4 dudes. At the same time.

April 4, 2011 1:07 PM

Super Mario Bros. 1, 2, 3, and Famicon Super Mario Bros. being beaten at the same time, using one controller for all four games simultaneously. Mega Man 3, 4, 5, and 6 being beaten at the same time, using one controller. Mega Man X and X2 at the same time. Final Fantasy V and VI (warning: 4 hours long).

via TAS Videos: Multiple Games page and @ryanqnorth.

posted by 0xFCAF (66 comments total)

These are some of my favorite TASes. Good post.
posted by codacorolla at 1:09 PM on April 4, 2011

How can that be one controller? I see one spot, early on, where Luigi in the lower left frame is just standing still and jumping in place, while the Mario in the upper right is running hard right.
posted by Malor at 1:10 PM on April 4, 2011 [1 favorite]

This is vaguely blasphemous...

Man should not be capable of such perfection. IT'S TOO MUCH POWER, I SAY!
posted by Krazor at 1:11 PM on April 4, 2011 [1 favorite]

I'm not sure "Famicon Super Mario Bros." is a thing. You mean "Super Mario Bros. 2 (Japan)" or "The Lost Levels"
posted by East Manitoba Regional Junior Kabaddi Champion '94 at 1:13 PM on April 4, 2011

There is a potential Donkey Kong kill screen coming up if anyone is interested
posted by phirleh at 1:13 PM on April 4, 2011 [7 favorites]

Voltaire's famous quip springs readily to mind: "What the shitting fuck?"
posted by Jofus at 1:14 PM on April 4, 2011 [13 favorites]

Oh, wow. That was awesome.
posted by wpenman at 1:17 PM on April 4, 2011

How can that be one controller?

Yeah, I'm missing something.
posted by muddgirl at 1:20 PM on April 4, 2011

How can that be one controller? I see one spot, early on, where Luigi in the lower left frame is just standing still and jumping in place, while the Mario in the upper right is running hard right.

One easy answer to that general type of situation is carried momentum, though I don't know if that covers the specific thing you're seeing. Subtler stoppages probably come into it all sorts of weird ways. I'd love to see someone deconstruct the specifics, though; I like the theory behind speedruns and related stunts almost more than the actual runs themselves.

And you know what I'd love to see? An overlay version of this with the camera pinned to Mario's (or luigi's or whoever's) center of mass. That'd be a weird, blurry world.
posted by cortex at 1:24 PM on April 4, 2011 [1 favorite]

Apparently, the trick to beating any Mario game is to constantly be jumping. Sounds about right.
posted by moviehawk at 1:25 PM on April 4, 2011

How does someone even come up with this insanity.
posted by shakespeherian at 1:28 PM on April 4, 2011

How can that be one controller?

There's a ton of reading you can do on the tasvideos website to learn about how people do this, but long story short, different games do wildly different things on a per-frame basis that let you pull of these kinds of shenanigans when you can issue input at super-human levels. For example, game 1 might only poll the controller state on even-numbered frames, and game 2 might only actually start your character moving if the controller has held down the right arrow for at least 2 frames. You send RLRLRLRLRLRLRLRLRLRLRL on the controller, game 1 sees RRRR and you run full-speed, game 2 sees RLRLRL and makes you stand still.

TAS Videos also do weird things like pressing L and R at the same time (or even UDLR), which makes most games behave in unexpected ways (e.g. game 1 might treat this as R and game 2 might treat it as "do nothing").

One example from the website:

This short time it takes for Mega Man to position himself after running was also used in other occasions. For example when one of the games has to jump to the left, and it was possible to keep another Mega Man which didn't need to move to the left on the ground. If pressing left is skipped for a frame, the Mega Man on the ground will position himself for like ten frames, so he won't move to the left. The Mega Man who is moving to the left will only stop moving for one frame since he is on the air. It was also used at instances where Mega Man had to move on a rushjet to the left, and a lot when multiple boss battles where going on at the same time.

posted by 0xFCAF at 1:28 PM on April 4, 2011 [28 favorites]

How can that be one controller?

Yeah, at first it looks like they may be out of sync but if you wait until about a minute in you can tell they are being played with the same commands to all four games.
posted by P.o.B. at 1:28 PM on April 4, 2011

How can that be one controller? I see one spot, early on, where Luigi in the lower left frame is just standing still and jumping in place, while the Mario in the upper right is running hard right.

You mean here? SMB3 Mario (upper left) starts sliding down a hill, which due to some precision jumping lets him fly through the rest of the level without running. SMB2 Luigi (lower left) and SMBF Mario (lower right) are jumping standing still. SMB1 Mario (upper right) is in between levels.
posted by burnmp3s at 1:30 PM on April 4, 2011

Right around 1:08 in the SMB video, SMB3 Mario is flitting back and forth and grabs a Koopa shell while the rest are ducking or standing still. I suppose on some level, it's possible that the "same" input is being used for all 4 games, but the same input isn't being sent to all 4 games, so boo to this.
posted by explosion at 1:33 PM on April 4, 2011

I suppose on some level, it's possible that the "same" input is being used for all 4 games, but the same input isn't being sent to all 4 games, so boo to this.

I think it's the other way around: the same literal controller input signals are being sent to all four games, but those games, per 0xFCAF above, won't necessarily treat that input the same, especially when the input is pretty weird and ephemeral.

So, it's sort of a "boo to this thing" on the purist assumption that some guy would have just sat down with a controller and a splitter and manually played the four games simultaneously with human fingers and no help. But it's also a pretty fantastic bit of forensic work and puzzle-solving and hackery AND some guy managing to put it all together.
posted by cortex at 1:37 PM on April 4, 2011 [7 favorites]

OMG OW MY HEAD. Internet, you win!
posted by Theta States at 1:38 PM on April 4, 2011

So, what the heck is going on in SMB and SMB2J where Mario is shimmying through the walls? Is that a machine-only glitch similar to the one that can be produced by humans playing in SMB that accesses World -1?
posted by Golfhaus at 1:39 PM on April 4, 2011

To put it another way, in principle this could probably be recreated by a robot with a controller. In practice building an actual physical robot to hold a controller and press the buttons correctly, and having a controller that lends itself to that kind of responsiveness reliably (or that will even physically permit a simultaneous press of L and R on the d-pad, say) would be a whole other engineering challenge of the sort that would make this video just plain not exist in all likelihood.
posted by cortex at 1:39 PM on April 4, 2011

I think I ate the wrong mushroom.
posted by tmt at 1:40 PM on April 4, 2011 [1 favorite]

How does someone even come up with this insanity.

Think of TASes more like a puzzle than actually playing a reflex based video game. People who create these things are absolute experts on the game. Not like they have good reflexes, like they know how the game passes information in memory around the console's architecture. The last TAS post I made people were getting angry that the person who was doing the run was somehow cheating, but that's missing the point. You're not playing a game, you're solving an equation. How does the game work? How does it render sprites, detect collision, and manage numbers (lives, energy, etc.)? Once you start to figure that out you have to go the next level up and figure how the levels are designed and decide what the most optimal way to traverse them is.

The fact that these are basically solving this equation times 8, at the same time... it's pretty impressive. One negative thing about TASes: how many cures for cancer (or more likely Wikipedia articles) do they beggar us of?
posted by codacorolla at 1:41 PM on April 4, 2011 [13 favorites]

So, it's sort of a "boo to this thing" on the purist assumption that some guy would have just sat down with a controller and a splitter and manually played the four games simultaneously with human fingers and no help. But it's also a pretty fantastic bit of forensic work and puzzle-solving and hackery AND some guy managing to put it all together.

I guess it's all in the presentation. It's like if someone was like, "Trent Reznor is an amazing one-man band!" and you've never heard of the dude. So you're expecting someone playing multiple instruments at once, and then it turns out it's just this mopey goth dude who plays all the instruments on the recordings of his Nine Inch Nails albums.
posted by explosion at 1:44 PM on April 4, 2011 [2 favorites]

Or to put it another way, it's not that the games are being beaten "using one controller for all four games simultaneously." They're being beaten with zero controllers, and one very precise string of input.

Definitely impressive, but doesn't do what it says on the box.
posted by explosion at 1:46 PM on April 4, 2011 [3 favorites]

Right. This isn't playing a game as we shallow earth organisms consider it. It's feeding signals to an emulator, a brain in a jar in a brain in a box, to manipulate its state. While this is fun and fun to watch, it should not be confused with playing a game sitting on your couch.

And it definitely is cheating; it necessarily uses save states and queries the virtual machines' simulated RAM to obtain information that would be hidden from a human player. It makes for an interesting video, but is a bit further away from the act of honestly playing the game like we sane people would than is obvious from watching it, or the language of the FPP for that matter. (Of possible interest: roguelike "borg" players are specifically designed not to use such information.)
posted by JHarris at 2:02 PM on April 4, 2011

And it definitely is cheating; it necessarily uses save states and queries the virtual machines' simulated RAM to obtain information that would be hidden from a human player. It makes for an interesting video, but is a bit further away from the act of honestly playing the game like we sane people would than is obvious from watching it, or the language of the FPP for that matter. (Of possible interest: roguelike "borg" players are specifically designed not to use such information.)

That's ridiculous. They're different things, and I don't understand why people have a purist attitude and feel the need to shit on TASes and call them cheating. It's like watching a TAS is some strange insult to your personal ability to play video games. Also: these people are as sane as anyone, and probably pretty smart too.

I'm going to stop myself from getting angry about TASes, because my neckbeard isn't that thick, but I think it's rude to shit up a thread because of your weirdo purity standards for video games.
posted by codacorolla at 2:11 PM on April 4, 2011 [6 favorites]

Apparently, the trick to beating any Mario game is to constantly be jumping. Sounds about right.

While playing the last Wii Mario title, I often found myself thinking "Mario Games favor the bold".
posted by sourwookie at 2:12 PM on April 4, 2011

Right around 1:08 in the SMB video, SMB3 Mario is flitting back and forth and grabs a Koopa shell while the rest are ducking or standing still. I suppose on some level, it's possible that the "same" input is being used for all 4 games, but the same input isn't being sent to all 4 games, so boo to this.

It is the same input. The Down control is being held while alternating between Left and Right, probably every frame, so while SMB3 Mario is flitting back and forth, SMB2US Luigi is ducking, which means he can't move. SMB1 and SMB2J are just standing there, which means either those older games can't process the input quickly enough, or Mario doesn't move while Down is being held.

So basically, what 0xFCAF said.
posted by neckro23 at 2:12 PM on April 4, 2011 [2 favorites]

To attempt to head this off at the curb: Regular speedruns try to beat games as fast as humanly possible. Tool-assisted speedruns try to do it as fast as theoretically possible. Both are "legitimate," neither are "cheating," and they each have their own merits.

The post is phrased a bit awkwardly, but it's not a lie -- all the games are played using the exact same set of controller inputs, exploiting quirks of each game's engines where needed.
posted by flatluigi at 2:19 PM on April 4, 2011 [4 favorites]

And it definitely is cheating; it necessarily uses save states and queries the virtual machines' simulated RAM to obtain information that would be hidden from a human player.

You think that's bad, IBM has been brazenly using computers when it plays Chess. And Jeopardy!

I mean, it's absolutely, unquestionably hackery. I really do get the purist disappointment at imagining that this was actually D.A.R.Y.L. sitting there getting his roboskills on, but they call them tool-assisted runs for a reason. There's a certain weird charged nature to the "it doesn't count" nature of the thing—this is the platform gamer's Kobayashi Maru, the whole point is that it's basically impossible but they figured out how to do it anyway.
posted by cortex at 2:21 PM on April 4, 2011 [6 favorites]

sourwookie: "
While playing the last Wii Mario title, I often found myself thinking "Mario Games favor the bold"."

Given that, in college, I was a lot better playing Mario Kart on the N64 when drinking but fairly bad at it if I'd been smoking... and given my reactions to said substances back then... I'd have to say I agree.

As for the thread/videos at hand, please add my vote to the 'exploding head' pile -- even given the knowledge of the technical howz-it-done behind-the-scenes.

And as for the question of "how many awesome cures for cancer" aren't being done because of the brain power spent on this awesome thing -- that doesn't bother me. I only find it depressing that I'm doing neither awesome thing.
posted by MCMikeNamara at 2:25 PM on April 4, 2011

Here's the actual submission on TASVideos. You can download it yourself and run it through, if you want.

Notes:

If you push left and right at the same time, SMB1 and SMB2J will walk toward the right, SMB2 will run toward the right, and SMB3 will stop accelerating. Even if SMB3 jumps, SMB3 won't accelerate.
If you push left or right and down, SMB1, SMB2J, and SMB2 will stop accelerating unless they're mid-jump and SMB3 will run toward the left or right.
In SMB1 and SMB2J, if you start to push A button to jump and when you stop pushing A button even if you stop pushing 1 frame, Mario won't rise up even if you push A button again. On the other hand, in SMB2 and SMB3, if you stop pushing A button for a moment, Mario will rise up when you push A button again.

posted by flatluigi at 2:26 PM on April 4, 2011

I posted the Mega Man X ones a while back. These are insane, can't wait to watch.
posted by BlackLeotardFront at 2:34 PM on April 4, 2011

So he presses left and right at the same time, something that is impossible to do on the real controller? Seems cheap to me.
posted by East Manitoba Regional Junior Kabaddi Champion '94 at 2:43 PM on April 4, 2011

I do have to give them props on thee Mega Man video for that message at the bottom saying "If you think this is being played at the same time, go to" an explanation page about tool-assisted speedruns.

That's ridiculous. They're different things, and I don't understand why people have a purist attitude and feel the need to shit on TASes and call them cheating.

But I agree, they are different things. It is cheating according to the ordinary rules of play, legitimate expectations are being broken; but by the rules of TASing they are valid and in fact essential to the form. The key is in expectations. However, people continue to confuse the two in descriptions, including the one that started off this very thread. They set expectations for the first kind of play, when they should be set for the second.

Tool assisted players may actually not be able to play their targeted games well at all, although I would expect most of them are at least very good for their understanding of the mechanisms that drive play. All hidden information is known, and there is no hand-eye coordination involved here at all. Plus due to the way items are dropped in games like Mega Man, a TAS player can generate any random pickups he wants.

Given that, in college, I was a lot better playing Mario Kart on the N64 when drinking but fairly bad at it if I'd been smoking... and given my reactions to said substances back then... I'd have to say I agree.

Oh man, Mario Kart battle modes look random, but can become such intense fights of wit and cunning that it's almost not fun to play anymore -- at least when I am losing.
posted by JHarris at 2:46 PM on April 4, 2011

GOD Final Fantasy VI is great.

Also, Mario Kart on SNES is way better than on N64, JHarris.
posted by BlackLeotardFront at 2:49 PM on April 4, 2011 [1 favorite]

I found myself far less interested in how he beat two Final Fantasy games at the same time as I was at how he managed to beat VI with six characters at level 10 or so.

I wish it wasn't in Japanese so I could gauge some of the glitches/tricks he used (seemed to use a lot of "let's refresh this state until Setzer casts Death successfully").
posted by XQUZYPHYR at 2:56 PM on April 4, 2011 [1 favorite]

But I agree, they are different things. It is cheating according to the ordinary rules of play, legitimate expectations are being broken; but by the rules of TASing they are valid and in fact essential to the form.

Totally. I think there's maybe an issue here that "cheating" carries a degree of moral charge with it that can suggest something other than a technical distinction. It's a cheat in the sense that stage magic or professional wrestling is a cheat: if you thought it was straight-faced, you can understandably feel like you've had your leg pulled, but that's as far as it goes.

So he presses left and right at the same time, something that is impossible to do on the real controller? Seems cheap to me.

My question is, if you broke the arms of a NES controller's d-pad such that you could press both at once, would the controller successfully read and transmit that state, or was the modal nature of U/D, L/R enshrined in the hardware by either intent or wiring limitations?

But then, the standard NES controller and the Advantage were very different physical experiences too. Setting aside even the question of turbo buttons. Is a run done with an Advantage a cheat? What about a modern USB gamepad that's functionally similar to a NES gamepad but is better machined or more responsive or etc?

I mostly am interested in the "what I did and how I did it" aspects of this stuff, and I dig that other people will come at it from different perspectives. But unless there's tourney money on the line I'm not sure why it needs to matter.
posted by cortex at 3:01 PM on April 4, 2011 [2 favorites]

You can actually, physically press both left and right on NES controllers.
posted by flatluigi at 3:26 PM on April 4, 2011

My question is, if you broke the arms of a NES controller's d-pad such that you could press both at once, would the controller successfully read and transmit that state, or was the modal nature of U/D, L/R enshrined in the hardware by either intent or wiring limitations?

My understand is that the controller would indeed be able to send that state.

(I haven't actually done the experiment, but the wire protocol between the console and the controller sends the button states individually for each button, and there are separate wires going from each button to the shift register chip, so I assume it's possible.)
posted by jcreigh at 3:27 PM on April 4, 2011

But then, the standard NES controller and the Advantage were very different physical experiences too. Setting aside even the question of turbo buttons. Is a run done with an Advantage a cheat?

The purpose of playing a game is, to a degree, divorced from the controller you use. It's to manipulate on-screen figures, not to push buttons (with some mostly obvious exceptions, such as Hyper Olympics/Track & Field). What is cheating, according to the rules of straight videogaming (not TAS-ing) is automation, whether that's produced by the controller or an algorithm in an emulator that plays back a string of inputs. That encompasses turbo buttons (automating pressing and releasing a button) and slow motion features (automating pressing and releasing the start button).

A well-made controller can help a little (although from what I've seen few are better than the NES Control Pad, including those pads supplied with some current game consoles), but there is a point beyond which controllers cannot help, where the speed of that mysterious wisp, the human spark of impulse, is what matters. A controller that helps in that regard is certainly cheating.

That leaves pressing opposite directions at once, which is something that doesn't necessarily have to be automated if your (possibly broken) controller supports it. It may still be cheating (Nintendo would certainly claim that it was), but it's more of a judgement call. The intent of the developer, who set the game and decide the rules, is arguably important in the general case, but some individuals or groups might disagree for their own purposes.
posted by JHarris at 4:23 PM on April 4, 2011

I'll quote from the TASVideos "cheat FAQ" (apparently now taken down):

Arguably, we do cheat the games. The games have been programmed assuming that the player is a human who has humanly limits. We surpass those limits with tool-assistance.

For example, there are very few humans who can produce a 30 Hz autofire by rapidly bashing the B button, but in tool-assistance, it is trivial to do that. We also abuse many programming errors in the games.

But we do not cheat the audience. The makers of tool-assisted speedruns are very open at how they made the movies. There certainly is no deceiving taking part. It is the audience's responsibility to read and understand how these movies are made.

As for rules, tool-assisted speedruns fall under the rules of tool-assisted speedruns, not under the rules of non-assisted speedruns. The purpose of tool-assisted speedruns is different from regular speedruns and there is no competition between these two different kinds of runs, so the set of rules is not, and does not have to be, the same.

We try to keep the difference from the games themselves minimal. For example, our rules dictate that the movies we make must be of non-modified games. We do not crack and hack the games. Basically, the only thing we do is produce unexpected kind of input to the real games, as if a god were playing the game.
posted by teraflop at 4:26 PM on April 4, 2011 [5 favorites]

It's more like blowing up a Death Star than bullseye'ing womp rats in your T-16 back home...

The feat isn't as impressive as the fact that the feat is even possible, with or without your dead guidance counselor and some magic bugs in your bloodstream.
posted by pokermonk at 4:35 PM on April 4, 2011 [1 favorite]

So he presses left and right at the same time, something that is impossible to do on the real controller? Seems cheap to me.

If you had a robot you might be able to simulate it with a robot pressing left and right really quickly.

The next step would be to do this without taking advantage of glitches, meaning, something a normal person could do with a real controller.
posted by delmoi at 4:44 PM on April 4, 2011

A member of the TAS community actually modded an NES controller to play back TASes on the real, physical hardware. See: http://www.youtube.com/watch?v=52xBCinOLSs&feature=player_embedded
posted by LSK at 4:58 PM on April 4, 2011 [2 favorites]

I'm also not sure how you were under any impression that is something other than a TAS, when the first thing under "more inside" is the source TAS Videos. Does every post about TAS have to have explanatory material so as to not dupe the audience in to thinking it's a human with a controller?
posted by codacorolla at 5:06 PM on April 4, 2011

Does every post about TAS have to have explanatory material so as to not dupe the audience in to thinking it's a human with a controller?

Never heard of them before. Also, the language in the OP doesn't just fail to provide explanatory material, it assumes that most readers know that "controller", when it comes to video games, means something other than this.
posted by muddgirl at 5:21 PM on April 4, 2011 [2 favorites]

I really don't understand the idea that TAS is cheating. There's no deception involved, they're no more cheating than a stage magician is cheating at poker when he deals himself four aces (less, because they show you how the trick was done). There's "ordinary rules of play" of card games as well, but no-one comes into posts about card tricks with accusations of cheating.
posted by Proofs and Refutations at 5:22 PM on April 4, 2011

To be fair, "TAS" is utterly opaque to people who don't already know about this stuff. "Tool-assisted" at least makes it clear that there's something going on there to people unfamiliar with speedruns and such, so for the sake of clarity it's probably a decent idea to be a little more explicit when posting to avoid confusion. Not a big deal, but worth factoring in when considering a post.
posted by cortex at 5:24 PM on April 4, 2011 [1 favorite]

codacorolla: Assuming you were directing that at me: the "TAS" signifier behind the fold was overwhelmed by the "with one controller" right there on the post. It seems relevant considering you don't have to click through to see the videos. Scanning the earlier comments shows it wasn't just me.

However, I am beginning to think I need to clarify what I mean by "cheating" every time I comment in the same thread:

Proofs and Refutations:
It unquestionably is cheating according to the normal rules of play because it so overwhelmingly violates designer, and normal player, expectations. Speedruns aren't made by recording actual controller inputs; they're constructed as one might compose a music score in a sequencer, the author deciding where impulses go and how long they should last. It isn't cheating according to the rules of TASes, literally "Tool Assisted Speedruns." A machine like a robot might be able to reproduce these things but no unaided human ever could.

Everything I touch jumps the track. SIGH
posted by JHarris at 5:31 PM on April 4, 2011

I like how the music doesn't clash in the FF TAS. And the Mario one made me realize I'd get killed in all 4 games in 10 seconds, my reflexes aren't up to par.
posted by dragonplayer at 5:36 PM on April 4, 2011

I was in the French Quarter early in my freshman year at Tulane when this dude walks up to me.

"I'll bet you a dollar I can throw this empty can over my back and make it land on either the top or the bottom."

"All right," I say. "I'll take that bet."

The man sets the can down on the ground and, *crunch*, he flattens the can with his boot so that it's half an inch thick. Then he picks it up, tosses it over his head and, without bothering to look at the can, asks me for my dollar.

TASes kinda remind me of that.
posted by bpm140 at 5:40 PM on April 4, 2011

I think it takes a little bit of the magic out of it to explain upfront what TAS is to a newcomer. Everyone should have the experience I had ~8 years ago watching the SMB3 TAS speedrun where my roommates and I crowded around the monitor debating whether or not it could have been done by a human.

Compare and contrast The Turk: A chess-playing machine that is supposedly an automaton, but is actually controlled by a human hiding inside. In 1770, a machine which could beat a human in chess would literally defy belief, but the fact that a human could be hidden inside the machine, operate it without being able to directly see the board, and be a strong chess player is still extremely impressive. If you just said "Hey there's a dude in the machine but check out the machine" it'd lose a lot of the mystique.
posted by 0xFCAF at 5:48 PM on April 4, 2011

JHarris:
I still don't see how your criteria wouldn't define many magic tricks as "cheating". For that matter TAS can only violate the expectations of players who don't know what it is, they go to great pains to describe what they are doing.

In my mind for something to be "cheating" there is a necessary component of a broken promise. A "cheater" has agreed to a code of conduct and then violated it. The TAS community make very explicit what code of conduct they adhere to and don't violate it. Thus when I see TAS being described "cheating" I get my hackles up because in my mind it carries and implicit accusation of dishonesty.

I agree that the post world have been improved by making clearer that this was a TAS post rather than an ordinary playthrough, I would not have put that information below the fold.
posted by Proofs and Refutations at 5:57 PM on April 4, 2011

I still don't see how your criteria wouldn't define many magic tricks as "cheating".

There are two different problems with your summation.

The first:
If you want to push it that far then yes, many magic tricks are cheating. However, they are usually announced as magic tricks. They are cheating in that you aren't really pulling a quarter out of someone's ear. But we have a cultural understanding of magic tricks, and the understanding that they are just fooling around and, for example, not worthy of calling in the Inquisition. There was a time when you didn't even want to hint that something might be magic; fortunately we're over that now, but in the process of getting over it we invented a new form of entertainment.

But magicians can't pull off the kinds of tricks exhibited here. They can only appear to. That brings us to the second:

Magic tricks are not the same thing at all; as far as we know, the laws of physics don't completely ignore what you're doing on even or odd moments, or cause you to walk through walls if you try to go in two directions at once, or any similar kind of reality-shattering phenomenon. These things are possible because the people who compose TASes have access to a higher order of reality than that of the game, the one that emulates the machine. They can use that access to gain insight into the game state, to figure out holes in the laws of physics (which are much more numerous anyway than those in our own reality), and take advantage of those. Magicians are unable to accomplish any of that; if they could, there are a lot of physicists who would have questions for them.
posted by JHarris at 6:14 PM on April 4, 2011

In my mind for something to be "cheating" there is a necessary component of a broken promise.

Okay, how about this. Many games rely on hidden information, knowledge that is kept from the player. But computers like the chip that powers the NES are deterministic machines, that cannot behave randomly as a fact of their operation. Events such as which item an enemy will drop are actually determined by the consequences of the ultra-precise timing of game inputs. Some of these inputs aren't even part of the "game," such as the number of elapsed frames from powerup to the game-starting keypress.

Peeking at the game state to determine things like this is obviously cheating, because the information is supposed to be hidden from the player. Else why even have random drops? But speedrunners not only peek at, they manipulate this data by choosing exactly the right timing for button presses to generate the items they want. What speedrunners do make a mockery of the whole idea, because what they do is not how the game was intended to be played.

My gaming history contains a lot of roguelikes, and these things may be more obvious to me than others. In a roguelike, you aren't even allowed to go back to an old save after you die without being called a cheater, and this is a perfectly valid given the foundation of expectations upon which those games are built. But similarly, games like Mega Man were not designed to be played with savestates, or knowledge of the game's pseudo-random generators, or frame-level manipulation of game timing, or pressing left and right at the same time. You are playing a different game at that level.

And that's okay. Except you're playing the game of tool-assisted speedrunning. Not the game that many people were expecting.
posted by JHarris at 6:24 PM on April 4, 2011

Did we break tasvideos.org, or has this multi-game-run made other sites at about the same time?
posted by jepler at 6:56 PM on April 4, 2011

JHarris:

I agree with most of your second comment after my last. I just suppose I don't agree that "Playing a different game" = "Cheating" any more than "Playing Draughts (Checkers) with a Chess set" = "Cheating". After all the TAS community makes it clear that they're playing the game of tool-assisted speedrunning. I don't see how it's a response to what you quoted however, what promise have the TAS community made and broken? Sure they don't play the game the way the designer intended (which is unknowable anyway) they never said they were.

For that matter neither do professional tournament players of many competitive video games (Marvel Vs Capcom 2, has several tournament-legal glitches). Hard to argue those guys are cheating, though.
posted by Proofs and Refutations at 7:04 PM on April 4, 2011 [1 favorite]

This is sort of a pedantic point to make, but the Mega Man X/X2 run was, if I recall correctly, made using an actual gamepad. It involved some custom software to tee the controller inputs to two processes, and a modification to the emulator to read the controller input from a pipe. He had three extra controller buttons mapped to save state, load state, and frame advance. So, yes, one controller, for both games simultaneously, very, very slowly.
posted by Kalthare at 8:33 PM on April 4, 2011

Anyway, watching the mario run(s) again makes me wonder what sort of shit Mario did to end up repeating the same nightmarish hell for 20 some odd years running. Fighting man-sized turtles, collecting meaningless wealth, falling into lava, and trying to reach his true love that stays forever out of reach. That's a cruel God, right there.
posted by codacorolla at 8:41 PM on April 4, 2011 [3 favorites]

I just suppose I don't agree that "Playing a different game" = "Cheating" any more than "Playing Draughts (Checkers) with a Chess set" = "Cheating".

It is, if you sometimes follow the rules to chess during your checkers game!

After all the TAS community makes it clear that they're playing the game of tool-assisted speedrunning. I don't see how it's a response to what you quoted however, what promise have the TAS community made and broken?

The implicit promise that it was a person sitting down and playing the game normally. But it is solitaire, and it is okay to cheat at solitaire, so long as you don't then show a video of you playing solitaire well and claim you were playing by the rules. Which the TAS people have NOT done, but this post does, slightly. Are we on the same page now?

This is sort of a pedantic point to make

Go ahead, why should we have all the "fun."
posted by JHarris at 9:30 PM on April 4, 2011

I find this fascinating. I'm not a very good gamer, but I love that people are pushing the limits and playing with what the game itself can do. And games like Mario and Mega Man are the classics of the gaming world, a lingua franca that still informs game design today. Most people don't play games with the constraints that Roguelikes demand. These videos are just removing more and more constraints from the 'player'.
posted by Lovecraft In Brooklyn at 12:26 AM on April 5, 2011

I hope this exists somewhere:

Bug: When right and left are pressed at the same time, Mario runs to the right.
Expected Result: Mario should not move.
Resolution: Closed, not a bug. This can never happen.

posted by Sibrax at 11:18 AM on April 5, 2011 [1 favorite]

Golfhaus: "So, what the heck is going on in SMB and SMB2J where Mario is shimmying through the walls?"

There are ways to make this no clipping mode happen in the game apparently, using a Game Genie code or otherwise. There are a few items like the "neverending water world" negative level than can be accessed if you do it right. I don't remember if my brothers and I ever managed to make it work, but the water world sounds familiar at any rate.
posted by caution live frogs at 1:37 PM on April 5, 2011

Pulled of World -1 exactly once. It took for fucking ever leaping and leaping and leaping against that warp pipe wall in 1-2, running through a bunch of guys trying to get it to happen, to the point where I started to believe it was just a hoax after all, and then, suddenly, I got it! I was through the wall! There was the pipe! I was in World -1!

And I swam and I swam and I swam and it never ended and I died.
posted by cortex at 2:23 PM on April 5, 2011

For those wondering they got through the walls:

Explanations for tricks and glitches used in Super Mario Bros, Super Mario Bros 2, and Super Mario Bros 3. There are multiple walking-through-walls methods, some of which are quite doable in realtime on real hardware, no Game Genie required.

The Minus World is actually World 36-1. It's reached by going through the wall, so as to enter the warp pipe before the level numbers get displayed.
posted by Kalthare at 4:22 PM on April 5, 2011

I actually got quite good at pulling off the minus world. I remember you having to do kinda a "hook" motion when jumping against the brick to get it to go through the wall. It became kinda like a party trick. "Oh, you haven't seen the minus world? Let me show you!"

But then, once you are there, you realize the payoff for all the effort sucks. It's like a really good story with a bad ending. The journey becomes the cool part.
posted by mysterpigg at 6:12 AM on April 7, 2011