EPTAt every level there are lapses in our logic; gaps in our trains of thinking. Pro players fail to identify out-of-the-box and non-standard play because they're so used to 'efficient' play. Casters judge player's actions because of their own lapse in realization that what they see isn't what the player has seen. And as spectator's we try to mimic our pro's without realizing why it works, or how it works, or what it's trying to accomplish, without the necessary information or backing to do it safely to begin with (at all levels). Simply, it's a failure in perspective, and it isn't unique to Starcraft in any way shape or form. However, I want to use it to analyze a specific concept: all-in.
First and foremost I guess the word must be defined. I'll use my own thinking. To be all-in is a situation in which you must, at least, have an end result of your opponent being at your level after you execute it, with the risk of being critically behind if failed. Less technical: An all-in is a place you put yourself in that requires you to put your opponent at your level when you do it, or you'll probably die.
There is a lot of difficulty in understanding the all-in because of it's misuse and misunderstanding. To do an all-in, for example, requires you to be all-in. But you can be all-in without doing an all in, and that's the critical piece a lot of confusion comes from. You will hear a lot of caster's say things like, "Oh he hasn't expanded yet he's obviously doing a 2-base all-in". At the pro level that's probably true because standard 2-base play has sort of died to the longer macro game, and at the top level if someone is doing two-base play, the most efficient effective way of doing it is probably to do a very specific all-in, that MIGHT have some hope of transition out of it.
Which brings up two more issues: 1. The lapse, and 2. If it can be transitioned out of, how is it an all-in?
So we'll address 1 first. The lapse involves a failure to bridge to your own level. I hear, "Oh he isn't taken his third by 15 minutes he's obviously doing an all-in" and I process that as, "Okay if I'm against an opponent and he hasn't expanded by 15 minutes he's doing something wonky." In some circumstances that's definitely true. In others however, your opponent is just committing himself to a strong two base game with a lot of focus on army and is probably going to try to kill you frequently, or mass a two base army and kill you. Then you have to take that leap and say, "well if he's committing to that, he IS all-in", which is true, but it doesn't mean he's doing an all-in, and that's really important to recognize.
2. If we look back at the definition of an all-in, it doesn't actually require the all-inner to win. If you do say...a 1-1-1 all-in against a protoss. You don't have to kill the protoss with your attack. It'd be awesome to. It'd be great to at least totally cripple the shit out of them. But what is actually required is that you either force them into making units while being cost efficient enough with your own that you can tech/expand/whatever really fast and catch up to them, OR you have to bring them down to your level in army/tech/economy. It's part of what makes an all-in really complicated to judge the success of; generally we judge an all-in based on whether or not we clearly win the game with it.
A two-rax all-in is what I'll use as an example (despite the actual definition of "what is a two-rax" being much more difficult, so for clarity, not a two rax pressure with a few marines, but a two-rax, 3 maurader and some marines, thing). If we start off with a two rax...we have no real tech started and are on one base. Let's say it's against another terran. Chances are they're making units anyway unless they're going for some fast mech build or a fast tech (banshee something) thing. So we take in our units and we kill a bunch of scvs and snipe his natural, giving us time to get our expansion half-up. Now, we've killed a couple scvs, and it was early enough that he probably didn't have a very big scv lead anyway, and our natural is almost done and he has to restart his. We are ahead in economy. We may or may not be ahead in gas. Depending on whether or not he has his factory at home and what he was doing he might still be ahead in tech. Our all-in then, probably paid off and put us at least even again. Let's say instead he kept his orbital and had to lift it, but we ran up his ramp, made him pull scvs, and we sniped his factory. We're probably going to be about even in economy, with him probably getting an extra MULE off but having 1-3 less scvs. But! We're definitely even on tech. And you have the advantage of knowing what he was hoping to go for (and possibly the advantage of having forced him into going for something else now that he wasn't planning to do). Our all-in was successful enough that we could transition from it into what we wanted without being behind or effectively dead. And that's not limited to super-early game situations; the same scenario could be created with a blink-all-in, but I'm not going to do it for the sake of over-explaining.
So in short, an all-in puts you all-in, but being all-in does not necessary mean you're doing an all-in, or that what you're seeing is someone doing an all-in.
All just, food for thought, and something to remember in your own games and scouting and thinking, and to notice and note when you're watching high level games and casts.
End.
