In Which I learn to Macro

To begin with, I want to clarify that--at least as I read it--most of the "how to get better if you're a lower-league player" advice on TL is emphasizing working on macro mechanics. The "mechanics" part is important to distinguish this from the idea of playing a "macro game", as it's called. The "macro game" is the sort of stable way SC should go, and it is incredibly complicated.

Working on mechanics makes sense: you can't possibly learn useful things about the strategy of SC if you're missing injects, not building enough drones, getting supply capped, forgetting to take gas, etc. You miss injects and you start thinking, say, speedling-only strategies don't work early on against such and such a thing Terran can do because I can't get enough speedlings because I don't have enough larva. This is one of the primary reasons you should work on mechanics first: you can't make useful conclusions about strategy until your mechanics are reasonably solid. (The other reason is that it is an easy way to make fast gains in skill.)

Working on mechanics means getting to the point where you're doing everything you need to do when you need to do it. This could mean a whole bunch of things depending on how complicated your strategy is. At its most basic, it just includes building drones on time (and enough of them), hitting overlords, hitting injects.

I don't think attempting to play a "macro game" makes any sense for a bronze-level player. The issue is the following: You want to work on mechanics, but the "macro style" is tremendously strategically difficult, requires incredibly good scouting, and on top of that is the most mechanically taxing style I can think of. In short, it's just too complicated for someone trying to work on mechanics. You should be trying to simplify the game strategically so that your mechanics have more sway over the outcome.

So the thing I am recommending against is your idea (2): not attacking to win until you have three bases up and running. With Zerg, I would instead do something like the following: build a bunch of drones for a while. Focus on not missing injects, not missing overlords, etc, and get some sort of basic tech up. (It can be whatever suits your fancy; I like roaches as a core unit because they're hardy and so you can 1a them into a lot more stuff as you said above.) Then, at a certain time in the game (and you just sort of pick some random time here) switch into only making attacking units. Then, after you have some number of attacking units, go try and kill your opponent. Don't stop until your opponent dies or you run out of money.

Note that this strategy, simple as it is, is already pretty hard to do effectively. On the one hand, you need to make sure you have perfect mechanics. You shouldn't be missing injects, or ovies, or drones, or putting guys on gas, or whatever. On the other hand, you need to make two really difficult decisions: when to switch into units, and when to attack. Then, after you play a game, you have basically two things to look at: (1) The mechanics and (2) the timing of the switch into units.

This strategy also has the advantage that you get a better idea, while playing, of what your opponent is doing--so you get a better idea of what the races you're playing with are capable of at early times in the game. For instance, maybe you find out that when your Protoss opponent expands at such and such a time and also sends a void ray into your base, you can just go kill him with a bunch of roaches. This is really nice to know! To generalize, being aggressive early deals effectively with a lot of unsafe strategies--both super-aggressive army play from your opponent and also economic- or tech-aggressive play.

But from my perspective the real reason to start with this kind of simple strategy is that after a while, you will start to see it evolve naturally into a more "macro game" style of strategy. For instance, at a certain point you'll decide "gee, if my opponent is making a lot of units early, I can't attack effectively, so it's better if I make more drones and be defensive in such and such a way". You'll want to be scouting, because you have a simple strategy that you know how to tweak effectively based on what you see. You'll start adding things in little by little--"gosh, my mass roaches don't deal very well with other Zergs when they get infestors; I guess I'll try getting infestors too".

The problem with committing yourself to playing defensively for a late-game is that all of a sudden you're trying to deal with way too many problems of strategy. You have to scout cheese so you can defend it. You have to scout when your opponent is trying to expand or tech too fast so you can punish it (otherwise you'll die in the late game). You have to scout hidden expansions. As the armies get closer to maxed, you have to worry a lot more about composition. You have to worry about tier 3 tech and upgrades and where are your opponents' expansions and where is his army and what's a good way to engage and all of this. All of this together is hard enough to understand that this game is deep enough to allow for professional play, and I don't see any way you're going to be able to focus effectively on improving your mechanics while trying to deal with all this other stuff.

On a personal note: The above just conforms to the advice that wound up working for me. I played Brood War with the "macro game" mentality and I was terrible at it. I copied what I "saw" professionals doing and was frustrated when it didn't work. Often my problem was that my mechanics were bad. But my most serious problem was that because I didn't have a grasp of the simple ideas of how my pro macro super three-base strategies should work, I eventually hit a wall and wasn't able to improve because I had no ability to adapt to what I saw. I thought about how to not die for the first 10 minutes of the game and then I just sort of played somewhat aimlessly. ("Pressure" my opponent. Whatever that meant.)

For some reason I took SC2 with a different attitude, perhaps because it was new and I felt like I could play around with it. I play Zerg, and I play a lot of aggressive strategies. But I also find a lot of my games developing naturally to the late game. I played lots of games where I just built a ton of roaches and threw them into my opponent's army and either won or lost. Eventually that stopped working. So I threw in some corruptors to deal with colossi, say. Now I feel a lot more comfortable going into the late game because I'm doing so with a plan that I've developed over a long time of playing that came about as a natural extension of what started as build a lot of units and go try and win. Frankly, that also has made the game a lot more fun.

TL;DR If you want to work on mechanics, you ought to start out with a strategy that is so simple that you can actually understand how it works, and build up from there. When you have a good grasp of your strategy, you know everything you're supposed to be doing, and then learning the mechanics is a simple matter of checking to see that you're doing it. In addition, this has the advantage that your strategies will develop to be more complicated, and start to move toward more "macro-oriented" strategies, not artificially, but in a way you understand, as you start to see the necessity for complicating them.