Well your last point is the easiest to show as false! Vapor Snag is strictly better than Void Snare and shows that instants are superior to sorceries, both because it can be fetched with Cunning Wish, which is not only amazing in itself but can power a Snapback or Disrupting Shoal countering Unlicensed Disintegration while Burning Wish can only alternately power a Pyrokinesis (which is an amazing card but inherently less flexible), it can also win the game against an opponent at 1 life who has an Angel's Grace in hand ready to cast!
Well, Void Snare can be used to kill a 0-life opponent who has Phyrexian Unlife on the battlefield. Vapor Snag cannot do this. Ergo, another win for sorceries.
You are right about the first set - on reflection, Legacy has a lot of ways in which turn based loses can be thrown out the window, which leaves the second set as a more vague outline even that my original formulation.
While that does seem to be true, the first set is still more informative. Sure, if I'm being pedantic (which, knowing me, seem likely) then I can state that the "running out of cards" route to ending the game can be thwarted if I control Yawgmoth's Bargain. Once I'm out of cards, I simply neglect to every pay life in order to draw more cards. I skip my draw step, so I don't lose. But even that isn't a complete answer: you could target me with Ancestral Recall, forcing me to draw cards. Ah, but if I have Pursuit of Knowledge, I can replace each of those draws with putting study counters on it, so no matter how many effects you generate that would make me draw cards, I'll just keep replacing them with meaningless study counters. Well, if you destroy my Pursuit of Knowledge, then you can get me to lose the game by drawing cards. Except that doesn't work out so well for you if I have Laboratory Maniac, because then instead of losing the game, I win the game. Fine then, destroy everything on my side of the board. I have no cards that skip my draw, I have no cards that replace my draws, I have no cards that prevent me from losing the game. Well, then I guess I would lose—if it weren't for your Abyssal Persecutor!
With all the exceptions and nuances, we could just stipulate: "If, as per Rule 104.3c, you are required to draw more cards than are remaining in your library, then you draw the remaining cards and, the next time a player would receive priority, you lose the game." Technically correct and to the point, but a little weird.
This "ending the game" business and your mention of Legacy reminds me of the "Four Horsemen" deck and the ensuing rules issue. The issue may be over and done with, but that was a matter of
tournament protocol and if we're speaking about Magic theory outside the scope of a tournament setting, then that tournament-specific aspect doesn't necessarily apply. This is another potential "extraordinary circumstance" that could be excepted from your second set of game outcomes. My qualm with that is that I don't think Four Horsemen, as an exception, is realistic: playing the loop out might
theoretically take more time than the players have, but that feels like a lame out because
realistically it won't take more than a few minutes (extra long shuffling sessions aside). But Four Horsemen isn't the only possible loop that runs into this issue: it's just the only one that saw tournament play.
Not knowing who all is reading this thread and what they do or don't know/remember about Legacy decks, I should go over the combo in the Four Horsemen deck. I won't talk about the whole decklist, which evolved over time and had various advantages and disadvantages. Strictly looking at the combo kill...
The deck aimed to get Basalt Monolith and Mesmeric Orb onto the battlefield so that it could mill its own library one card at a time, at will (tap Monolith for mana, use the mana to pay for Monolith's ability to untap itself, mill one card due to Mesmeric Orb). It would then, going one card at a time, attempt to get at least 3 copies of Narcomoeba onto the battlefield in preparation for a Dread Return. Next, it would continue using the Basalt Monolith + Mesmeric Orb engine to mill cards and get Dread Return, Sharuum the Hegemon, and Blasting Station all into the graveyard
while the singleton copy of Emrakul, the Aeons Torn was still tucked away in the library. If Emrakul showed up before all three were in the graveyard, then the player would have to shuffle up graveyard and library and try again, but you had infinite self-milling available so that wasn't a problem. Eventually all three cards would hit the graveyard (in whatever order) before Emrakul showed up, at which point Dread Return put Sharuum on to the battlefield and Sharuum put Blasting Station onto the battlefield. The final step was to self-mill repeatedly to put Narcomoebas down, untapping Blasting Station, then sacrificing the Narcomoebas to Blasting Station for damage to the opponent, and then Emrakul shuffles everything up to reload. A point in the deck's favor was the it could easily use Narcomoeba with Cabal Therapy to punch away at combo-hate cards. But what the deck was famous for was the amount of setup to get that final infinite combo. This cause a bit of a controversy because some judges allowed the deck to go off in tournaments and others cited the players for slow play (I think "slow play" was the official term but I cannot remember) on the basis that they were not really doing anything. The idea was that this was different from a traditional infinite loop because in the traditional loop, all of the variables are known. I sacrifice Ornithopter to Fallen Angel and Enduring Renewal brings it back. I sacrifice Ornithopter to Fallen Angel and Enduring Renewal brings it back. Over and over. Same thing every time. So I specify that I'll do it X times and that's an acceptable shortcut. But with Four Horsemen, I don't know how many iterations it will take. I might hit all three of the needed cards on the first try. Or I might hit Emrakul before hitting the last of them 10 trillion times in a row. Ridiculously implausible,
but theoretically possible. People did probability analysis on this, they used real trials and showed that the time to complete the loop was averaging on the order of a few minutes. But the sticking point was that uncertainty. Sure, if your library randomizes so that you get all three of those cards before you get the other one, then you win, but you cannot say when that will happen, so you can't shortcut it. And if you can't shortcut it, then executing the loop deliberately is apparently, under tournament rules, not allowed. Eventually, some mandate from on high ruled that shortcutting wasn't permissible for this and that the loop was slow play (or whatever), killing the deck. I'm not much of a tournament player and do not feel qualified to speak to the rightness or wrongness of the ruling. But rulings aside, this presents an interesting case, conceptually, of a situation which, until resolved, has neither playing winning or losing. The game is "stuck." In practice, this loop is unlikely to remain stuck for very long. Like I said, it really should only a take a few minutes.
But that's tournament Magic! I'd bet, somewhere out there, it's possible to get a configuration of cards that takes a lot longer, a loop that could theoretically win, but one that's too cumbersome to pull off. In fact, I encountered something like this by accident, which I'll relate at some point.
As to distillation, I was thinking primarily of the premises underpinning your ABC formulation of thought. What I have for now after your initial feedback is something like, "Each deck is a combination of cards that achieves an outcome from the following options: set a player state to win or lose, or set the game state to draw. Analyzing the performance of decks in relation to one another is a matter of determining the probabilities regarding which deck determines an outcome when the decks interact." Still a kernel with some utility for starting a larger theory from, though as you say, extremely broad and vague in its inclusiveness.
Also, though I do not wish to incur your wrath, Tempest Efreet now technically lives outside the main rules for the game in an "optional variation."
As far as I'm aware, it does not and there is no such thing as "lives outside the main rules for the game." It is a black/white bordered Magic card from real Magic sets and is perfectly legal. It's banned from tournament play, of course, but I was talking about outside of tournament play. It also instructs you to take it out of your deck before playing if not playing for ante, but in my silly scenario,
the card was never in my deck. I fetched it from outside the game with Ring of Ma'Ruf. In tournament play, Ring of Ma'Ruf can only fetch cards from one's sideboard, but same deal, blah blah blah. Casual play has no built-in protection against the use of Ring (or a Wish card) to bring Tempest Efreet into a non-ante game. Opponents would then have to pay 10 life, concede, or make the exchange. I have even made
token copies of Tempest Efreet and attempted to trade those for my opponent's cards (opponent didn't go for it).