I am now aware of that which has taken place here. I am rather upset at the loss. Actually, potential loss -- It isn't lost yet, but is certainly headed that direction.
I wish to argue for keeping this site running. It may well be possible that everything I say here will not impact the ultimate result (closure of site), but I'm certain that all I say here will be noticed.
Your argument (the collective of the admins and judges I observe speaking here before this was opened to public), as far as the concept I have received from reading this thread, is that the site must represent perfection, and the emulation can't achieve this perfection. With conflicting goal and possibility, the site should dissolve.
I will soon attempt to attack this reasoning through my own belief of emulation. Through this, I will attempt to inject a different thought, one that no one else has considered. I will also pose a question about emulation of the Next-Gen consoles, which may well reveal that the issue does not show up in that field, or it may end up further confirming what is feared.
Finally, I will move on to the community itself. They are as good a reason as anything to bring up, and they do have a large impact even outside of Tool-Assisted Superplays.
Now that I said what I will say, I shall go on to say it.
The core randomness of the actual consoles. The timing wanders. Reading from uninitialized memory. All sorts of systems for randomness within the hardware itself. Successfully emulating this randomness will prevent movie files from working (without a measure to reproduce the same randomness from the movie file), yet it is this randomness that we don't emulate with complete precision. Hence, everything we do is inherently flawed.
But the argument that I see is that the emulator can't match the console without going into the utterly bizarre. Perfectly emulate the sound, video, each and every opcode, defined or not, everything else, but the hardware itself still is random.
What I ask is not that we must match the emulator to the console. This is asking for far too much. However, I will ask, is it possible, with all this randomness, that the console itself can match the emulator, however slim?
The impression I got from each and every TAS is that, given perfect conditions, how fast we can complete each game. Given perfect conditions! All the randomness lined up to make our input file work, the timing is spot on, and the memory at power-on had initial values that were exactly that. What's so impossible here that the emulator is doing that the console can't do?
I will predict one question that comes up that will poke a hole in my argument: Couldn't someone adjust the emulator to a certain, highly favorable randomness that does utterly break some games and lets them complete it far faster than what should be possible? Oh, no, it's not outside the realm of possibility for the console itself. I was just lucky, how can you prove otherwise?
My response to that? I have no valid argument. I must concede at this point. But is this point enough to break the rest of my argument? The best I can say is that the emulator should attempt to produce the average case, whatever the average can be defined as.
I hear all about the timing crystals and all that for older consoles. What about newer ones? XBox360, Wii, PS3, DSi? Are they subject to the exact same rules of hardware randomness that the older consoles have? Maybe. I don't know the details. If they are more "protected" from hardware randomness, then perhaps these consoles can still be emulated "perfectly."
If so, TASVideos can still have a future through those consoles, given enough time for the emulators to finally develop. If not, your original point stands.
I have put forth my thoughts on emulation, and a thought that it's not the fact we must match the console, but can the console match the emulation? I have also put forth a question about later consoles.
But I don't feel the emulation itself is the whole issue. There is also a community here, and there has definitely been an impact on things that aren't a TAS. I note on the front page: "I've spent the past few days figuring out how to word this, and I still don't know" -- Are words a problem now?
A great many people have come. One way or another, someone has successfully brought many people together into one site. Is it right to tell them to leave all because you declare that one concept is flawed? Granted, it's the founding concept of the site, but it is not the sole concept that everyone unanimously shares when coming here. I can say this believing my own concept and reasons may well be different from the site's.
There has also been some impact in other areas. Actual speedrunners who actually do runs on the console can use some of the techniques discovered through TASing. Losing this resource (I am under the impression that TASVideos is a significant resource here) will certainly affect them. Obviously, what we're doing here has enough similarity to the actual console, correct?
TASing can involve disassembling the game. We develop emulator tools for that. TASing might need assistance from scripts. Lua is added to it. Related communities benefit from our work on these improvements on our road to TASing.
Even if the emulation itself is never perfect for TASing, we are a benefit to others that work on their own things. We are worrying far too much over perfection to.
If there's a new concept that you want to have for the site, don't make English words a problem. Make it implicit until someone can come up with an especially well-fitting phrase that works for it. If it's all down to a single sentence that determines whether the whole site lives, then that isn't a useful foundation for the site.
I suggest that you leave language out of whether the site lives.
I have given my thoughts on community. And my thoughts on words with the concept, which I think is silly. Maybe I'm completely wrong about the community. Please correct me if I am.
I may have been overzealous in defending the site, to keep it running indefinitely, in spite of my habits not showing the sort of activity that would call for this sort of argument from me. All my writing here can attest to that. I can always turn around and never look back to this site, and my own life will not be impacted much. But fact is, I want to keep the site going. I wouldn't know how to revive the site should it go down.
If the decision here is final, fine. I am saddened, but I will simply move on. Scarcely any further mention from me of TASing, and almost certainly no drive to make any further attempts. That is what I believe I will do.
It's a nice black ball right? Quite clearly.
However, on our real console when we turn on from a cold start we get this:
Even more bizarrely, when we do it again we get this:
What's going on? The emulator is executing the game's code 100% correctly. Where are the colours coming from?
What's happening is the game is initializing a ball and getting the colour information from a source or part of the hardware that is constantly random. An entropy, if you will. It's possible it's reading such information entirely from uninitialized RAM which may have left over residue information from a previously played game. This is how Banjo Kazooie's Stop 'N' Swop was going to work.
How far does this stretch? Almost every game, and it causes the entirety of our work to be invalid under our single premise of theoretically working on real hardware.
