without the decompilation. But in recent years, there has been a rise of decompilation projects across many classic games, which in my opinion makes this method a lot more relevant. It's now possible for any programmer to target any part of the game and no reverse-engineering skills are required. I also just think it's interesting to show off these cheats and demonstrate what is possible - a lot of people weren't aware of the possibilities and how these mods can be targeted at speedrun strats. 4: "It's the same as just using macros". This is totally false for a number of reasons. For one, a macro does not have access to the state of the game/reading the game's RAM. To take the BLJ cheat in this video as an example, it adjusts the mash speed based on Mario's velocity, which makes the BLJ much more consistent. A macro cannot do that because it cannot react to the live state of the game. Also, a macro plays back the exact same input sequence every time. This means that it would quickly become obvious because you would have the same exact input sequence across multiple runs. On the other hand, the types of cheats in this video can dynamically adjust to the live game and so you still get a unique run/attempt every time (very useful for streamers who need to farm hours of content worth of attempts). 5. "A better method is to just create a TAS". Creating a TAS is a time-consuming process and you would need to individually pre-prepare each TAS run behind the scenes. You would also have to completely fake your reactions and that you're playing live. On the other hand, the cheats in this video seamlessly integrate with your normal gameplay. You can still mostly play as normal and farm hours of attempts, but simply gain the advantage that boosts your gameplay to the next level. You still get a unique run every time, unlike a TAS which you could only use once and then have to keep creating more. So these types of cheats work far better with the modern "live stream meta" of speedrunning. I would also remind people that the cheats shown in this video are only an example. There are countless possibilities with this type of method across many different classic games. You can probably think of some for your favourite speedrun game to target specific glitches/chokepoints. And these cheats are increasingly easier to do as many of these older games are getting decompilation projects (not to criticise those decomp projects, they are awesome and I have contributed to some myself).
A macro can most definitely read the games ram. smwcentral even has some ram addresses mapped for the public. If you are savage enough to change the mario 64 source code and recompile it you can probably make a macro by reading game ram.
I think we have much different definition of the word macro if you think that macros can read RAM. Generally a macro is just understood to be a pre-recorded button sequence. Also, all the serious Mario players play on the original N64 console, not emulators anyway.
Carpetless difficulty is a little overstated/inaccurate (can wallkick on different frames, pause early and adjust, it's 30fps game so frame perfect isn't as hard). But it certainly is a very dangerous/costly strat, especially with where it's located in the run. I guess at some point certain games need input display or something to handle this stuff. There's really no limit to the extent people can go to if they really want to. We're lucky that the cross section of people knowledgeable enough to do advanced cheating, and people who actually want to seems to be pretty small. But every now and then...
the cheats can be built to show an input display that would make the vanilla game show exactly what is written on screen. There is no fix to this problem except for runners gathering evidence of their runs being legit. Similar to anti-cheating measures in online chess tournaments. Players have microphone open, a camera shows all of their screens as well as their hands moving the mouse. Not all runners can afford webcams I guess
a handcam requirement would probably work better, I bet this could be repurposed to work on a modded version of retrospy, where the retrospy itself would be changing the inputs, thus showing the cheated inputs through the input display and also bypassing the need for a hacked rom entirely.
i've been saying since 2014 that i could just make a tas input hack (that maybe even has 20 different playbacks per star,?) and just get world record like that. maybe even put a "press L to let Mario take the wheel" button that just does one star really well for you. it'd be trivial to do this for any game honestly.
@@TheGaming100 if you just prerecorded a good version of the star you did earlier, there would be absolutely nothing to point to as it would just be a recording of you doing the star legit
As feather ruffling as this video may be to the community, nothing has ever been solved through ignorance and knowledge suppression. I hope this video is the first step to finding a solution to detect hacked runs.
Agreed. Not to mention, if Retro knows about this there's a good possibility cheaters already know as well. It's more important than the community be made aware what cheaters are potentially up to.
Thing is even most modern games with online capabilities and anti-cheat still have cheaters. So it's near impossible to fight those same cheaters when the game has zero anti-cheat and zero tools for moderators to verify a run wasn't cheated. It sucks but it's a never ending war imo.
I disagree. If you could make a Thermonuclear device capable of creating an explosion which can destroy an entire city, simply by using 3 liters of coke, 2 onions, and 1 liter of chocolate milk...And only 3 dozen people in the world knew how to do it...Would you come out and tell everyone? What you say is wishful thinking about everyone being decent and completely ignoring mental illness, depression, and a lot of things. There are bad people, there are mentally ill people, there are people who crave attention they do not earn. I do not see how this information does any good to the community. The people moderating and maintaining the communities are the only ones who could make good use of this.
You should always show the cheats and hacks you guys find. ALWAYS. Never hide them or keep it secret. We can't develop ways to detect or know what to look for if we are kept in the dark. It's easier to stay alive when you know where the sniper is watching from.
So one thing i find interesting you point out at the end of the video is that established runners are ones to be less worried about, but in my experience of seeing cheaters exposed its often is some of the very top runners who are legitimately good at the game that end up also cheating(for a mix of reasons). So i dont think just because a person can show up to a live performance and get a close wr should prevent them from still being under scrutiny, the pressure to stay at the top can corrupt a lot of people.
Totally agree. Someone sitting in the top 10 but never quite hitting the record might say "I can hit the BLJ/throws 90% of the time, why should I lose WR over some BS" or even "I'm just going to have this hack as a backup for practice runs so I can get my nerves down". You reach the point where getting a record isn't about whether you can do it, but about how many hundreds or thousands of hours of your life you have to burn repeating the same actions to get it, and the temptation gets pretty high.
I was about to comment this, JohnDoe360 will raise all sorts of red flags, but established runners who just need that one perfect run, but have been denied it due to one or two mishaps would be very tempted to cheat.
If you want a great example of this, look up "The Story of the Greatest Rocket Jumper of All Time". That's exactly what happened. If he wasn't stupid, he never would of got caught. Really makes you wonder if there's pros that aren't dumb as rocks and just have some fool proof cheating that can't be found out unless they told us. They'd just take it to their graves.
Exactly joe shmoes cheating means nothing and its always super blatant the most prominent cheaters in competitive gaming were always good players who cheated and were good at cheating as well and sometimes they started theyre comp career with cheats and sometimes they get corrupted
I think it's a good that these types of cheating methods are out there. Rasing awareness is the first part of the battle. If people think this truly is getting out of hand, we might need to do what Pokemon red/blue speed runners do and only allow a very specific emulator of the game that must need to be examined for changes.
I'd say requiring the use of a hand-cam like SMB runs have, an input capture device like a RetroSpy (although runners might object if it'd make switching between OEM and HORI controllers more difficult), or require some sort of tool that verifies the checksum of the version of the game they're running with an ED64 or SC64 could all be reliable options for helping to verify the run if things reached that point (although I'm sure requiring a checksum every run would be going a bit far given the tools needed and if people even know how to do so).
@@renerpho custom firmware for the ED64 to just pretend it's doing the checksum correctly would be trivial at the point someone is compiling their own custom rom to cheat. handcam is currently beyond reasonable to fake and is likely the only way to deal with this currently. If hand-cam falls then the only resort will be live in person runs on validated hardware, being generous 20 years maximum until that happens (assuming speedrunning persists that long)
@@Jalaeeven handcam is only reliable within ~1 frame at 60 fps- it will catch anything blatant, but if a program checks for slightly early inputs and adds a frame of delay, it would be nearly undetectable. Means you need to be very nearly frame perfect anyway, but most cheaters who make big news are already very decent players in their own right
This can apply to any make-or-break trick in any game that has a decomp project. Pretty huge, and good to keep in mind when watching suspicious runners. I hope speedrunning is small (and honest) enough that anything like the CS:GO private cheats market can't flourish enough to be worth existing, but you never know. Thanks for always uploading quality.
It would have been possible before the decomp project. People have been modifying game code in ROMs for decades, the decomp just makes the process easier.
There have always been cheaters in speedrunning. The easier it is to cheat the more cheaters there will be. Many online multiplayer games are ruined by cheating and the same can be said for speedrunning as well. I think it's best to keep gaming as a casual affair for this and many other reasons. Gaming is a meaningless endeavor anyway and to spend so much energy to become good at a game is pathetic to say the least.
Yes, I setup a cheat like this for Minecraft speed runs during the dream drama. I knew it was undetectable, but since I’m not a cheater, I didn’t actually use it. Basically I patched the raw Minecraft binary and that allowed me to fake not having mods installed. The mods were compiled into the game itself :) Edit: oh yeah, and mc has a check that the game is modded that displays in the corner, but since I had the whole source code, I just turned that feature off
The biggest driver behind the CS:GO and CS2 private cheat markets are the skins having financial value, along with the streamers, esports pros & others who make money from the games - people are incentivised to make more subtle, less detectable cheats. It's not to say there aren't cheaters in speedrunning, but it's traditionally been more about video manipulation of runs by splicing & editing with some unauthorised mods to improve things like RNG thrown into the mix... fortunately when the incentive is only bragging rights, cheating tends to be less pervasive.
Yeah anyone with (deep) knowledge of MIPS disassembly could've done this before but this has certainly lowered the potential barrier to cheating. Obviously, this is all a moot point if someone doesn't play on original hardware as there are no guarantees there.
what the Game Gennie back in the day did was effectively "code injection" these de-compilation steps just inject that code right into the binary running on the carriage, and the only way to maybe detect them would be to do some kind of check-sum-hash, but that would require an interface of "after each run insert cartridge into this, and if the test fails then tampered" (almost like testing performance enhancers after sporting events...)
this video is very appreciated! as someone with a strong interest in speedrunning this is a very strong case for, especially decompiled games absolutely requiring input display/handcam at top level. but personally i also think that speedrunning communities as a whole have done a pretty good job at enforcing high standards for their players. A system of increasing standard enforcement proportionally to placement/time not too dissimilar to how SMB1 does it is probably enough to dissuade people from using mods to hack their way up the list until suddenly they have to show handcam/input display and have to hit the tricks they've been cheating on to this point. the fact you're playing on modded software is undetectable, but if you need to show that you actually hit the tricks in your run with input display/handcam while playing live, something which you've shown is possible and part of the insidiousness of the program, your run is now... as difficult as a non-modded run. and if you forget to manually execute even one, the fact you *are* playing on modded software becomes instantly detectable.
Just adjust the code so it only executes the next step if a button is pressed roughly at the correct time. Now you still need to perform the correct inputs, but your timing can be way more relaxed.
"film your hands or it doesn't count" has always been in the back of my mind anyway. Both the tech itself and technical literacy is only going to make cheating more possible and varied in future so in my mind if you are spending hours on trying for records, you owe it to yourself to at least aim a cheap webcam at your hands to remove all doubt (including VS any future methods that might be revealed).
I can understand the "why" (has been installed from past times when running 2 cams was more expensive/technical, has made it the norm in those scenes)... These are small communities each with their own little meta. It's not like there's a central verification body with set methods of anti-cheat across all games that are speed-run (not advocating for one either, Guinness sucks, for example). I feel it's those games that have been stung by proven cheaters are more likely to introduce things like enforced hand cams. The smaller the community, the less likely there is to be someone with the technical knowledge and free time to act as police, therefore the simplest solution becomes the best, IMO... Problematically, I think there's a lot of smaller games with only a few runners where these players have the scene on lock, so change can be slow - until it becomes the norm across speedruns in general to have hand-cams and mics. The hacks are always one step ahead of the anti-hacks, not just in speedrunning!
It seems speedrunning has finally come around to be in the same position the previous generation of competitive gaming culture was in during the corrupt Twin Galaxies era. Establishing how the industry could ever be legitimately regulated seems a Herculean task requiring an astronomical degree of specialization to truly verify, that would also require a veritable army of those specialized to keep up with all the various categories in such a massive diversity of games. Mercy
Not really. Most people still understand how stupid cheating is and requiring a handcam or maybe even mic audio makes this method impossible. This video is just fearmongering for the sake of views.
@@quilllllyit does punish and price out poorer people for not having the money to establish this setup. While there's certainly more nuance than that, accessibility to the top of the leaderboards goes down majorly and discourages the average speedrunner for aiming so high
@@justinhansen6069 accessibility to the top of the leaderboards? In order to get your times on the main leaderboards you need an n64, a working controller, a copy of the game, a capture card, a device to record on eg a pc, and ideally a crt monitor and a second controller for throws. The leaderboards aren't accessible. A Webcam and microphone is the least of anyone's concerns
@@quillllly i gotta say u cooked with this reply, exactly what i was thinking. You’ve already spent way more money just getting access to the original console and game, not to even mention dropping like $100+ on a capture card, all before you even begin to think about webcam and mics lol. Plus, you can get a cheap webcam and mic for probably less than $50, so in actuality that would probably be the cheapest part of a mario 64 speedrunning setup 😂
@@quillllly I strongly disagree this video is “fearmongering.” The rhetoric on that is clearly an exaggeration, and would imply this analysis brings up no facts whatsoever. I also agree with Justin Hansen. The existence of private experts who can help someone rig themselves a record for the right price, will have a socioeconomic effect on the industry. Saying that people “understand how stupid cheating is” is deeply naive. The bigger the industry grows-and it is certainly huge now; much, much bigger than competitive gaming ever was from the 1980s to 2000s-the more the risk of cheating becomes one that many are willing to take. The problem with competition and so many financial incentives to get to the top is that many people be willing to do whatever it takes to get there, including cheating. Having a robust system of regulation and being able to objectively verify people are following the rules it goes without saying is extremely necessary. To return to my original point, it is starting to seem like it will be immensely difficult to really verify and say confidently that the industry will be free from manipulation going forward. If you don’t believe me on the propensity for power and an in-group to manipulate an industry, simply look into the history of Twin Galaxies and how much insider manipulation was going on when Walter Day was the head of the organization.
I think at this point it would mean all officially submitted speedruns would need to require at the very least an input reader visible on screen or a hand-cam
Adding a buffer to inputs for roll outs (and maybe elsewhere) would be something very useful and very difficult to detect, even with a hand cam...for that a hand cam AND a synced input reader would be needed... but for top top level runs it might be needed.
Make it a requirement for world record attempts lmao. If your just going for PB's or like top 10's who cares that's way to many people to force into using this stuff. But if you force world record holders or that level of player it's a much smaller amount and a more reasonable request.
I think these could be detected by requiring an input and controller display for people with WR level times, as well as requiring a longer video with a full session of runs. For a game like SM64 where top runners have practiced and grinded for thousands of hours, by the time you have that level of dedication to the game you should be willing to provide this kind of proof.
TLDR: Using code injection or modified game-code, it's possible to add cheats that are invisible to the outsider, since the game appears to run normally. This is similar to adjusting drop rates, like Dream did in Minecraft and, indeed, MUCH harder to detect.
@@JeffGoldblum64 My brother in christ, that is literally every mod in existance. I am currently for fun speedrunning the 1st level of ultrakill against my friend. He uses Bomb movement tech, I dont. I beat him now to 33.3 and am 3 seconds ahead of him. Neither of us modded the game here. But the best part is: I actually cant say for sure that he didnt. And I cant say for sure that any speedrunner modded their game. Cause a mod by nature isnt supposed to be detectable unless it quite literally adds things to the game. But there are a lot of mods that simply enhance FPS, unlock FPS even, overhauls of stats and more. There is a TON of stuff in the modding scene that you cant detect with an untrained eye.
requiring audio from controller inputs (which seems to already be done for the most part) and (maybe handcam) could be a good solution to catch most of these
While I agree with that given how they're implemented, I don't think it will be too long until these kinds of hacks either ignore your inputs (so you can mash / spin and it plays the "right" inputs behind the scenes) or even simply buffers them so a 1 frame window becomes a 3 frame window, making these incredibly hard to detect even with a hand cam / audio
I understand the danger of showing cheats but the community being aware of the possibilities is a good thing. Credibility of the runner and having a camera showing the controller should be the new standard. Also live streaming attempts would make it harder to consistently cheat and not get caught.
Only for the top level. Holding the bottom of the leaderboard to the same standard as the top is crazy and kills speedgames. If you're good enough at SM64 to get the world record, you've spent long enough at it that this stuff is a pittance.
I'm sensing a SpaceUK type situation in the future. Someone shows up out of the blue and immediately makes it to top 50. Soon after that, top 25. Then 20, 15, 10, runner-up. A few people are skeptical, so they make claims of cheating. But since this guy streams his runs and his gameplay seems legit, no one bats an eye as he finally makes it to the top. And as our champ keeps improving, time barriers are simply destroyed in matters of weeks... if not just days. However on one fateful day, a top runner will catch something peculiar. His button presses have seemingly inconsistent delays compared to the gameplay. A moderator is contacted and an investigation begins. It doesn't take long before a pattern is found. Every PB that this "champion" gets has a random C-UP input on the intro text box. Turns out, he was using a modified copy that basically turned top competition into child's play. One seemingly random button tap turns Mario into a TAS. This guy wasn't a champ, but rather a SCAM! And now, because of his disgraceful actions, we may not trust any up and coming top runner for a long time.
I think the amount of trust that's historically been put into just recordings is insanely high. There have already been many examples of simple splices being almost impossible to detect when done right. I think that official records will at some point have to be done in front of some sort of an event for it to count as a wr, similar to rubix cubes.
You’re completely right, and that wouldn’t be a problem if that were the only way to "officially" measure records. The most popular speedrun leaderboards are by and large unofficial anyway.
that would be incredibly unfair though. there are people who can't travel for a multitude of reasons. speedrunning should not be only for rich able people.
Iunno who you are, nor am I a serious gamer, let alone a speedrunner. I also don't know how this video ended up in my feed, but I thoroughly enjoyed this. Very well executed, even to those of us who aren't involved in the community. Have a sub my dude, I'm interested in seeing more of this type of stuff.
While not speedrunning, a Melee player back in 2016 was caught using a modified version of the game to win tournaments. His version of the ISO had been modified so that you could enable a bunch of major buffs for the character Pichu if you selected Pichu with a specific outfit while holding a specific controller input. He was eventually caught because the buffs were too obvious and he behaved suspiciously by refusing to play on any set-up other than his own, prompting others to secretly copy his copy of the game so they could examine it.
Secretly copy of his game? Dude I know that story and the people who exposed himself did bad stuff to him by stealing his sd card and revealing personal info.
@@NumbHydro I'm not saying eye for and eye is the right way to go about it. That will leave everyone blind. That being said, dude deserved getting served in one way or another.
seeing cheats like this breaks my heart but on the other hand , i am glad decomp exists and i guess in a technical level this is showing me things that can be done (when doing it for good reasons of course , not for cheating)
I’ve been a dev for 25+ years. Seeing this brings back an idea I had to build a live handcam generator based on a prerecorded sample set. Because of the low resolution and frame rate used on handcams, you would only need to morph 2-4 partial low quality frames per button press to create a fluid, photographically unique gesture. With a few seconds delay, you could “live” stream with audience interaction and nobody would know. My point is handcams need to have higher resolution and frame rate to be safe. That will not be enough soon.
It's an interesting video, but honestly I think you're maybe exaggerating a bit that people will be mad you've "revealed" it. While the full decomp does make editing the code here significantly easier than it would otherwise be, editing any game to give desired results is almost as old as gaming itself, and this isn't the first time such a method has been abuse-able. Speedrunning has literally always required some degree of an honor system. Once upon a time it was rare for people to even provide footage at all. I don't feel like anything has changed. I mean, the fact you literally wrote out the code and showed it for the whole internet does take literally any skill or knowledge out of the process of successfully cheating now, so maybe in that respect one could argue this problem has now been heightened for mario 64 specifically. But honestly I think the people who would want to do such a thing would have eventually figured it out just the same.
That's what I was thinking. With learning assembly you can pretty much do whatever you want in terms of editing, you don't even need a decompilation, so this isn't limited to just SM64. ROM hacking has been a thing since forever, so when they revealed game modification as if it were a huge discovery that nobody knows about I chuckled a bit.
I do think this video should be deleted. Nothing is gained by drawing attention to the dark underbelly of speedrunning like this. It will only result in more people cheating, thus hurting speedrun communities.
@@TheFinalChapters People with the coding knowledge to make mods like this were certainly already aware of the cheating possibilities the decomp had to offer. In my opinion all this video does is inform more casual viewers of a possible leak in the speedrun scene, and “drawing attention” to potential cheating can only lead to positive change, right?
These videos bring joy to my heart whenever I see them. Every time a new speedrun cheat method is discovered, it becomes harder and harder to verify a "real" run. The hobby started off as just that, with just gamers seeing how fast they could beat the game, but it's become so competitive and serious that it has lost its soul. Maybe when all the top speedruns are fake people will come to their senses and realize that they've wasted their lives by speedrunning.
AFAIK there is on most emulators and they are required. Helps show "too perfect" stuff like a TAS but doesn't help when it's not too perfect, but very close. That's why he talks about cred in live events. Like bluescuti could have faked his beating of tetris but we can be confident is it real because he has played live with random seeds and shows the ability of doing that given a reasonable amount of chances. If I say I can beat metroid in 4 minutes (I didn't look at the record don't kill me) and never played live only have a bunch of runs without being live (with some randomness to prove legitness) or showing the controller it would be sus.
just watched this after watching your cod cheating video. You make great content bro. I would love to see more deep dives in to cheating and can't wait to see your channel grow.
This is true but their are exceptions. I guess it would depend how consistent a player might be livestreaming and recording. Personally, I always do better when doing offline attempts because streaming makes me nervous and I'm distracted by chat.
I mean, maybe they would be caught eventually, but initially people would just think the person was a good player and praise how consistent they are. I have absolutely heard players get praised for having very consistent BLJs. (I don't have any reason to think any of those players are cheating, of course, but I've definitely heard that kind of praise tossed around).
Using this video as an example, they could just do the BLJs normally most of the time then hold the button when they're on a good pace, taking the difficulty out when it matters
@@skeletonorchestra2900 Most cheaters from my experience don't tend to balance things out well enough to avoid being caught by making statistical anomalies.
Speed running has always been full of cheaters and nowadays it's so available and easy to do that if communities want to eliminate it they will need to enhance the requirements for a valid speed run. There are simple things they can do to prevent any possible cheating.
I'm not a runner but a few ideas for potential requirements: ROM Checksum Video of controller to verify inputs I know you don't like adding cost but an inline input recorder could be made and distributed so inputs can be played back (like DOOM). I know there are aspects of all these in various games already.
Any verification device can itself be modified to cheat as well. Input recorder works in theory, but requires players to buy an extra, third-party device, and even legit runs won't always play back correctly due to subtle differences between consoles/cartridges. If one console's CPU is running at 93.0000007 MHz and another's is running at 93.0000002 MHz, there's no perceptible difference, but it can change RNG, timing between chips, etc. This slight variance is normal; every console will have a few Hz difference, even brand new. Look into Super Metroid TASes for more about that. They had to modify a console to restore it to precise factory specs, because it was old enough that the timing crystals were slightly off. That didn't affect normal gameplay at all, but meant occasionally the main CPU would have to wait slightly longer for the audio CPU to finish a task, and that was enough to introduce a lag frame and throw off the whole TAS.
Changing source code just would require a checksum hash to verify or key logger of inputs. That does get into the range of how much you need to trust/distrust someone in a hobby space
@@reallifeistoflat also im talking about the n64, which is what most top speedrunners are playing on. who in the hell is going through the trouble to have the cartridge just to play it in an emulator? and also that doesn't stop anyone from inserting their own code.
@@JeffGoldblum64 if this is something the community thinks is valid you could make it a requirement for wr validation to checksum on camera after the run. I believe runs like Minecraft require you to show your last few world generations, i might be wrong about the specifics. It's just another step that might be required for high level. Would also be easy for in person events.
I'd say a pretty decent standard would be to require input detection overlayed over the game whenever possible, and have a hand cam showing the real world controller inputs as well. This way not only can moderators see what inputs are being put in by the player to verify that it's all legit with what's happening on screen, but this can also be compared to the hand cam to confirm that the same inputs are what's being put into the physical controller the player is using at the same time. It would be very hard to, for example, simulate a fake BLJ while still having to do the correct inputs anyway. If the player is just holding down the A button while doing BLJ's or just holding the analog stick straight up during the Bowser fight, it's a clear tell that they're playing a modified version of the game that gives them an advantage. Just these requirements alone would do a lot to make this kind of cheat much more difficult to pull off.
These cheat codes add additional CPU instructions, so they should be detectable, but not in a 30 fps recording. I think, speedrunners should provide an input trace which can be replayed and would always lead to the same result (as long as the RNG seeds is also exactly the same).
These would be adding a handful of microseconds here and there. You couldn't detect it by frame counting, because those few extra cycles just mean it spends a few less microseconds waiting for the frame to finish. Also, you could craft your code carefully - or only change number constants - to take the exact same length.
One another thing that could verify them is somehow verifying the checksum of the cartridge. Even if a single byte is changed, that will dramatically change the checksum. And using multiple checksums (especially modern ones) will make brute-force attack of them less viable.
We know the checksum of a legit rom. Surely someone with the knowledge to make these types of hacks will have no issue feeding whatever device you're using to determine a checksum the real rom info.
@@thomasmathew13 Good luck implementing a device that checks the rom for legitimacy in the official rules as a required item (That you would presumably need to purchase), it would kill the community almost instantly
Damn man, this hurts to see, although I don’t find this kind of thing that surprising given how advanced and how far people will go with this kind of thing. You did good by exposing this, don’t feel bad for making this known.
The best solution I can think of, is to add some sort of secret watermark that's not well known. But like, maybe a certain thing in the game is sized differently but not easy to notice unless you know. Examples could be changing the color of something to a slightly different color. Or changing something's size by just a few pixels. something that you can tell if you're looking for it but it's invisible otherwise
Your idea will only work for "script kiddies". Adding something to original game is impossible, on the other hand, adding something to public sm64 source code is... public. So cheater can remove that change simply by taking older version of source code.
It’s called hashing. Having a known valid digest value (the result of a hash) means that we can show whether or not a game has been modified. Hashes provide unique values for a set of data, aka “a fingerprint”
@@cmspteam it's not that it's only for script kiddies, it's just something that has to be kept secret. Obviously it's too late to do so now, but it's just the only solution I have.
@@soybeanrice I know hashes, but how would we confirm? "Hey, send us the hash of your game with every run!" All they'd do is send a hash made by the real game while they run a separate modified copy. I think it should be something like, if you do an emulator run, you have to use this specific cheat that generates something like a hash based on the code, current room, and playtime or something and have it display somewhere discreet on screen. That way someone should be able to recreate the run and have the exact same hash. Although someone dedicated could probably make it run on an N64 and say they're playing an official copy so they couldn't have modified it and can't run the hashing code. Therefore I'd say my original idea is still better, have something slightly off that's hard to notice. Say the git had some issues and they had to remake it, so they just reupload the code with the slight differences that'll be dead giveaway for mods
@@collinlawson713You bring up very good questions that the community will need to address at some point. I have ideas. However, the point of the video is that code modification is "UNDETECTABLE" and to that I strongly disagree.
The fact you can make it an n64 cartridge... this is tragic. You absolutely had to show this though, otherwise cheaters would've abused it under the radar and no one would know this was even possible.
Makes you think. Someone could bring their own cartridge or controller to a live event and get a record or victory. At least I can brace for the inevitable drama.
Or you could modify the console itself. Overclock, or have it read from an embedded ROM instead of the actual cartridge... end of the day, there are a million ways to cheat.
As scary as this is I think a lot of it could make really sick bonus categories. Auto blj and spins etc are really cool in terms of how fun they look lol
While these are interesting ways to cheat using the decompilation, the same could've been easily accomplished by using macros, turbo button, or even just a gameshark/AR code. In fact, using macros would be the more reliable way to cheat than modifying the code directly. Taking a look at your first example with the BLJ, you could easily just use a turbo button and hold down A instead of making the game automatically input it for you. Using a turbo button would have a strong benefit over the game itself, as input viewers would show that the A button is repeatedly being spammed instead of being held (with the latter obviously meaning there's a cheat going on somewhere). Your second example doesn't even need the decompilation. It's literally just a simple ASM mov instruction (or setting the value to the max (aka a gameshark/AR code will easily do it.)) It also suffers from the exact drawbacks as the first example, where the input viewer will show that the stick isn't being spun fast enough to match up with Bowser's spin speed. A macro with the analog stick being set to spin super fast and the cheater just needing to time the B button would look more legit on an input viewer. Finally, for your third example, it suffers from the same thing as the first two did. It's overly complicated for something that any simple input viewer will clearly show that its cheated. A far better way of cheating that doesn't even need the decompilation at all with the bunch of checks would be to just find a consistent setup strat, then using TAS tools, making a macro that repeats the same set of actions every time. This will also have the benefit of actually showing the inputs being pressed on the input viewer. Overall, using the decompilation project specifically to cheat is very inefficient. The drawbacks FAR outweigh the benefits: - Any simple input viewer that displays the button being pressed on the controller won't show buttons that are getting pressed by the game itself - Needing to set up an entire project, compiler, and w/e else to be able to create a ROM with your changes - Needing to know how to write C code and know where to edit to actually make the cheats - Most cheating changes can either be easily done with a gameshark/AR code and a macro or turbo button that'll show up on an input viewer In the end, this is a pretty niche way to cheat that'll end up as nothing more than a novelty (at least given the examples you've shown.) The only thing you could argue with using a hacked decomp over macro/turbo is that it's "cheaper", but given that to run it on console you need an Everdrive 64 (which is expensive as hell), it's not worth it.
Imagine you are streaming 5 hours of attempts. With the TAS approach you mention, you would be doing the trick (e.g. carpetless) in the exact same way every run because the input sequence would be the same. Therefore, it would be very obvious unless you created a bunch of possible permutations. You'd also have to manually switch the inputs between your controller and TASbot mid-run and ensure there are no TAS desyncs which is tricky, plus if there was a desync because you didn't start the TASbot at the exact right spot then it would look fishy. The decomp cheats can dynamically respond to any situation in the game and are much more flexible. Remember that what was shown in this video is only an example. Also, in regards to the BLJ, I guarantee using a macro would not work very well. It's not just about mashing at the fastest possible speed - there's a rhythm to it. In particular, the code shown in this video will mash with a 3-frame delay until Mario starts picking up speed, and then switches to a 4-frame delay. Of course, you need to be able to read the RAM to do things like that. So it's far more powerful than what a macro can do.
Maybe it's possible to come up with some way to enforce a ROM on all runners for verification purposes. After a run, someone can run something in the menu or otherwise that prints out all executable memory, logs it, hashes it against a legitimate ROM... is there a way to verify rom size through hardware? I'm not at all knowledgeable about N64 hardware or software, so this is far from a proposal, but if someone can find the right missing pieces there could be a way to substantially raise the bar for cheating through ROM manipulation...
Something like the above might be verifiable through counting clock cycles... if someone is trying to make a fake "good" output, maybe it'd be hard to get CPU time to align. Obviously, you would worry about someone reading values from a separate copy within the same ROM, but maybe there is some limitation in hardware that makes this more realistic, i.e. different memory fetch time for different parts of memory... someone might be able to do something with this kind of approach?
Seeing as it's a console game, it would be really hard to tie any give recording to a signature that says "yep, that's a legit sm64 cartridge." If it's a machine that you add to your N64, how do you know the recording was done on it. If you make GameShark code to do a checksum or hash on the game and show it on screen, how do you know it wasn't modded to spoof that?
@@herrabanani obviously I don't really know the implementation details, but obviously a checksum by itself wouldn't be enough. You could sleep for some amount of time and then display the correct one. If you could display some output as this is being calculated, maybe there isn't a way to spoof this *with the same amount of CPU clock cycles*. Or, if you were just printing out the executable memory, maybe there is a performance reason (maybe n64 carts ex mem is set at an absolute address. You need to therefore either have a copy of the entire game [which might not be possible due to size constraints, or otherwise inducing performance issues, or fails an OS check for size] or conditionally read memory upon verification, causing a performance drop [once you get to this address, stop calculating] OR maybe you could edit your own executable memory real time [must reload your rom every run!!!]) why this is hard to deal with. Anyways, to reiterate, I'm just spitballing here.
@@xininja8167any potential check could be spoofed. It's the equivalent of having a green checkmark on screen if your cartradge is legit. If you can't trust the game is legit, how can you trust that the checker is legit? Not to mention having to require every player to modify their hardware to check for the modification with the use of everdrive carts or game sharks. Imo the best way to legitimize runs is to require people to show hand cams and stream/record all of their run attempts. That isn't perfect either but it's the best you can do imo.
It's starting to feel like we need cameras (with audio) on everyone's actual hands, for every game, which kind of sucks because this is an extra barrier to speedrunning. It would be interesting if there was an automatic way to cross check the inputs as detected on camera, vs the inputs in the game
You wouldn't need this for every runner - just at moderator discretion. Someone in the top 5 cheating is obviously much worse than someone who's ranked 350. For larger communities like mario 64, the ability for top runners to make enough money off of twitch donations to get a setup like this is almost certain - and if for some reason it's not, some sort of crowdfunding system would likely be possible. Only issues are for runners that are legitimate, but are generally disliked for reasons outside of the game, or for games that have no larger audience to fund their speedrunners - but the number of games like that which have decomp projects are likely quite low.
@@peter-rhodes True but a lot of times people need their phone for other things while running. Buying two cameras, a decent mic, and having a pretty good PC does add up
i think your perspective on what to reveal / what not to reveal is a bit odd. carpetless is op yes, but its also hard. some rando with a hacked ROM getting carpetless out of nowhere is gonna raise lots of suspicion, but a rando getting good BLJs / good bowser throws wont nearly draw the same suspicion
the cheats aren't for low level players but top level players. I believe it was karl jobst that said speed runners cheat to get a pb faster when dream's cheated minecraft speed run came out. Meaning its for some one who already had a good pb and knows they have the skills to shave 1 second or half a second off of their pb but they get impatient with them selves at their minor time losses.
@@anacreon212 cheats are cheats, people will use them no matter what. its more of a matter of what is easy to spot from a top player level that matters, top players can get perfect bljs multiple times every time they play, but almost nobody can get carpetless without a single mistake. so using a cheat that gets carpetless perfect is gonna look way more sus than a cheat that gives you perfect bljs / owless / similar strats that top players have mostly consistent but up and coming players might not another thing with carpetless; since theres tons of inputs youd be able to see if multiple people are using the exact same inputs on the exact same frames, whereas a real player might buffer inputs, or have variances in the frames they hit inputs that arent frame perfect (of which I think carpetless has a few that are a few frame windows) another thing too, if someone uses the same cheat for carpetless twice in a stream, thats way easier to notice than it is if they use a perfect blj cheat
you have to spin your stick 8 times per second to get that throw and that's physically impossible for many people. Some rando getting it would raise suspicion. Also controllers are loud, you hear them. This wouldnt work
@@CatOnVenus183 problem is, it *is* possible for some as you mentioned. And a good mic can be set up to pick up only voice audio. So I'm not entirely sure it's that unfeasible
It's sort of expected that a highly popular game for speedrunning having its source code leaked like this would just turn into a cheater's haven. It's highly sought after, a nice quick run, in and out. I could've figured it would be like this as soon as I heard the game was decompiled. After all, like how speedruns and techniques evolve and refine, so too would cheats and other unsavory tactics. It's the natural flow of things. When the optimal bar is approached, it's only natural people looking to hit that peak would look into tools to better imitate that "perfect run" they seek. I personally think it's interesting, the idea of a Real Time TAS (Or otherwise modified). Depends on the rules of the game, but SM64 is unfortunately pretty cut and dry in that regard. Still, speedruns evolve, how it might evolve to handle this situation will be interesting to see (assuming it does find an answer).
Important distinction: no leak was involved. The code was reverse engineered from only the binary. (A leak _happened_ but that code was considered off-limits for legal reasons. Also I don't think it contained the actual code, only intermediate files?)
As cheaters progress, so do the non cheaters need to progress. Its a cat and mouse game. I wish i was technologically savvy enough to articulate this any better than how im about to, but the best example i can give is in super mario bros for the nes, you can play a modified version that displays your coordinates in the spot where the time remaining would have been, which is used for perfection of wall clipping. Using this same idea, it sounds like a mod of this game could be made that displays the code of the game's coordinates live as their playing it and if they are using a modification, you would be able to tell because of the automatic response of the game when certain buttons in concession are hit. Because over the span of an entire game, its highly likely that they would reveal themselves.
I think another precaution the speedrunner community should take is recording the inputs from the controller. these cheats rely on detecting inputs and making it easier for you to do these skips. if you were to be able to do a perfect bowser spin in the game but replayed through the inputs you weren't, that could qualify as you cheating.
@@Gamebuster hmmm... that's a tougher one... I would assume you'd use a Radio receiver or maybe a flipper zero to record all of the inputs. because you can also buy non-nintendo switch controllers online and connect them via Bluetooth, maybe it would be easrier to create a bugged controller that could record all of the inputs
The backwards longjump trick would require mashing the button to actually look like you're doing the technique legit though, and even then if your controller inputs shown on screen don't match up to the correct frame in relation to Mario's jumping you can still be found out.
so the Controller input UI on the screen reads instead what the game believes you pressed, including the auto-presses, instead of the controller. now its impossible to tell.
@@isodifbrakiul6387 How are you able to get that information from the game itself without using an emulator or some kind of hacked console? Most controller input displays I see read from the controller itself or the controller connection, particularly for something like N64 where you have to splice and tap into the cable wires and use a microcontroller to decode the signal.
You seem to have a very cynical take on this. I think you're forgetting that while people can advance in cheating, there are ways to advance anti-cheating. Showcasing this predicament was a good thing because it shows what verifiers need to be aware of. I still believe requiring handcams, input trackers, providing ROM hashes, etc, is egregious for a majority of speedrunners (in any game) and is definitely gatekeep-y, but providing it at the very top level (WR, top 10, maybe even top 25) is a compromise worth taking. Every game is different, of course, but the tone this video had is very doomsday-like and discouraging, and I don't think trying to scare casuals off is a good idea.
@@Espik_23 As you get better and better times, verification proof gets stricter and stricter. You may be expected to provide full session attempt videos, input files, face cam, hand cam, and more. Times below five minutes are treated with extreme suspicion, and you will need to show a lot of proof there, whereas slower runs are more relaxed. The idea is that casual speedruns will not require too much proof (a video of the run, maybe showing off the second quest) and by the time they demand the kitchen sink, you're already pushing into the upper echelon of times and can justify the investment.
Reminds me of how speedcubing is required to be done at events, since its too easy to cheat otherwise. This is almost certainly the direction speedruns are gonna go.
It's the end of history, speedrunning history. I mean I know several people that all think speedrunners are just cheaters anyway. It's kind of a poetic little bow on it if they are right and nobody can tell the difference.
I still just see speedrunning as a hobby which you will ruin for yourself if you cheat but the thing is that people specifically cheat for the potential fame which brings twitch viewers which brings money.
Hot take: cheating is as much a part of speedrunning as speedrunning itself. It also has done a ton for speedrunning’s popularity. The never ending quest to cheat runs and have others catch these cheaters is in my mind equivalent to fighting in hockey. It’s not supposed to be a part of the game, but it’s a huge part of the appeal.
From the very first seconds of the video when you took a retired leaderboard mod's words out of context, this video was shown to be in bad faith and some level of fearmongering. Obviously it's possible to modify the game's code or play back input to cheat, and the community has been aware of it for a while. But, making this sound like a conspiracy the mods are trying to stop from getting out is ridiculous. My personal experience with the SM64 mods is that they are very honest and intelligent people who have established verification standards and methods above and beyond what should be expected from a few volunteers in the community. The amount of work they do to verify everyone's ordinary runs is immense, and they have a very good track record of catching bad actors while keeping a low barrier of entry for legitimate players. Saying that "it's the right thing to do" to speculate and cast distrust about the whole community is nonsense, you're just making the mods' work harder and influencing people to think cheating is a much bigger problem than there is any evidence for. A few idiots could watch this video and go cheat low-level times without being caught, but if someone was trying to fake being a top player, it would absolutely throw up red flags across the community. Players take thousands of hours to reach an elite level, if ever, through dedication, struggle, and failure. People are accustomed to these patterns and can raise reasonable suspicion when there are outliers, at which point higher scrutiny or submission requirements (e.g., hand cam) could be considered. If the author really wanted to find out and present the truth, they would have at least reached out to the mods. The entire message of this video should be taken with a grain of salt.
I understand your perspective but I have to disagree with a few things here. Firstly, I don't really see your point about how the beginning quote indicates bad faith. It's a real quote and was not out of context - it was exactly within the context of me talking about making this video. It was also not meant to be a criticism of that mod or any other mod. I was simply trying to portray the difference in philosophy that I don't really believe in the "security by obscurity" principle (reliance on obscuring/hiding knowledge). I apologise for any possible interpretation about the mods being bad. But I did show this video to a number of people before releasing, and nobody gave me any reaction of any hostile interpretation towards the mods, and I have not seen any comments/concerns to that effect (except from comments like yours and from the mods themselves). Honestly it's hard to predict all possible interpretations of a video before it goes out but that does not seem to be a common interpretation from regular viewers. So, I don't disagree with your defence of the mods, but I wonder who exactly you are defending them against. The final point that you make, about the difficulty of faking being an elite player, is a very similar point that I make towards the end of the video. I made it clear towards the end that we have to rely on a runner's credibility, and that runners like cheese have demonstrated very high credibility. Finally, you mention I should have "at least reached out to the moderators" - where do you think I got the quote from at the beginning? I did reach out to a few mods, even offered them to be on the video for an interview. But they declined and did not seem too interested in talking beyond the quote I put at the beginning which you didn't like me showing.
@aijx4223 The quote at the beginning of the video that says that the lb mod doesn't want the information to be shared is bc it does not provide any benefit at all and only serves to harm the community. The dude's just being an asshat pretending that it's for the "moral dignity" of the leaderboard. He's not providing any information that anyone with more than 5 brain cells knew about and he's exactly going against the words of a former lb mod (retro lied for clickbait).
@LFoxyIsBlue If everyone with 5 brain cells knew about this, then why are you saying im harming the community and why are people mad at me for revealing this? Surely it would not harm the community if everyone already knows about this. And what did I "lie for clickbait" about?
Wouldn't the bowser throws cheat be pretty detectable? Like people are aware of the exact max speed and can tell if the speed of throws is constant or done by a human. Might be detectable. Very good video though. I was thinking the most likely undetectable cheating method would be changing RNG values through modding, if you didn't make it too obvious. You could even stream runs with it, granted you don't don't make the changes in RNG values too obvious.
I'd love to hear people's honest feedback about this video, how it could be improved, and inspiration/ideas about future videos. discord.gg/pcYs2tHTCE
I also want to clear up a few things about this video:
1. Some people might interpret the intro as an attack against the mods or that the mods are covering things up. That is not what I meant to imply and the mods do a great job as volunteers to verify runs. I would also note that they are the best people to determine what the proof standards should be, and increasing the proof standards to demand things like handcam might just not be viable for something as niche as speedrunning. That's perfectly understandable.
2. "Just use a hash/checksum to verify the ROM". The problem with this is that it just shifts the problem to having to verify the hash. How do you stop someone from just providing a fake hash? Any program that outputs the hash could be modded to spoof it.
3. "This is old, it's just the same as ROM hacking". This is mostly true, and it's also true to say that these cheats could've been done many years ago without the decompilation. But in recent years, there has been a rise of decompilation projects across many classic games, which in my opinion makes this method a lot more relevant. It's now possible for any programmer to target any part of the game and no reverse-engineering skills are required. I also just think it's interesting to show off these cheats and demonstrate what is possible - a lot of people weren't aware of the possibilities and how these mods can be targeted at speedrun strats.
4: "It's the same as just using macros". This is totally false for a number of reasons. For one, a macro does not have access to the state of the game/reading the game's RAM. To take the BLJ cheat in this video as an example, it adjusts the mash speed based on Mario's velocity, which makes the BLJ much more consistent. A macro cannot do that because it cannot react to the live state of the game. Also, a macro plays back the exact same input sequence every time. This means that it would quickly become obvious because you would have the same exact input sequence across multiple runs. On the other hand, the types of cheats in this video can dynamically adjust to the live game and so you still get a unique run/attempt every time (very useful for streamers who need to farm hours of content worth of attempts).
5. "A better method is to just create a TAS". Creating a TAS is a time-consuming process and you would need to individually pre-prepare each TAS run behind the scenes. You would also have to completely fake your reactions and that you're playing live. On the other hand, the cheats in this video seamlessly integrate with your normal gameplay. You can still mostly play as normal and farm hours of attempts, but simply gain the advantage that boosts your gameplay to the next level. You still get a unique run every time, unlike a TAS which you could only use once and then have to keep creating more. So these types of cheats work far better with the modern "live stream meta" of speedrunning.
I would also remind people that the cheats shown in this video are only an example. There are countless possibilities with this type of method across many different classic games. You can probably think of some for your favourite speedrun game to target specific glitches/chokepoints. And these cheats are increasingly easier to do as many of these older games are getting decompilation projects (not to criticise those decomp projects, they are awesome and I have contributed to some myself).
to be completely honest, this most likely could have killed sm64 speedrunning forever.
that example is in no way invisible to handcam lmao. people can just check if your timing with your inputs lined up with the actions in game
@@Maker_7 gd has much more advanced cheating methods than this and it certainly isnt dead
A macro can most definitely read the games ram. smwcentral even has some ram addresses mapped for the public. If you are savage enough to change the mario 64 source code and recompile it you can probably make a macro by reading game ram.
I think we have much different definition of the word macro if you think that macros can read RAM. Generally a macro is just understood to be a pre-recorded button sequence. Also, all the serious Mario players play on the original N64 console, not emulators anyway.
Using Billy Mitchell's (Silly Bitchell)'s forehead as "the mind of a cheater" was beautiful.
innocent until proven guilty
@@d.d5619 He’s proven dude. Take the L.
LMFAO
Inb4 he gets sued by Billy
lol, good one xD
Carpetless difficulty is a little overstated/inaccurate (can wallkick on different frames, pause early and adjust, it's 30fps game so frame perfect isn't as hard). But it certainly is a very dangerous/costly strat, especially with where it's located in the run.
I guess at some point certain games need input display or something to handle this stuff. There's really no limit to the extent people can go to if they really want to. We're lucky that the cross section of people knowledgeable enough to do advanced cheating, and people who actually want to seems to be pretty small. But every now and then...
burger
the cheats can be built to show an input display that would make the vanilla game show exactly what is written on screen. There is no fix to this problem except for runners gathering evidence of their runs being legit. Similar to anti-cheating measures in online chess tournaments. Players have microphone open, a camera shows all of their screens as well as their hands moving the mouse. Not all runners can afford webcams I guess
@@snared_At a certain point, input displays and input cams will become required, just like in many speedgames.
I love how ppl are talking about cheating in a nearly 30 year old game like it’s bank robbery or something 😆
a handcam requirement would probably work better, I bet this could be repurposed to work on a modded version of retrospy, where the retrospy itself would be changing the inputs, thus showing the cheated inputs through the input display and also bypassing the need for a hacked rom entirely.
i've been saying since 2014 that i could just make a tas input hack (that maybe even has 20 different playbacks per star,?) and just get world record like that. maybe even put a "press L to let Mario take the wheel" button that just does one star really well for you. it'd be trivial to do this for any game honestly.
Yoshi man out here breaking the whole emulator speedrunning scene when that one drops.
Kaze when is Mario Maker 3
@@Katx- you can just put it on a real n64 and break the console scene as well
wouldnt replaying the tas be way too obvious though?
@@TheGaming100 if you just prerecorded a good version of the star you did earlier, there would be absolutely nothing to point to as it would just be a recording of you doing the star legit
As feather ruffling as this video may be to the community, nothing has ever been solved through ignorance and knowledge suppression. I hope this video is the first step to finding a solution to detect hacked runs.
agreed
Agreed. Not to mention, if Retro knows about this there's a good possibility cheaters already know as well. It's more important than the community be made aware what cheaters are potentially up to.
Thing is even most modern games with online capabilities and anti-cheat still have cheaters.
So it's near impossible to fight those same cheaters when the game has zero anti-cheat and zero tools for moderators to verify a run wasn't cheated.
It sucks but it's a never ending war imo.
@@Rrezz hands camera seems like it could be a solution and maybe some input tracking but only with the hands camera
I disagree.
If you could make a Thermonuclear device capable of creating an explosion which can destroy an entire city, simply by using 3 liters of coke, 2 onions, and 1 liter of chocolate milk...And only 3 dozen people in the world knew how to do it...Would you come out and tell everyone?
What you say is wishful thinking about everyone being decent and completely ignoring mental illness, depression, and a lot of things.
There are bad people, there are mentally ill people, there are people who crave attention they do not earn. I do not see how this information does any good to the community. The people moderating and maintaining the communities are the only ones who could make good use of this.
You should always show the cheats and hacks you guys find. ALWAYS. Never hide them or keep it secret. We can't develop ways to detect or know what to look for if we are kept in the dark. It's easier to stay alive when you know where the sniper is watching from.
That’s exactly what a cheater would say
the argument for and against open source code.
So one thing i find interesting you point out at the end of the video is that established runners are ones to be less worried about, but in my experience of seeing cheaters exposed its often is some of the very top runners who are legitimately good at the game that end up also cheating(for a mix of reasons). So i dont think just because a person can show up to a live performance and get a close wr should prevent them from still being under scrutiny, the pressure to stay at the top can corrupt a lot of people.
Totally agree. Someone sitting in the top 10 but never quite hitting the record might say "I can hit the BLJ/throws 90% of the time, why should I lose WR over some BS" or even "I'm just going to have this hack as a backup for practice runs so I can get my nerves down". You reach the point where getting a record isn't about whether you can do it, but about how many hundreds or thousands of hours of your life you have to burn repeating the same actions to get it, and the temptation gets pretty high.
It's the ones who know how to play the game that are harder to catch. For all we know the top runners in most games could just be really good cheaters
I was about to comment this, JohnDoe360 will raise all sorts of red flags, but established runners who just need that one perfect run, but have been denied it due to one or two mishaps would be very tempted to cheat.
If you want a great example of this, look up "The Story of the Greatest Rocket Jumper of All Time". That's exactly what happened. If he wasn't stupid, he never would of got caught.
Really makes you wonder if there's pros that aren't dumb as rocks and just have some fool proof cheating that can't be found out unless they told us. They'd just take it to their graves.
Exactly joe shmoes cheating means nothing and its always super blatant the most prominent cheaters in competitive gaming were always good players who cheated and were good at cheating as well and sometimes they started theyre comp career with cheats and sometimes they get corrupted
What hurt me inside the most is not the cheat itself, but the act of pulling the cartridge out of the N64 while the system was still powered on. 😬
That is what I was thinking
Why's that bad? I'm assuming it maybe messes with the data in the cartridge and console like tilting the cartridge does but I'd still like to know-
@@Alichu-05 it can damage the memory ram of the cartridge, permanently corrupting the game or destroying its ability to create saves.
@@caetanolunardi3429 ohhh I see, thank you!
It's a a kids game machine, if it can't handle that and drool then it truely was just crap
I think it's a good that these types of cheating methods are out there. Rasing awareness is the first part of the battle. If people think this truly is getting out of hand, we might need to do what Pokemon red/blue speed runners do and only allow a very specific emulator of the game that must need to be examined for changes.
I'd say requiring the use of a hand-cam like SMB runs have, an input capture device like a RetroSpy (although runners might object if it'd make switching between OEM and HORI controllers more difficult), or require some sort of tool that verifies the checksum of the version of the game they're running with an ED64 or SC64 could all be reliable options for helping to verify the run if things reached that point (although I'm sure requiring a checksum every run would be going a bit far given the tools needed and if people even know how to do so).
Then they'd just fake the checksums and the input capture device, maybe even find a way to fake the hand-cam.
@@CouchPotator Faking the checksum is, I believe, considered impossible.
@@renerpho custom firmware for the ED64 to just pretend it's doing the checksum correctly would be trivial at the point someone is compiling their own custom rom to cheat.
handcam is currently beyond reasonable to fake and is likely the only way to deal with this currently. If hand-cam falls then the only resort will be live in person runs on validated hardware, being generous 20 years maximum until that happens (assuming speedrunning persists that long)
@@CouchPotator I don't think it is possible to fake handcam in real time, otherwise there would be TASes with faked hands
@@Jalaeeven handcam is only reliable within ~1 frame at 60 fps- it will catch anything blatant, but if a program checks for slightly early inputs and adds a frame of delay, it would be nearly undetectable. Means you need to be very nearly frame perfect anyway, but most cheaters who make big news are already very decent players in their own right
Holy shit retro its been 5 years
This can apply to any make-or-break trick in any game that has a decomp project. Pretty huge, and good to keep in mind when watching suspicious runners. I hope speedrunning is small (and honest) enough that anything like the CS:GO private cheats market can't flourish enough to be worth existing, but you never know.
Thanks for always uploading quality.
It would have been possible before the decomp project. People have been modifying game code in ROMs for decades, the decomp just makes the process easier.
@@BainesMkII True
There have always been cheaters in speedrunning. The easier it is to cheat the more cheaters there will be. Many online multiplayer games are ruined by cheating and the same can be said for speedrunning as well. I think it's best to keep gaming as a casual affair for this and many other reasons. Gaming is a meaningless endeavor anyway and to spend so much energy to become good at a game is pathetic to say the least.
Yes, I setup a cheat like this for Minecraft speed runs during the dream drama. I knew it was undetectable, but since I’m not a cheater, I didn’t actually use it. Basically I patched the raw Minecraft binary and that allowed me to fake not having mods installed. The mods were compiled into the game itself :)
Edit: oh yeah, and mc has a check that the game is modded that displays in the corner, but since I had the whole source code, I just turned that feature off
The biggest driver behind the CS:GO and CS2 private cheat markets are the skins having financial value, along with the streamers, esports pros & others who make money from the games - people are incentivised to make more subtle, less detectable cheats. It's not to say there aren't cheaters in speedrunning, but it's traditionally been more about video manipulation of runs by splicing & editing with some unauthorised mods to improve things like RNG thrown into the mix... fortunately when the incentive is only bragging rights, cheating tends to be less pervasive.
Decomp isn't what enabled this, but it certainly made it highly accessible to the masses.
Yeah anyone with (deep) knowledge of MIPS disassembly could've done this before but this has certainly lowered the potential barrier to cheating. Obviously, this is all a moot point if someone doesn't play on original hardware as there are no guarantees there.
what the Game Gennie back in the day did was effectively "code injection" these de-compilation steps just inject that code right into the binary running on the carriage, and the only way to maybe detect them would be to do some kind of check-sum-hash, but that would require an interface of "after each run insert cartridge into this, and if the test fails then tampered" (almost like testing performance enhancers after sporting events...)
@@gardian06_85 You have to know how to write the code lmao, not just inject it.
The BLJ cheat could very easily be confused for using a controller with turbo capabilities by someone just watching from the outside
exactly. If theres no rule against turbo controllers,one could just...use a turbo controller to further mask it,with the turbo off
@@alexisverity47 turbo is banned lol
@@CatOnVenus183 disappointing but understandable
this video is very appreciated! as someone with a strong interest in speedrunning this is a very strong case for, especially decompiled games absolutely requiring input display/handcam at top level.
but personally i also think that speedrunning communities as a whole have done a pretty good job at enforcing high standards for their players. A system of increasing standard enforcement proportionally to placement/time not too dissimilar to how SMB1 does it is probably enough to dissuade people from using mods to hack their way up the list until suddenly they have to show handcam/input display and have to hit the tricks they've been cheating on to this point.
the fact you're playing on modded software is undetectable, but if you need to show that you actually hit the tricks in your run with input display/handcam while playing live, something which you've shown is possible and part of the insidiousness of the program, your run is now... as difficult as a non-modded run. and if you forget to manually execute even one, the fact you *are* playing on modded software becomes instantly detectable.
Just adjust the code so it only executes the next step if a button is pressed roughly at the correct time. Now you still need to perform the correct inputs, but your timing can be way more relaxed.
"film your hands or it doesn't count" has always been in the back of my mind anyway. Both the tech itself and technical literacy is only going to make cheating more possible and varied in future so in my mind if you are spending hours on trying for records, you owe it to yourself to at least aim a cheap webcam at your hands to remove all doubt (including VS any future methods that might be revealed).
yep this sounds right to me.
Yeah one of the most insane things to me is that many world record runners do not even have camera on. In some cases (marlene) not even mic.
I can understand the "why" (has been installed from past times when running 2 cams was more expensive/technical, has made it the norm in those scenes)... These are small communities each with their own little meta. It's not like there's a central verification body with set methods of anti-cheat across all games that are speed-run (not advocating for one either, Guinness sucks, for example). I feel it's those games that have been stung by proven cheaters are more likely to introduce things like enforced hand cams. The smaller the community, the less likely there is to be someone with the technical knowledge and free time to act as police, therefore the simplest solution becomes the best, IMO... Problematically, I think there's a lot of smaller games with only a few runners where these players have the scene on lock, so change can be slow - until it becomes the norm across speedruns in general to have hand-cams and mics. The hacks are always one step ahead of the anti-hacks, not just in speedrunning!
It seems speedrunning has finally come around to be in the same position the previous generation of competitive gaming culture was in during the corrupt Twin Galaxies era. Establishing how the industry could ever be legitimately regulated seems a Herculean task requiring an astronomical degree of specialization to truly verify, that would also require a veritable army of those specialized to keep up with all the various categories in such a massive diversity of games. Mercy
Not really. Most people still understand how stupid cheating is and requiring a handcam or maybe even mic audio makes this method impossible. This video is just fearmongering for the sake of views.
@@quilllllyit does punish and price out poorer people for not having the money to establish this setup. While there's certainly more nuance than that, accessibility to the top of the leaderboards goes down majorly and discourages the average speedrunner for aiming so high
@@justinhansen6069 accessibility to the top of the leaderboards? In order to get your times on the main leaderboards you need an n64, a working controller, a copy of the game, a capture card, a device to record on eg a pc, and ideally a crt monitor and a second controller for throws. The leaderboards aren't accessible. A Webcam and microphone is the least of anyone's concerns
@@quillllly i gotta say u cooked with this reply, exactly what i was thinking. You’ve already spent way more money just getting access to the original console and game, not to even mention dropping like $100+ on a capture card, all before you even begin to think about webcam and mics lol. Plus, you can get a cheap webcam and mic for probably less than $50, so in actuality that would probably be the cheapest part of a mario 64 speedrunning setup 😂
@@quillllly I strongly disagree this video is “fearmongering.” The rhetoric on that is clearly an exaggeration, and would imply this analysis brings up no facts whatsoever. I also agree with Justin Hansen. The existence of private experts who can help someone rig themselves a record for the right price, will have a socioeconomic effect on the industry. Saying that people “understand how stupid cheating is” is deeply naive. The bigger the industry grows-and it is certainly huge now; much, much bigger than competitive gaming ever was from the 1980s to 2000s-the more the risk of cheating becomes one that many are willing to take. The problem with competition and so many financial incentives to get to the top is that many people be willing to do whatever it takes to get there, including cheating. Having a robust system of regulation and being able to objectively verify people are following the rules it goes without saying is extremely necessary. To return to my original point, it is starting to seem like it will be immensely difficult to really verify and say confidently that the industry will be free from manipulation going forward. If you don’t believe me on the propensity for power and an in-group to manipulate an industry, simply look into the history of Twin Galaxies and how much insider manipulation was going on when Walter Day was the head of the organization.
I think at this point it would mean all officially submitted speedruns would need to require at the very least an input reader visible on screen or a hand-cam
I feel like that carpetless one would be insanely hard to detect even with a handcam or input reader
@@isaiahromero9861 any of these would with mic audio
That shot to Billy Mitchell's forehead almost made me fall on my ass laughing
Adding a buffer to inputs for roll outs (and maybe elsewhere) would be something very useful and very difficult to detect, even with a hand cam...for that a hand cam AND a synced input reader would be needed... but for top top level runs it might be needed.
Make it a requirement for world record attempts lmao.
If your just going for PB's or like top 10's who cares that's way to many people to force into using this stuff.
But if you force world record holders or that level of player it's a much smaller amount and a more reasonable request.
Handcams are now mandatory to submit runs
I think these could be detected by requiring an input and controller display for people with WR level times, as well as requiring a longer video with a full session of runs. For a game like SM64 where top runners have practiced and grinded for thousands of hours, by the time you have that level of dedication to the game you should be willing to provide this kind of proof.
TLDR: Using code injection or modified game-code, it's possible to add cheats that are invisible to the outsider, since the game appears to run normally.
This is similar to adjusting drop rates, like Dream did in Minecraft and, indeed, MUCH harder to detect.
Unironically mad this was presented as something new when its literally "Bro... what if cheaters... mod their game?!?!?!?" Like yeah no shit.
@@halokinger This is about it being undetectable, though. And as advanced as this.
@@JeffGoldblum64 As most modifications are...?
@@JeffGoldblum64 My brother in christ, that is literally every mod in existance. I am currently for fun speedrunning the 1st level of ultrakill against my friend. He uses Bomb movement tech, I dont. I beat him now to 33.3 and am 3 seconds ahead of him. Neither of us modded the game here. But the best part is: I actually cant say for sure that he didnt. And I cant say for sure that any speedrunner modded their game. Cause a mod by nature isnt supposed to be detectable unless it quite literally adds things to the game. But there are a lot of mods that simply enhance FPS, unlock FPS even, overhauls of stats and more. There is a TON of stuff in the modding scene that you cant detect with an untrained eye.
yeah I thought this video was going to be some crazy genius cheating method but then it's literally just "mod the game" for 10 minutes...
This is why I never unsub to legendary channels, always a chance they're back:>
"The mind of a cheater"
Chefs kiss my friend lol
Glad to see you’re still making videos. Looking forward to watching this. You’re a legend hahaha
Clickbait + not getting to the fucking point, name a more iconic duo.
requiring audio from controller inputs (which seems to already be done for the most part) and (maybe handcam) could be a good solution to catch most of these
Thegaming100
While I agree with that given how they're implemented, I don't think it will be too long until these kinds of hacks either ignore your inputs (so you can mash / spin and it plays the "right" inputs behind the scenes) or even simply buffers them so a 1 frame window becomes a 3 frame window, making these incredibly hard to detect even with a hand cam / audio
I understand the danger of showing cheats but the community being aware of the possibilities is a good thing. Credibility of the runner and having a camera showing the controller should be the new standard. Also live streaming attempts would make it harder to consistently cheat and not get caught.
Good lord retro its been years
This feels like hand cam + visual input is gonna be required for speedruns now
Only for the top level.
Holding the bottom of the leaderboard to the same standard as the top is crazy and kills speedgames. If you're good enough at SM64 to get the world record, you've spent long enough at it that this stuff is a pittance.
I'm sensing a SpaceUK type situation in the future.
Someone shows up out of the blue and immediately makes it to top 50. Soon after that, top 25. Then 20, 15, 10, runner-up. A few people are skeptical, so they make claims of cheating. But since this guy streams his runs and his gameplay seems legit, no one bats an eye as he finally makes it to the top. And as our champ keeps improving, time barriers are simply destroyed in matters of weeks... if not just days.
However on one fateful day, a top runner will catch something peculiar. His button presses have seemingly inconsistent delays compared to the gameplay. A moderator is contacted and an investigation begins. It doesn't take long before a pattern is found. Every PB that this "champion" gets has a random C-UP input on the intro text box. Turns out, he was using a modified copy that basically turned top competition into child's play. One seemingly random button tap turns Mario into a TAS. This guy wasn't a champ, but rather a SCAM! And now, because of his disgraceful actions, we may not trust any up and coming top runner for a long time.
I think the amount of trust that's historically been put into just recordings is insanely high. There have already been many examples of simple splices being almost impossible to detect when done right. I think that official records will at some point have to be done in front of some sort of an event for it to count as a wr, similar to rubix cubes.
Given that speedrun WRs take hundreds of attempts to achieve, this is effectively impossible.
You’re completely right, and that wouldn’t be a problem if that were the only way to "officially" measure records. The most popular speedrun leaderboards are by and large unofficial anyway.
@@chrisprice8112 well, would have to just reset the WR to a lower record, which would be a true 'verified' record
@@thewiseowl8804 Just introduce a new category: Public%, and make it the default.
that would be incredibly unfair though. there are people who can't travel for a multitude of reasons. speedrunning should not be only for rich able people.
I'd never thought I'd see the day you uploaded again. Welcome back, Retro!
Never thought I would see more videos from you nice to see you back
Iunno who you are, nor am I a serious gamer, let alone a speedrunner. I also don't know how this video ended up in my feed, but I thoroughly enjoyed this. Very well executed, even to those of us who aren't involved in the community. Have a sub my dude, I'm interested in seeing more of this type of stuff.
While not speedrunning, a Melee player back in 2016 was caught using a modified version of the game to win tournaments. His version of the ISO had been modified so that you could enable a bunch of major buffs for the character Pichu if you selected Pichu with a specific outfit while holding a specific controller input. He was eventually caught because the buffs were too obvious and he behaved suspiciously by refusing to play on any set-up other than his own, prompting others to secretly copy his copy of the game so they could examine it.
Secretly copy of his game? Dude I know that story and the people who exposed himself did bad stuff to him by stealing his sd card and revealing personal info.
Even if cheating is bad, stealing stuff and revealing his files is very wrong.
@NumbHydro They only did it because it was super suspicious, and they to prove it before he just deleted it. I don't blame them one bit.
@@NumbHydro I'm not saying eye for and eye is the right way to go about it. That will leave everyone blind. That being said, dude deserved getting served in one way or another.
@@isocle Stealing is stealing
Thanks for coming back after 5 years to expose this ridiculous shit
seeing cheats like this breaks my heart but on the other hand , i am glad decomp exists and i guess in a technical level this is showing me things that can be done (when doing it for good reasons of course , not for cheating)
"... the mind of a cheater," followed by a close-up of Billy Mitchell's head. LMAOOOOOOO
So glad you uploaded another video haha, brilliant content.
I’ve been a dev for 25+ years. Seeing this brings back an idea I had to build a live handcam generator based on a prerecorded sample set. Because of the low resolution and frame rate used on handcams, you would only need to morph 2-4 partial low quality frames per button press to create a fluid, photographically unique gesture. With a few seconds delay, you could “live” stream with audience interaction and nobody would know. My point is handcams need to have higher resolution and frame rate to be safe. That will not be enough soon.
It's an interesting video, but honestly I think you're maybe exaggerating a bit that people will be mad you've "revealed" it. While the full decomp does make editing the code here significantly easier than it would otherwise be, editing any game to give desired results is almost as old as gaming itself, and this isn't the first time such a method has been abuse-able. Speedrunning has literally always required some degree of an honor system. Once upon a time it was rare for people to even provide footage at all. I don't feel like anything has changed.
I mean, the fact you literally wrote out the code and showed it for the whole internet does take literally any skill or knowledge out of the process of successfully cheating now, so maybe in that respect one could argue this problem has now been heightened for mario 64 specifically. But honestly I think the people who would want to do such a thing would have eventually figured it out just the same.
That's what I was thinking. With learning assembly you can pretty much do whatever you want in terms of editing, you don't even need a decompilation, so this isn't limited to just SM64. ROM hacking has been a thing since forever, so when they revealed game modification as if it were a huge discovery that nobody knows about I chuckled a bit.
@@nippo4512 that just seems unfair
I do think this video should be deleted.
Nothing is gained by drawing attention to the dark underbelly of speedrunning like this. It will only result in more people cheating, thus hurting speedrun communities.
@@TheFinalChapters
People with the coding knowledge to make mods like this were certainly already aware of the cheating possibilities the decomp had to offer.
In my opinion all this video does is inform more casual viewers of a possible leak in the speedrun scene, and “drawing attention” to potential cheating can only lead to positive change, right?
I can assure you, people were mad that I revealed this lol
These videos bring joy to my heart whenever I see them. Every time a new speedrun cheat method is discovered, it becomes harder and harder to verify a "real" run. The hobby started off as just that, with just gamers seeing how fast they could beat the game, but it's become so competitive and serious that it has lost its soul. Maybe when all the top speedruns are fake people will come to their senses and realize that they've wasted their lives by speedrunning.
I feel like there could be an input recording tool that could be used to verify speedruns
AFAIK there is on most emulators and they are required. Helps show "too perfect" stuff like a TAS but doesn't help when it's not too perfect, but very close. That's why he talks about cred in live events. Like bluescuti could have faked his beating of tetris but we can be confident is it real because he has played live with random seeds and shows the ability of doing that given a reasonable amount of chances. If I say I can beat metroid in 4 minutes (I didn't look at the record don't kill me) and never played live only have a bunch of runs without being live (with some randomness to prove legitness) or showing the controller it would be sus.
@@Needed4Reddit however, there couldn't be a tool for consoles to record inputs?
@@varietychanyou could just modify the game to make the displayed inputs identical to doing the trick legit
@@dsobransingh you can't, input recording only should take inputs from the controllers, not from the game
@@varietychan but how do you verify that the inputs came from the controller and weren't generated by the modified ROM?
just watched this after watching your cod cheating video. You make great content bro. I would love to see more deep dives in to cheating and can't wait to see your channel grow.
if someone live streams all their attempts. It is going to get suspicious when they pull off those strats with such alarming consistency.
Same with an obviously amateur player
This is true but their are exceptions. I guess it would depend how consistent a player might be livestreaming and recording.
Personally, I always do better when doing offline attempts because streaming makes me nervous and I'm distracted by chat.
I mean, maybe they would be caught eventually, but initially people would just think the person was a good player and praise how consistent they are.
I have absolutely heard players get praised for having very consistent BLJs. (I don't have any reason to think any of those players are cheating, of course, but I've definitely heard that kind of praise tossed around).
Using this video as an example, they could just do the BLJs normally most of the time then hold the button when they're on a good pace, taking the difficulty out when it matters
@@skeletonorchestra2900 Most cheaters from my experience don't tend to balance things out well enough to avoid being caught by making statistical anomalies.
Wow. This is crazy.
Blistering ep! Well worth the wait. Thank you, retro.
Incredibly well made video. Thank you.
Speed running has always been full of cheaters and nowadays it's so available and easy to do that if communities want to eliminate it they will need to enhance the requirements for a valid speed run. There are simple things they can do to prevent any possible cheating.
I love how when you said "think like a cheater", you showed Billy Mitchell.
I'm not a runner but a few ideas for potential requirements:
ROM Checksum
Video of controller to verify inputs
I know you don't like adding cost but an inline input recorder could be made and distributed so inputs can be played back (like DOOM).
I know there are aspects of all these in various games already.
Any verification device can itself be modified to cheat as well. Input recorder works in theory, but requires players to buy an extra, third-party device, and even legit runs won't always play back correctly due to subtle differences between consoles/cartridges. If one console's CPU is running at 93.0000007 MHz and another's is running at 93.0000002 MHz, there's no perceptible difference, but it can change RNG, timing between chips, etc. This slight variance is normal; every console will have a few Hz difference, even brand new.
Look into Super Metroid TASes for more about that. They had to modify a console to restore it to precise factory specs, because it was old enough that the timing crystals were slightly off. That didn't affect normal gameplay at all, but meant occasionally the main CPU would have to wait slightly longer for the audio CPU to finish a task, and that was enough to introduce a lag frame and throw off the whole TAS.
Even more shocking is seeing another video from this channel, holy shit
I SAW THAT!!! Bittcy Milleell is going to sue you. Get your wallet and red joysticks ready.
Lol that Silly Bitchel
Love when channels i thought were dead upload again. Ready to dive into this
Changing source code just would require a checksum hash to verify or key logger of inputs. That does get into the range of how much you need to trust/distrust someone in a hobby space
how are you going to get a checksum of a cartridge running on an N64 without modifying the game or the hardware?
@@JeffGoldblum64 adaptors like retrode don't modify anything and once the rom is downloaded it's trivial to checksum it.
@@reallifeistoflat how can you trust the checksum you are getting from someone who might be cheating? explain how you can do that
@@reallifeistoflat also im talking about the n64, which is what most top speedrunners are playing on. who in the hell is going through the trouble to have the cartridge just to play it in an emulator? and also that doesn't stop anyone from inserting their own code.
@@JeffGoldblum64 if this is something the community thinks is valid you could make it a requirement for wr validation to checksum on camera after the run. I believe runs like Minecraft require you to show your last few world generations, i might be wrong about the specifics. It's just another step that might be required for high level. Would also be easy for in person events.
I'd say a pretty decent standard would be to require input detection overlayed over the game whenever possible, and have a hand cam showing the real world controller inputs as well. This way not only can moderators see what inputs are being put in by the player to verify that it's all legit with what's happening on screen, but this can also be compared to the hand cam to confirm that the same inputs are what's being put into the physical controller the player is using at the same time. It would be very hard to, for example, simulate a fake BLJ while still having to do the correct inputs anyway. If the player is just holding down the A button while doing BLJ's or just holding the analog stick straight up during the Bowser fight, it's a clear tell that they're playing a modified version of the game that gives them an advantage. Just these requirements alone would do a lot to make this kind of cheat much more difficult to pull off.
These cheat codes add additional CPU instructions, so they should be detectable, but not in a 30 fps recording. I think, speedrunners should provide an input trace which can be replayed and would always lead to the same result (as long as the RNG seeds is also exactly the same).
1 branch per frame is not all that much though, 30fps is already the game's full framerate, and RNG is already deterministic by the game's logic.
These would be adding a handful of microseconds here and there. You couldn't detect it by frame counting, because those few extra cycles just mean it spends a few less microseconds waiting for the frame to finish. Also, you could craft your code carefully - or only change number constants - to take the exact same length.
One another thing that could verify them is somehow verifying the checksum of the cartridge.
Even if a single byte is changed, that will dramatically change the checksum. And using multiple checksums (especially modern ones) will make brute-force attack of them less viable.
how are you going to verify the checksum of a running 64 game without modifying the game or the hardware
We know the checksum of a legit rom. Surely someone with the knowledge to make these types of hacks will have no issue feeding whatever device you're using to determine a checksum the real rom info.
@@thomasmathew13 Good luck implementing a device that checks the rom for legitimacy in the official rules as a required item (That you would presumably need to purchase), it would kill the community almost instantly
New carpetless? I snore.
Bob-Omb carpetless? I kneel!
ah yes kneel for a much harder and slower method
@@cay7809
It's called swag. You wouldn't understand.
Yea check out my cheated bob omb carpetless i did that was wr once
Damn man, this hurts to see, although I don’t find this kind of thing that surprising given how advanced and how far people will go with this kind of thing. You did good by exposing this, don’t feel bad for making this known.
When the world needed him most.. he returned 🙏
That ai Mario wojak gives me like hypertension
The best solution I can think of, is to add some sort of secret watermark that's not well known. But like, maybe a certain thing in the game is sized differently but not easy to notice unless you know. Examples could be changing the color of something to a slightly different color. Or changing something's size by just a few pixels. something that you can tell if you're looking for it but it's invisible otherwise
Your idea will only work for "script kiddies". Adding something to original game is impossible, on the other hand, adding something to public sm64 source code is... public. So cheater can remove that change simply by taking older version of source code.
It’s called hashing. Having a known valid digest value (the result of a hash) means that we can show whether or not a game has been modified. Hashes provide unique values for a set of data, aka “a fingerprint”
@@cmspteam it's not that it's only for script kiddies, it's just something that has to be kept secret. Obviously it's too late to do so now, but it's just the only solution I have.
@@soybeanrice I know hashes, but how would we confirm? "Hey, send us the hash of your game with every run!" All they'd do is send a hash made by the real game while they run a separate modified copy. I think it should be something like, if you do an emulator run, you have to use this specific cheat that generates something like a hash based on the code, current room, and playtime or something and have it display somewhere discreet on screen. That way someone should be able to recreate the run and have the exact same hash. Although someone dedicated could probably make it run on an N64 and say they're playing an official copy so they couldn't have modified it and can't run the hashing code. Therefore I'd say my original idea is still better, have something slightly off that's hard to notice. Say the git had some issues and they had to remake it, so they just reupload the code with the slight differences that'll be dead giveaway for mods
@@collinlawson713You bring up very good questions that the community will need to address at some point. I have ideas. However, the point of the video is that code modification is "UNDETECTABLE" and to that I strongly disagree.
Bro, just dropped this video after a five-year hiatus, like nothing ever happened.
The fact you can make it an n64 cartridge... this is tragic. You absolutely had to show this though, otherwise cheaters would've abused it under the radar and no one would know this was even possible.
Makes you think. Someone could bring their own cartridge or controller to a live event and get a record or victory. At least I can brace for the inevitable drama.
@@Yo2cool i think thats exactly why most good live events would supply their own cartridges/controllers and not allow the runner to use theirs
Flashcards are not accepted.
Or you could modify the console itself. Overclock, or have it read from an embedded ROM instead of the actual cartridge... end of the day, there are a million ways to cheat.
@@renakunisaki that won't happen at events are you dumb?
i think having an input cam should be proof enough since it's easily provable if what happens in the game was actually inputted by the player
Yeah. That way, ANYTHING off equals cheater.
Thanks for the tutorial I can finally get my first world record! 🔥
As scary as this is I think a lot of it could make really sick bonus categories. Auto blj and spins etc are really cool in terms of how fun they look lol
While these are interesting ways to cheat using the decompilation, the same could've been easily accomplished by using macros, turbo button, or even just a gameshark/AR code. In fact, using macros would be the more reliable way to cheat than modifying the code directly.
Taking a look at your first example with the BLJ, you could easily just use a turbo button and hold down A instead of making the game automatically input it for you. Using a turbo button would have a strong benefit over the game itself, as input viewers would show that the A button is repeatedly being spammed instead of being held (with the latter obviously meaning there's a cheat going on somewhere).
Your second example doesn't even need the decompilation. It's literally just a simple ASM mov instruction (or setting the value to the max (aka a gameshark/AR code will easily do it.)) It also suffers from the exact drawbacks as the first example, where the input viewer will show that the stick isn't being spun fast enough to match up with Bowser's spin speed. A macro with the analog stick being set to spin super fast and the cheater just needing to time the B button would look more legit on an input viewer.
Finally, for your third example, it suffers from the same thing as the first two did. It's overly complicated for something that any simple input viewer will clearly show that its cheated. A far better way of cheating that doesn't even need the decompilation at all with the bunch of checks would be to just find a consistent setup strat, then using TAS tools, making a macro that repeats the same set of actions every time. This will also have the benefit of actually showing the inputs being pressed on the input viewer.
Overall, using the decompilation project specifically to cheat is very inefficient. The drawbacks FAR outweigh the benefits:
- Any simple input viewer that displays the button being pressed on the controller won't show buttons that are getting pressed by the game itself
- Needing to set up an entire project, compiler, and w/e else to be able to create a ROM with your changes
- Needing to know how to write C code and know where to edit to actually make the cheats
- Most cheating changes can either be easily done with a gameshark/AR code and a macro or turbo button that'll show up on an input viewer
In the end, this is a pretty niche way to cheat that'll end up as nothing more than a novelty (at least given the examples you've shown.) The only thing you could argue with using a hacked decomp over macro/turbo is that it's "cheaper", but given that to run it on console you need an Everdrive 64 (which is expensive as hell), it's not worth it.
Imagine you are streaming 5 hours of attempts. With the TAS approach you mention, you would be doing the trick (e.g. carpetless) in the exact same way every run because the input sequence would be the same. Therefore, it would be very obvious unless you created a bunch of possible permutations. You'd also have to manually switch the inputs between your controller and TASbot mid-run and ensure there are no TAS desyncs which is tricky, plus if there was a desync because you didn't start the TASbot at the exact right spot then it would look fishy. The decomp cheats can dynamically respond to any situation in the game and are much more flexible. Remember that what was shown in this video is only an example.
Also, in regards to the BLJ, I guarantee using a macro would not work very well. It's not just about mashing at the fastest possible speed - there's a rhythm to it. In particular, the code shown in this video will mash with a 3-frame delay until Mario starts picking up speed, and then switches to a 4-frame delay. Of course, you need to be able to read the RAM to do things like that. So it's far more powerful than what a macro can do.
With this information I now have the desire to speed run Mario 64.
Lawsuit coming at 2:56
The wild Billy Mitchell is great!
Maybe it's possible to come up with some way to enforce a ROM on all runners for verification purposes. After a run, someone can run something in the menu or otherwise that prints out all executable memory, logs it, hashes it against a legitimate ROM... is there a way to verify rom size through hardware? I'm not at all knowledgeable about N64 hardware or software, so this is far from a proposal, but if someone can find the right missing pieces there could be a way to substantially raise the bar for cheating through ROM manipulation...
Something like the above might be verifiable through counting clock cycles... if someone is trying to make a fake "good" output, maybe it'd be hard to get CPU time to align. Obviously, you would worry about someone reading values from a separate copy within the same ROM, but maybe there is some limitation in hardware that makes this more realistic, i.e. different memory fetch time for different parts of memory... someone might be able to do something with this kind of approach?
Seeing as it's a console game, it would be really hard to tie any give recording to a signature that says "yep, that's a legit sm64 cartridge."
If it's a machine that you add to your N64, how do you know the recording was done on it.
If you make GameShark code to do a checksum or hash on the game and show it on screen, how do you know it wasn't modded to spoof that?
@@herrabanani obviously I don't really know the implementation details, but obviously a checksum by itself wouldn't be enough. You could sleep for some amount of time and then display the correct one. If you could display some output as this is being calculated, maybe there isn't a way to spoof this *with the same amount of CPU clock cycles*. Or, if you were just printing out the executable memory, maybe there is a performance reason (maybe n64 carts ex mem is set at an absolute address. You need to therefore either have a copy of the entire game [which might not be possible due to size constraints, or otherwise inducing performance issues, or fails an OS check for size] or conditionally read memory upon verification, causing a performance drop [once you get to this address, stop calculating] OR maybe you could edit your own executable memory real time [must reload your rom every run!!!]) why this is hard to deal with. Anyways, to reiterate, I'm just spitballing here.
@@xininja8167any potential check could be spoofed. It's the equivalent of having a green checkmark on screen if your cartradge is legit. If you can't trust the game is legit, how can you trust that the checker is legit?
Not to mention having to require every player to modify their hardware to check for the modification with the use of everdrive carts or game sharks.
Imo the best way to legitimize runs is to require people to show hand cams and stream/record all of their run attempts. That isn't perfect either but it's the best you can do imo.
This would be a good solution, but the only issue is that distributing this hypothetical rom would be illegal, unfortunately
This is a good video, definitely informative for those whom do not know anything about cheating.
It's starting to feel like we need cameras (with audio) on everyone's actual hands, for every game, which kind of sucks because this is an extra barrier to speedrunning. It would be interesting if there was an automatic way to cross check the inputs as detected on camera, vs the inputs in the game
You wouldn't need this for every runner - just at moderator discretion. Someone in the top 5 cheating is obviously much worse than someone who's ranked 350. For larger communities like mario 64, the ability for top runners to make enough money off of twitch donations to get a setup like this is almost certain - and if for some reason it's not, some sort of crowdfunding system would likely be possible. Only issues are for runners that are legitimate, but are generally disliked for reasons outside of the game, or for games that have no larger audience to fund their speedrunners - but the number of games like that which have decomp projects are likely quite low.
Not much of an extra barrier. Everyone has a smartphone
@@peter-rhodes True but a lot of times people need their phone for other things while running. Buying two cameras, a decent mic, and having a pretty good PC does add up
you dont need a good webcam for anti-cheat, $5 spend.
Yooo!!! I was wondering if you were okay out there, but I couldn't remember what your channel was to check. I'm glad you're okay :)
Absolutely brilliant return to UA-cam. Warfare and Ridge approved.
As a computer science student I’m not surprised but amazed that no one published this bodes earlier on.
i think your perspective on what to reveal / what not to reveal is a bit odd. carpetless is op yes, but its also hard. some rando with a hacked ROM getting carpetless out of nowhere is gonna raise lots of suspicion, but a rando getting good BLJs / good bowser throws wont nearly draw the same suspicion
the cheats aren't for low level players but top level players. I believe it was karl jobst that said speed runners cheat to get a pb faster when dream's cheated minecraft speed run came out. Meaning its for some one who already had a good pb and knows they have the skills to shave 1 second or half a second off of their pb but they get impatient with them selves at their minor time losses.
@@anacreon212 cheats are cheats, people will use them no matter what. its more of a matter of what is easy to spot from a top player level that matters, top players can get perfect bljs multiple times every time they play, but almost nobody can get carpetless without a single mistake.
so using a cheat that gets carpetless perfect is gonna look way more sus than a cheat that gives you perfect bljs / owless / similar strats that top players have mostly consistent but up and coming players might not
another thing with carpetless; since theres tons of inputs youd be able to see if multiple people are using the exact same inputs on the exact same frames, whereas a real player might buffer inputs, or have variances in the frames they hit inputs that arent frame perfect (of which I think carpetless has a few that are a few frame windows)
another thing too, if someone uses the same cheat for carpetless twice in a stream, thats way easier to notice than it is if they use a perfect blj cheat
you have to spin your stick 8 times per second to get that throw and that's physically impossible for many people. Some rando getting it would raise suspicion. Also controllers are loud, you hear them. This wouldnt work
@@CatOnVenus183 problem is, it *is* possible for some as you mentioned. And a good mic can be set up to pick up only voice audio. So I'm not entirely sure it's that unfeasible
It's sort of expected that a highly popular game for speedrunning having its source code leaked like this would just turn into a cheater's haven. It's highly sought after, a nice quick run, in and out. I could've figured it would be like this as soon as I heard the game was decompiled. After all, like how speedruns and techniques evolve and refine, so too would cheats and other unsavory tactics. It's the natural flow of things. When the optimal bar is approached, it's only natural people looking to hit that peak would look into tools to better imitate that "perfect run" they seek. I personally think it's interesting, the idea of a Real Time TAS (Or otherwise modified). Depends on the rules of the game, but SM64 is unfortunately pretty cut and dry in that regard. Still, speedruns evolve, how it might evolve to handle this situation will be interesting to see (assuming it does find an answer).
Important distinction: no leak was involved. The code was reverse engineered from only the binary. (A leak _happened_ but that code was considered off-limits for legal reasons. Also I don't think it contained the actual code, only intermediate files?)
One could possibly also slightly increase the hit boxes of the bombs for Bowser in order to make it easier to hit them.
That would not work as people can verify that. Programming it to release at certain angle ranged would be safer
As cheaters progress, so do the non cheaters need to progress. Its a cat and mouse game.
I wish i was technologically savvy enough to articulate this any better than how im about to, but the best example i can give is in super mario bros for the nes, you can play a modified version that displays your coordinates in the spot where the time remaining would have been, which is used for perfection of wall clipping.
Using this same idea, it sounds like a mod of this game could be made that displays the code of the game's coordinates live as their playing it and if they are using a modification, you would be able to tell because of the automatic response of the game when certain buttons in concession are hit. Because over the span of an entire game, its highly likely that they would reveal themselves.
excellent 👍 I'm not quite sure I will ever trust another mario record again
😂
It's been so long since you uploaded I didn't even know who you were in my sub box. Glad to see you upload again.
I think another precaution the speedrunner community should take is recording the inputs from the controller. these cheats rely on detecting inputs and making it easier for you to do these skips. if you were to be able to do a perfect bowser spin in the game but replayed through the inputs you weren't, that could qualify as you cheating.
How would you detect inputs from wireless controllers like the wii or switch?
@@Gamebuster hmmm... that's a tougher one...
I would assume you'd use a Radio receiver or maybe a flipper zero to record all of the inputs.
because you can also buy non-nintendo switch controllers online and connect them via Bluetooth, maybe it would be easrier to create a bugged controller that could record all of the inputs
The British VA makes it very foreboding. Good job. I don't even care much about speed running.
The backwards longjump trick would require mashing the button to actually look like you're doing the technique legit though, and even then if your controller inputs shown on screen don't match up to the correct frame in relation to Mario's jumping you can still be found out.
so the Controller input UI on the screen reads instead what the game believes you pressed, including the auto-presses, instead of the controller. now its impossible to tell.
@@isodifbrakiul6387 How are you able to get that information from the game itself without using an emulator or some kind of hacked console? Most controller input displays I see read from the controller itself or the controller connection, particularly for something like N64 where you have to splice and tap into the cable wires and use a microcontroller to decode the signal.
@@Scribblersys If you are already running a hacked game cilent, then reading from memory the player inputs should be rather trivial.
@@isodifbrakiul6387 "game client" Do you even know how a Nintendo 64 works?
The best possible chance of learning to detect it is to know about it.
You seem to have a very cynical take on this. I think you're forgetting that while people can advance in cheating, there are ways to advance anti-cheating. Showcasing this predicament was a good thing because it shows what verifiers need to be aware of. I still believe requiring handcams, input trackers, providing ROM hashes, etc, is egregious for a majority of speedrunners (in any game) and is definitely gatekeep-y, but providing it at the very top level (WR, top 10, maybe even top 25) is a compromise worth taking. Every game is different, of course, but the tone this video had is very doomsday-like and discouraging, and I don't think trying to scare casuals off is a good idea.
So basically what they do for SMB1?
@@alaeriia01 I’m not aware of what they do for SMB1. What do they do?
@@Espik_23 As you get better and better times, verification proof gets stricter and stricter. You may be expected to provide full session attempt videos, input files, face cam, hand cam, and more. Times below five minutes are treated with extreme suspicion, and you will need to show a lot of proof there, whereas slower runs are more relaxed.
The idea is that casual speedruns will not require too much proof (a video of the run, maybe showing off the second quest) and by the time they demand the kitchen sink, you're already pushing into the upper echelon of times and can justify the investment.
@@alaeriia01 Okay yeah, that’s pretty much what I suggested
Maybe there could be an "event WR" as a separate leaderboard. That would be lower but 100% verified on the speedrun events.
Reminds me of how speedcubing is required to be done at events, since its too easy to cheat otherwise. This is almost certainly the direction speedruns are gonna go.
It's the end of history, speedrunning history. I mean I know several people that all think speedrunners are just cheaters anyway. It's kind of a poetic little bow on it if they are right and nobody can tell the difference.
Those people are fucking stupid
those people are a bit goofy tbh
thinking that *everyone* who wants to get good at playing a game quickly is a cheater
"But i still think it is the right thing to do"
The klicks. You want the klicks.
"click" is the word
Now fuck off
I still just see speedrunning as a hobby which you will ruin for yourself if you cheat but the thing is that people specifically cheat for the potential fame which brings twitch viewers which brings money.
Hot take: cheating is as much a part of speedrunning as speedrunning itself. It also has done a ton for speedrunning’s popularity. The never ending quest to cheat runs and have others catch these cheaters is in my mind equivalent to fighting in hockey. It’s not supposed to be a part of the game, but it’s a huge part of the appeal.
Nah, it would be better if it just didn’t happen, there’s nothing to appreciate about it and I would rather do without it.
Cheating does not mean playing a completely different game. This is in effect playing a different game.
Hand cam should be a requirement in 2024
From the very first seconds of the video when you took a retired leaderboard mod's words out of context, this video was shown to be in bad faith and some level of fearmongering. Obviously it's possible to modify the game's code or play back input to cheat, and the community has been aware of it for a while. But, making this sound like a conspiracy the mods are trying to stop from getting out is ridiculous.
My personal experience with the SM64 mods is that they are very honest and intelligent people who have established verification standards and methods above and beyond what should be expected from a few volunteers in the community. The amount of work they do to verify everyone's ordinary runs is immense, and they have a very good track record of catching bad actors while keeping a low barrier of entry for legitimate players. Saying that "it's the right thing to do" to speculate and cast distrust about the whole community is nonsense, you're just making the mods' work harder and influencing people to think cheating is a much bigger problem than there is any evidence for.
A few idiots could watch this video and go cheat low-level times without being caught, but if someone was trying to fake being a top player, it would absolutely throw up red flags across the community. Players take thousands of hours to reach an elite level, if ever, through dedication, struggle, and failure. People are accustomed to these patterns and can raise reasonable suspicion when there are outliers, at which point higher scrutiny or submission requirements (e.g., hand cam) could be considered. If the author really wanted to find out and present the truth, they would have at least reached out to the mods. The entire message of this video should be taken with a grain of salt.
couldnt have sait it better, 1000% agree
I understand your perspective but I have to disagree with a few things here.
Firstly, I don't really see your point about how the beginning quote indicates bad faith. It's a real quote and was not out of context - it was exactly within the context of me talking about making this video. It was also not meant to be a criticism of that mod or any other mod. I was simply trying to portray the difference in philosophy that I don't really believe in the "security by obscurity" principle (reliance on obscuring/hiding knowledge).
I apologise for any possible interpretation about the mods being bad. But I did show this video to a number of people before releasing, and nobody gave me any reaction of any hostile interpretation towards the mods, and I have not seen any comments/concerns to that effect (except from comments like yours and from the mods themselves). Honestly it's hard to predict all possible interpretations of a video before it goes out but that does not seem to be a common interpretation from regular viewers. So, I don't disagree with your defence of the mods, but I wonder who exactly you are defending them against.
The final point that you make, about the difficulty of faking being an elite player, is a very similar point that I make towards the end of the video. I made it clear towards the end that we have to rely on a runner's credibility, and that runners like cheese have demonstrated very high credibility.
Finally, you mention I should have "at least reached out to the moderators" - where do you think I got the quote from at the beginning? I did reach out to a few mods, even offered them to be on the video for an interview. But they declined and did not seem too interested in talking beyond the quote I put at the beginning which you didn't like me showing.
@@retro-meisteryou aren't sorry
@aijx4223 The quote at the beginning of the video that says that the lb mod doesn't want the information to be shared is bc it does not provide any benefit at all and only serves to harm the community.
The dude's just being an asshat pretending that it's for the "moral dignity" of the leaderboard.
He's not providing any information that anyone with more than 5 brain cells knew about and he's exactly going against the words of a former lb mod (retro lied for clickbait).
@LFoxyIsBlue If everyone with 5 brain cells knew about this, then why are you saying im harming the community and why are people mad at me for revealing this? Surely it would not harm the community if everyone already knows about this. And what did I "lie for clickbait" about?
Oh man, that's really scary. But it's kind of beautiful how somebody has done all the work.
Wouldn't the bowser throws cheat be pretty detectable? Like people are aware of the exact max speed and can tell if the speed of throws is constant or done by a human. Might be detectable. Very good video though. I was thinking the most likely undetectable cheating method would be changing RNG values through modding, if you didn't make it too obvious. You could even stream runs with it, granted you don't don't make the changes in RNG values too obvious.