We could sit here all day and dream up scenarios that handicap the 2700. The only possible way a 1300 could win.
That's how humans are, though. Where do we draw the line? If you think about it, who plays at their exact rating? Maybe a 2700 will play like a 2697 on one day, and a 2699.98 player on another. In fact, a 2700 will rarely play exactly like a 2700, because that would require perfect consistency, which is something that humans just don't have.
Using the logic of Elubas, that anything is possible, the worst player ever, with a rating of 99 could eventually win. This is absurd
You said it yourself one time, binary. Ratings are a relative measure. So all you know for absolute sure is that a 2700 just does a lot better than people rated a lot less than him. Obviously that means that you can say a much lower rated player beating him is unlikely... but can you truly say anything beyond that? 
Maybe there is some kind of magic power the 2700 has, but that can't be proven just by saying "he has a high rating" because all that literally means is "he has better results than those lower than him." It says nothing about why he has better results.
Yeah, so we'd have to agree on what does and doesn't make up a human chess player. There are common elements like known positions, relatively short sequences of calculation, evaluation, etc.
Moves that make no sense AND appear fundamentally bad (like knight takes pawn, when it's defended by another pawn) will be considered 0% of the time for example.
Now, if the sacrifice is part of a known theme or idea, like freeing up a piece or square or removing a defender, creating a passed pawn, etc, then sure, it might be considered at least in a fantasy variation.
You might say it could be considered, but honestly, think back to your last tournament game. Surely over 50% of legal moves (1 ply deep) weren't considered by you. Maybe as high as 80 or 90%.
So I have a feeling that when we say a person doesn't play their rating, what we're really saying is that, based on their rating, they didn't play how you predicted. They played very far from how you predicted they will play. And sometimes it's so far from your prediction that you get confused and question the legitimacy of the ratings and players :)
But what's really happening is that, the ratings don't give perfect information on how a person will play -- so it's possible you might be shocked by their play in some rare cases. Humans will never play in the exact same way on any given day, that's just how they are. So if we're talking about a human 2700, you simply have to talk about variance. In fact you have to regardless of how close or lopsided the matchup is. Humans always vary, and their rating is just an average of that.
Even the current 2700 players aren't that way because their mind literally restricts certain bad moves from happening. They're just very unlikely to play certain kinds of bad moves. But they still can, and have.
All we know from watching 2700s play is that they are likely to do certain kinds of things and unlikely to do others. That doesn't mean they can't do these things.
Humans will be humans regardless of what rating system we create. You can't physically get rid of having a blind spot. All you can do is measure how rarely it happens :)
Ah, but it's not random, it just appears random. The same way flipping a coin isn't random. The same way oversights aren't random. It appears all moves are possible, but they aren't. I'll argue that stable ratings are evidence of this. (At least when we're assuming both players are serious and trying to win or draw.)
Going by your coin example, what I'm saying might be something like, technically, every coin flip ever done could have been heads. But in practice it's 50/50. But it didn't have to be that way -- there isn't this magic force that keeps that from happening. It just turned out that way.
And of course when I say random, I mean, based on the information we have. I'm saying that the knowledge we have about what 2700s are like, still nevertheless don't rule out them getting outplayed by a 1300. It just means, you'd be really stupid to ever bet on it.
I'm saying the knowledge we don't have may reveal that for two specific players (one 2700 and one 1300) it may be impossible.
I think it will be impossible for a much great number of pairs than in the 1 move mate blunder example. That's why I'd stress the possibility of impossibility in the outplaying example!
Using the logic of Elubas, that anything is possible, the worst player ever, with a rating of 99 could eventually win. This is absurd
I suspect so, yes. Unless it can be one of those "possible to be impossible" moments. I have respect for that argument, but personally, I do think a 99 could win. As you might imagine, his chances wouldn't be too good, though :) Incomprehensibly bad.
You said it yourself one time, binary. Ratings are a relative measure. So all you know for absolute sure is that a 2700 just does a lot better than people rated a lot less than him. Obviously that means that you can say a much lower rated player beating him is unlikely... but can you truly say anything beyond that? Maybe there is some kind of magic power the 2700 has, but that can't be proven just by saying "he has a high rating" because all that literally means is "he has better results than those lower than him." It says nothing about why he has better results.
Yeah, so we'd have to agree on what does and doesn't make up a human chess player. There are common elements like known positions, relatively short sequences of calculation, evaluation, etc.
Moves that make no sense AND appear fundamentally bad (like knight takes pawn, when it's defended by another pawn) will be considered 0% of the time for example.
I still wouldn't say zero, though. Maybe the way I'm feeling on a particular day (or just how I feel about the position) will get me to rationalize a ridiculous move. Very probably not. Maybe unlikely enough that it'll never happen in my lifetime. But that doesn't mean it couldn't happen.
Well, apart from the fact that some old chaps are bouncing back from time to time.
I see little truth in the assumptions coming out from the fact that someone is capable of maintaining a stable rating, since he is falling in the overall standings his rating should be increasing in the first place to be considered stable.
I still wouldn't say zero, though. Maybe the way I'm feeling on a particular day (or just how I feel about the position) will get me to rationalize a ridiculous move. Very probably not. Maybe unlikely enough that it'll never happen in my lifetime. But that doesn't mean it couldn't happen.
Yeah, it's not impossible... but it could be 
I still smell a bit of a no true scotsman though, I don't know. I feel like, some people here consider certain types of bad moves incompatible with certain ratings right from the start. Like any scenario I give where a 2700 does something crazy, it's like they become demoted from 2700 and don't count for the example anymore. In other words, 2700s are by definition "not crazy" or something like that.
So yeah, I guess we do need to define 2700s and 1300s very carefully.
Or maybe just skip the rating thing, and just talk about people in general. A four digit number can't play chess. It's a collection of patterns, evaluations, calculation ability, and experience.
Can this collection of knowledge and ability really produce any move when it's trying to win? I'm not so sure.
There are 40 players in the world rated 2700+. Their play can be predicted as being at the highest level. They will not suddenly play like patzer, under any circumstances
Like wise a 1300 is predictable (for established ratings) They will not nor can they suddenly play like a master for an entire game.
I mean... anyone is capable of playing extremely badly. Like, I could play like an 800 player if I just didn't care at all. And I don't mean trying to blunder, but just, playing a move, and not thinking about it, or checking if it's protected, etc. It's only my discipline that keeps me from doing that. Even in a casual setting, I don't want to play like an 800, lol. I want to check if pieces are hanging. I'm just saying, if I was too lazy to do that, then I might drop a piece, even though I'm a 2000 player.
In other words... it's not like I'm magically playing good moves. I still have to put in some kind of energy to play good moves. I still have a human impulse to make an 800 level move, it's just that I push it back by my desire to not play bad moves.
Sorry if this is a bit unclear. I don't know exactly how to phrase this, haha.
I mean... anyone is capable of playing extremely badly. Like, I could play like an 800 player if I just didn't care at all. And I don't mean trying to blunder, but just, playing a move, and not thinking about it. It's only my discipline that keeps me from doing that. Even in a casual setting, I don't want to play like an 800, lol.
In other words... it's not like I'm magically playing good moves. I still have to put in some kind of energy to play good moves. I still have a human impulse to make an 800 level move, it's just that I push it back by my desire to not play bad moves.
Sorry if this is a bit unclear. I don't know exactly how to phrase this, haha.
I tried some blitz last night and played surprisingly poorly. "I still have a human impulse to make an 800 level move, it's just that I push it back . . ." interesting.
And all of the above because of some stupid insinuations. I don't care that much, his assumptions were ridiculous, but we ought to be dead and crazy as well anyway.
Or maybe just skip the rating thing, and just talk about people in general.
Yeah, good point.
I mean... anyone is capable of playing extremely badly. Like, I could play like an 800 player if I just didn't care at all. And I don't mean trying to blunder, but just, playing a move, and not thinking about it. It's only my discipline that keeps me from doing that. Even in a casual setting, I don't want to play like an 800, lol.
In other words... it's not like I'm magically playing good moves. I still have to put in some kind of energy to play good moves. I still have a human impulse to make an 800 level move, it's just that I push it back by my desire to not play bad moves.
Sorry if this is a bit unclear. I don't know exactly how to phrase this, haha.
I tried some blitz last night and played surprisingly poorly. "I still have a human impulse to make an 800 level move, it's just that I push it back . . ." interesting.
I dunno. Maybe it's kind of like when you're watching a game between two GMs, and you think about moves they could play, oh, Qg4 looks interesting oh wait nevermind it's covered. But of course you would never (.000001% chance) hang a queen like that in a real game because you would filter that out before you play it. But you had that very momentary impulse to play that move before you saw it was hanging. I mean, I know I do this. Again it's not a big deal as long as when I'm actually playing a move, I quickly find out it hangs the queen.
Even the current 2700 players aren't that way because their mind literally restricts certain bad moves from happening. They're just very unlikely to play certain kinds of bad moves. But they still can, and have.
All we know from watching 2700s play is that they are likely to do certain kinds of things and unlikely to do others. That doesn't mean they can't do these things.
Humans will be humans regardless of what rating system we create. You can't physically get rid of having a blind spot. All you can do is measure how rarely it happens :)
Ah, but it's not random, it just appears random. The same way flipping a coin isn't random. The same way oversights aren't random. It appears all moves are possible, but they aren't. I'll argue that stable ratings are evidence of this. (At least when we're assuming both players are serious and trying to win or draw.)
Going by your coin example, what I'm saying might be something like, technically, every coin flip ever done could have been heads. But in practice it's 50/50. But it didn't have to be that way -- there isn't this magic force that keeps that from happening. It just turned out that way.
And of course when I say random, I mean, based on the information we have. I'm saying that the knowledge we have about what 2700s are like, still nevertheless don't rule out them getting outplayed by a 1300. It just means, you'd be really stupid to ever bet on it.