once you can program into the games RAM.

You make this sound like a trivial, yawn-worthy event. I'm not sure if it's clear that this is the second game ever, and first game on the SNES, that this has been done. Please correct me if I'm wrong on those numbers.

Let me start off by apologizing for raining on everyone's parade, and say that I think this is a cool video. I have to vote no because this video is in-optimal. It was cute putting in a Snake and Pong game, but why play them? It just wastes time that could've been better spent jumping directly to the end on the menu. If we say that Pong and Snake are trade-offs for entertainment, why are they played so poorly? At the very least, in Snake, the randomness should've been manipulated, and the game played till the snake filled up the entire screen. In good judgment, I don't see where we can accept this. It's slower than it should be, and the trade-offs could be better. Great demonstration video for gruefood delight.

In good judgment, I don't see where we can accept this. It's slower than it should be, and the trade-offs could be better. Great demonstration video for gruefood delight.

OH MY YES Here's where I want to switch to No vote. SNAKE could be better. A full gameplay of it, maxing out and then loosing will be more entertaining. I cannot comment on PONG, I say another game should've taken it's place.

Wow! I honestly don't know what else to write right now. I think this opens a new chapter somehow. Great job!

Soon TASes will be subject to copyright law. Oh dear.

I'm not sure if it's clear that this is the second game ever, and first game on the SNES, that this has been done. Please correct me if I'm wrong on those numbers.

Yes, although from my understanding, Yoshi's Island total control is also possible since the glitched run jumps to controller data. It may also be possible with Pokémon Gold by using a similar method as #4126: TheZZAZZGlitch's GBC Pokémon: Gold Version "game end glitch" in 59:36.02. I can't wait until more software exploits like this are found though! It's the next best thing after normal TASes! :)

I have to vote no because this video is in-optimal. It was cute putting in a Snake and Pong game, but why play them? It just wastes time that could've been better spent jumping directly to the end on the menu. If we say that Pong and Snake are trade-offs for entertainment, why are they played so poorly? At the very least, in Snake, the randomness should've been manipulated, and the game played till the snake filled up the entire screen.

This TAS doesn't aim for fastest speed, the only objective is the entertainment, so I guess that jumping directly to the end would be a bad thing, because he wouldn't show the games he created.

In comparison to something like this. There is a huge difference. Pi was done for it's time entertainment trade-off and it optimization over that full stop up there. Sure, pi is possibly unoptimized but this in comparison to what happened in snake is nothing. All that happened in snake was a "d" then "→" then "THE END". Snake can be better. If snake was only shown for demonstration purposes to the people, that makes me more upset about this run. I don't really have any comments on the performance of demonstrating PONG.

haha, I can’t believe this run is getting shit from some people Snake was obviously played just to demonstrate that the coding was legitimate, and not just a screen with 2 options that did nothing; it wasn’t played up to the end because the focal point of this TAS is reprogramming the game through controller input, not wasting viewer’s time to TAS a Snake game which could probably be done in a much more entertaining way on an actual original Snake game

haha, I can’t believe this run is getting shit from some people

Disagreeing does not equal shitting on something. Don't be so juvenile.

I am hoping that this will eventually be obsoleted by a run with better payload. Until then, I think this deserves publication. Not only is it awesome enough, but its display in agdq2014 makes it an important milestone for both our communities, and it'd be sad to see it buried in the rejected submissions archive. I also suggest linking the agdq presentation along with the publication.

I am hoping that this will eventually be obsoleted by a run with better payload.

Actually. I will agree with you on that one. [2187] GBC Pokémon: Yellow Version "arbitrary code execution" by bortreb in 12:51.87

Suddenly I'm requesting the snake minigame be played with better luck (apples every block) and snake filling the whole screen. Because without apples every frame it would take too much time. What do you think?

Well things got a bit ugly here didn't they? The first couple of pages of comments are mostly "zomg wtf this is da best tas EVA masterjun you fking god"-type comments, and now there's this big old argument about whether or not it should even be published. The problem is that arbitrary execution TASes essentially come in two parts. The first part is internally gaining control of the software to allow you to execute code, the second is the code you choose to execute. The relative importance of these is different for different people. For some people, the very fact that code can be executed is what is so fascinating, and what is done with it is a minor consideration, while for others, the payload can make or break the TAS. Each group has a complaint. For those focusing on the point before code execution only, up to that point the TAS is identical to the previous SMW TAS and as a result this run offers nothing new. For those interested in the payload, well, they can find the payload boring, although obviously some will be entertained by the payload and honestly I feel this would have been much more entertaining in the context of AGDQ, especially from the point of view of a watching real-time runner unaware of the concept of arbitrary code execution. Personally, I am more intrigued by the ability to execute code rather than what is done with it, as the whole point is that literally anything can be done once you have got the game into that state, so seeing what somebody picks isn't that impressive. I want to see this published, but what I do not want to see is ten other people subsequently copying masterjun's SMW run and using it to execute their own code. because none of this really adds anything to TASing as a whole.

The snake game softlocks if the snake is at full lenght and eat the last apple. You can do anything, not even get back to the menu, the snake doesn't die either. I testet this out yesterday night.

Then it must avoid the last apple only and die from something.

While I agree that this is a massively important run for the TAS community, the run itself does not bring or show much new. The first Pokemon total control was huge. The improved Pi version was also quite nice, but at that point, the total control hacks reached the limit. Once you can program anything you want, what makes Snake more impressive than a Pi showcase?

I think that from a layman spectator's point of view this is much better than either of those. Showing a static picture and playing some simple music is ok alright, but it's rather... well, static. Nothing much happens. As for the Pi hack, it just shows random numbers on screen. Personally I thought it was completely sub-par in comparison and didn't make any kind of sense. This, however, shows something actually happening on screen. Some activity, some game logic, and using something that even the casual viewer will understand what's happening. Thus I consider this idea much superior in this sense.

In good judgment, I don't see where we can accept this. It's slower than it should be, and the trade-offs could be better. Great demonstration video for gruefood delight.

I hope this was a joke.

Showing a static picture and playing some simple music is ok alright, but it's rather... well, static. Nothing much happens. As for the Pi hack, it just shows random numbers on screen. Personally I thought it was completely sub-par in comparison and didn't make any kind of sense.

Stopping you riiiiiiiiiiiiiiiiiiiiiiiiiight there. Ponies. Even though it was a long run because it was completely unknown where the total control could happen til pi tas came up, it was a run which showed more values that are unkown and it played the My Little Pony theme song, with what people believed it was 3 balloons until they figured out it was Pinkie's "cutie mark". Pi day in comparison was going to the beat of the music to bring up the numbers and not random numbers the digits of pi, with random stuff going at the bottom of the screen. Unless this is the point you're trying to get across then whoops.

It makes me wonder; what if we could do something like that in every game? How crazy!

Oh wow, how could I miss this! It’s incredible, I love it ♥ yes vote, of course… Is it the first time when the TASer succeeded to inject another game to a given one?

It makes me wonder; what if we could do something like that in every game? How crazy!

It might be possible to pull this off in Chrono Trigger. "Executes Arbitrary Code" is a rare category that I would love to see more of. :D You know, the only thing that could make this run more perfect is that, instead of Mario's head in Pong, it should have a picture of Bowser's balls to show how much Mario owns this game. XD

I don't think the video will appeal the average viewer more, if you reprogram that snake program so that the next apple always appears one field in front of your head. It wouldn't even look like snake to most people.

In good judgment, I don't see where we can accept this. It's slower than it should be, and the trade-offs could be better. Great demonstration video for gruefood delight.

I hope this was a joke.

Why should it be a joke? I genuinely believe that this is probably the greatest TAS ever made. However, that doesn't mean it doesn't include what looks like some sloppy play. The payload is playable, since it's playable, it falls into one of two categories: Speed - The fastest way to complete this is to jump straight to the end. Entertaining - To fill this with entertainment, the play of the games present should appear to be super human and not contain amateurish gameplay mistakes. Since it took the time to play the two games, then it should play them as expertly as possible. The Snake game should be played trying to manipulate luck if at all possible, or with foreknowledge of what the fixed apple appearing pattern is. Then proceeds to get the snake as long as possible as quick as possible. Do you honestly think the games are played "perfectly"? If so, explain why. If not, why do you think this should be ignored? Do you see another option here other than speed or entertainment? Our rules state that what looks like sloppy play to the average viewer is grounds for rejection. This run is an amazing accomplishment and one of the greatest things I've ever seen, but that doesn't mean we should overlook its faults. TASVideos is about getting as close to perfection as possible. To attack this problem from a different angle. Say we did accept this. Someone could come along tomorrow, start rerecording from the game menu, and then play Snake and Pong better and submit, and we'd be force to accept as an obsoletion. In fact, you can do that right now and still submit it while this is on the queue, and since those two games are simple, well, it'd be a rather simple way to *improve* this run right now. I imagine a bunch of our members are considering doing this as we speak, if they aren't already.

Since it took the time to play the two games, then it should play them as expertly as possible. The Snake game should be played trying to manipulate luck if at all possible, or with foreknowledge of what the fixed apple appearing pattern is. Then proceeds to get the snake as long as possible as quick as possible.

But the purpose of playing them at all was to simply show that they are playable games. Not to beat them. And using TAS precision on a game that you just programmed, might not actually be the best possible way to demonstrate that the games are playable.

This is a precedent-setting run in one very specific way - after the initial game has been TAS'ed and total control is initially established the payload (in this case, the target game) is programmed; as soon as *that* controller input is complete and the new payload starts executing, the controllers can be completely removed in a physical console verification scenario. This means that whatever follows would be a new game in its own right which could then be TAS'ed. Now, if someone figures out a way to perform controller input on the game within the game that causes arbitrary code to be executed, you could then create a new payload to create a game within a game within a game (in the same vein as Inception). For those concerned that the game-within-the-game (specifically Snake) was not TAS'ed optimally, please note that the primary purpose of this TAS was to demonstrate total control in front of a live audience during AGDQ 2014. During the development process we even talked about popping up an infobox along the lines of "Hi AGDQ 2014! We have total control of this console. Here is what we can do with it." The payload that Masterjun settled on was a good balance between providing a payload that was appropriate in length for the time we were alloted in our block as well as something that was clearly arbitrary code that crated a standalone game which could then be played. The payload Masterjun crated intentionally included an "insert coins" arcade game style demo of the gameplay, which is what people are arguing about here; most of those style demos of arcade games from the 80's demonstrated "suboptimal" play much the way this demo did, but when the user takes control themselves (which *can* be done immediately after the payload is fully executed; perhaps Masterjun can provide the exact frame number for this) the human player can play as optimally as they like, as TASEditor appears to have done. In my opinion, the payload for AGDQ was perfect - being able to have Tompa, who had no idea what was happening, compete against micro500 in Pong was absolutely amazing and added to the experience. Had the movie been designed with only a TASVideos submission in mind, I suspect Masterjun would have presented the payload differently. In conclusion, I feel that a TAS of the payload of a game-within-a-game is its own TAS and while the arguments put forth above have merit they should be argued within the context of what we now know is possible, not just for this run but for future runs with different payload. Thanks for reading this tome and I'd love to know what others think! A.C. ******