EPTSo my last entry was http://www.teamliquid.net/blogs/viewblog.php?id=147995 about the seeds of an idea germinating into an opener and finally a strategy. That article talks about the income and production differential between zerg and other races, and what is necessary to overcome it.
So I tried that strategy for quite a while, and for a while it was very good. I won a lot of matches in my range. Of course that moved me up to a range where I started losing. A lot.
I am not one to just abandon my strategies. So I started looking at what was going wrong. In case my previous post is too long, the build I was using there was a 12 pool, 15 overlord, 16 gas, 19 hatch in base into lots of lings. The premise behind it was to have enough military to counter early aggression, but still manage to keep up with economy of my opponent. I would expand after securing an advantage over my opponent, either by winning an early push, or noticing him commit to defense. If he early expanded himself, I would attack the expansion and then expand myself.
It worked reasonably well against fast expand plays, or early aggression plays, but it had trouble with large later pushes. For instance, it was great against a 2-gate into 4 warpgate, but sucked if the 2-gate push never manifested, and it was just a stronger 4-warpgate push later. Or a large marauder push. Or preignitor hellions. Or various other delayed 1-base pushes.
So I started to look at what was wrong on the surface. The first thing I noticed about the build is I was committing resources to military early. But I was keeping that zerg force at home and just waiting and hoping to be attacked. Especially since I don't want to go start hitting the wall and lose my zerglings to no gain.
The second thing I noticed is while I could keep up in harvester count, and I could keep up in production, which were my main focuses in the first article, I always fell behind in army strength. I thought this was actually a bit odd, because I had the same number of workers as my opponent, so I should have the same income, and I had similar production capacity, so I should have the same army. Looking more deeply, I did have the same income, and I was spending it all, but my army value was lower because of early tech requirements.
I'll start with an analysis of the second point:
If a zerg player wants to get roaches and hydras against a 4-warpgate push and remain competitive in production rate they need to have 2 hatcheries, 2 queens, a spawning pool, a roach warren, a lair, a hydralisk den, and they will probably be getting zergling speed. This means the zerg player has used 4 drones for buildings, 1/2 of an overlord's supply for queens, plus the building costs. This is 200 + 50 + 200 + 150 + 150/100 + 100/100 + 350 + 100/100. That's 1300 minerals and 300 gas committed to tech and production.
For a protoss player to get the same infrastructure, they need to build 4 warpgates, a cybernetics core, and research warp gates. This means they spend 800/50 for a similar level of technology.
(I use protoss a lot for my examples, it's not because I have more trouble with protoss, they are just easier to predict, a terran opening with reapers is very different than one opening with marines and different than one opening with hellions. There's a lot less variation with protoss)
So I'm learning that I can keep up with economy, and keep up with production, and even if I manage to do that perfectly, I'm going to be like 500 minerals and 250 gas behind on army compared to a 4-warpgate push due to that initial investment.
So then I started to consider the first part a bit more strongly. My issue with my original strategy was that I'm not putting those early zerglings to use unless my opponent attacks me. They are great to have because they keep me from dying to a 2-gate zealot push, they are awesome because they're way more mobile than a spine crawler. But they kind of suck if they're just mulling about in my base.
A worker mines .75 minerals every second or so. That means 12 zerglings hanging around in my base for 2 minutes actually ends up costing me about 540 minerals in lost mining time, as I could have made 6 drones instead. That's another expansion and two spine crawlers. I obviously don't want to die to an early push, and I never found turtling up on spine crawlers to be terribly effective the way I play, so I need to make those zerglings. But I don't want to build a bunch of zerglings sitting munching creep for 10 minutes while I wait for my opponent to build up a force to take me out.
So I started to consider my early-game options to make use of those zerglings. The simplest way to make good use of them, and something I was already doing, was simply for vision. Put one at the Xel'Naga towers, station them outside his base, deny scouting, deny expansions. Still, they aren't a lot of use. I'm spending minerals on them, and my opponent hasn't felt it.
One option is banelings. This is very effective against a terran player, because he's either fast teching, in which case he's got a smaller force of bio to deal with lings, or he's going bio, which is going to get melted by banelings. But it's also highly effective against protoss. A protoss player will often wall off with 1 or 2 zealots blocking his choke, maybe partially walled off with a pylon, to get fast gateways or tech up. A smallish investment into banelings will break that defense and cause a reasonable amount of havoc with all of those zerglings that you'd otherwise have sitting around waiting for your opponent to reach critical mass.
I'm going to go into a divergent topic now, but it is related. In my previous post I talked about production rates. I said that a hatchery produces about 0.1667 larva per second, so you can make about 0.1481 drones or zerglings per second as you need to get an overlord for every unit you build. Something kind of unique to zerg is the fact that they actually reinforce a lost army more quickly than they can build an army in the first place. If you're reaching supply cap, yeah, you build a unit every 6.75 seconds. But if you're simply reinforcing a dead army, you can build a unit every 6 seconds. This means that as a zerg army dies, your production rate actually goes up. This is different from terran and protoss because they can keep themselves from being supply capped without sacrificing barracks time or gateway build time. It's actually significant as it's 12.5% more production if you're producing 1 supply units, 25% more if you're producing 2 supply units, and 75% more if you're producing ultralisks. (Producing ultralisks when you're near supply cap means you're making 3 overlords for every 4 ultras, so you spend 7 larva just for 3 ultras, but when you're far from capped, you can make 7 ultras with 7 larva)
So, I increase my production capacity when I lose units. My enemy doesn't. I'm in a more difficult position when my opponent has a critical mass of units. I can't easily tell what's going on behind a wall. Due to these 3 points, it's incredibly effective for me to try and deal some early damage. Banelings aren't always the answer, and the times they are not are reasonably obvious. But if you can deal some damage inefficiently to your opponent this way, you actually might be in a better position. That sounds odd, actually counterintuitive, but it's pretty true.
For instance, say your opponent is well prepared for you. You blow up a pylon with 5 banelings, dealing some damage to a stalker and a gateway, the zealots that he has are kept away until the banelings explode, you end up finishing off the stalker, kill a zealot, decide to retreat, he gets a forcefield on your ramp as you retreat, you get 4 lings trapped and save about 4 lings and retreat to your base.
It seems like a failed bust. But assume you had 16 lings, and morphed 5 of them into banelings and you save 4 lings. You've lost 250/125 minerals worth of banelings, and 175 minerals worth of lings. So 425/125 all in all. You've killed a pylon, a stalker, a zealot, and possibly supply capped him for a moment. You've scouted his base and have an idea of what he's going to be doing. You have killed 125/50 worth of stalker, 100 minerals of zealot, and 100 minerals of pylon. So you dealt 325/50 worth of damage. But aside from that, you've increased your production by 25% by freeing that supply. You can gauge by that attack whether you're at risk of a counter attack. If you are at risk of a counterattack, you have that increased production capacity to reinforce your army. If you're not at risk of a counterattack, you can drone up. In the case of a counterattack, your opponent is losing the advantage of his base and his ramp.
Awesome, so even a baneling bust that looks like it's failed is doing a lot of good. It's giving those lings something to do. It's giving you the opportunity to free up that expensive supply, at a larger cost to your opponent, and it's giving you some of that scouting that is so hard to earn early on. Obviously things like a sentry forcefielding the ramp is going to cause this big problems, but then again, juking some forcefields out of the protoss is spending that energy so it can't be used during a push, and accounting for some gas so giving you reasonable information.
Ok, so I've dealt with my early-game army dilemma. What about the fact that he just has more stuff than me? I mean, assume he does just go for a 4-warpgate push and he uses the sentries to keep me from doing any baneling funny stuff. The fact that I'm spending some sentry energy doesn't mean much when he ends up with a much more expensive army
There I had to look at some of my original assumptions. In my first post I was trying to determine how to keep up with the opponent. This would maybe be useful if keeping up with my opponent would keep me competitive. Unfortunately, due to frontloaded tech costs, this isn't the case. So the next question is, how do I deal with the fact that keeping up is not really viable?
Well, the first thing I do is take a look at what everyone else is doing. Typical zerg play is fast expand. People like to do that because it works, though I expect many people who can execute it well don't necessarily know why it works. I'm all about the why, and challenging those assumptions. On the surface fast expand makes perfect sense, You're basically doing what I was doing before, except instead of hatching in your main, you hatch at your nat. Excellent. My issue with FE was always that I found it more difficult to defend against early pressure while keeping a strong economy.
Armed with the knowledge in this post, I started looking for the solution to this problem. My issue was early aggression, something like 2-gate zealot, or proxy 2-gate would give my FE trouble. Hatching in my main meant I had a queen to help out while zerglings finished, I didn't have a ramp to deal with, I had creep around, and I could basically do more with less. The fast expand I felt took a while to become saturated, and that risk wasn't paid off for a while. But that risk does get paid off, and it pays off well.
So since I know if I'm on one base, I'll be behind in economy because of these early tech costs, I pretty much need to be on 2 bases against anything but the most all-in early plays. Ok, so this FE thing is starting to make sense. So then the question becomes how do I FE and not die to this early aggression?
The answer is what makes zerg interesting. I can't keep up a protoss or terran with army and economy in the early game. But I can overwhelm them in either. Losing units makes my production capability go up. Preventing my opponent from achieving a critical mass of units or surprising me with tech is imperative. This made the solution reasonably obvious.
I can make a lot of zerglings. I can make enough zerglings that it's unreasonable for my opponent to have a similar sized force. This means that any early open air encounter I will win. I do cut drones to do it, but my alternatives are to match my opponent in drone production, which I've already identified as a bad idea, or drone up hard, which puts me in trouble against early aggression. (The alternative is spine crawlers for defense, but if those are avoided there's not much to fall back on, and I just like a more active style.)
If the zerglings can't do anything in the open, they can be banelings and break into the base. Before that happens though, I'm aware that there is no early push coming. I can now drone up hard. If I decide to attempt the bust, I will lose a lot of units worth of supply. The bust will give me enough knowledge to know how to follow up. The loss of units will increase my production rate temporarily.
Awesome, so I get this zergling army early which means I'm in a suppressed economic situation, but protected against early aggression. I use the zergling mobility to confirm that there's no early aggression and then can drone up equalizing my economy. I can then use those zerglings to scout, attack, or defend, and if I lose them, I get a temporary boost to production, which I can use to either rebuild an army, or pull myself into an advantageous economic situation.
Great, so I've FE'd, I've protected my FE from early aggression by frontloading army. I've recovered economically once my army is out, I've applied pressure to my opponent and exerted map control. I've quite possibly got a lot of scouting information, and possibly thinned his defense. I could possibly have forced him to reconsider a fast tech build as well. If he's tried to do something like a 2 gate zealot push, or a fast expand, I'm going to probably be ahead. If he's done something safe and economic like a forge FE, I can recover from that ling investment quickly and safely by powering drones.
All the information in my previous article is still relevant and I still consider it heavily when I play. I know that a 12 pool is a great time, but I want that ling speed quickly. 13 gas, 12 pool gives an almost identical economic situation to the 12 pool/later gas, except the pool completes a bit later. Any 13+x gas, 12+y pool setup works as well with a slight excess in minerals and later pool time. Pulling drones off of gas once you get 100 or 150 really helps mineral acquisition. This lets you get that expansion up that much faster, which means you can drone that much sooner, and you're not spending much gas at this juncture anyways.
So all in all, I realize I'm going http://wiki.teamliquid.net/starcraft2/Speedling_Expand and that's cool. I don't mind using the same strategy, especially one that Sheth advocates. The cool thing is that I came to this conclusion with my own stats, numbers and understanding. I know exactly why I'm doing this speedling expand, but not only that, I understand what impact subtle variations on it will have. For instance, in some situations, a 14/14 or 14/13 gas/pool might be better, or a 13/12 gas/pool, or the 13/13 advocated in the link. Each has subtle differences and impacts on your ling speed and hatchery timing. 13/12 will get lings out slightly sooner, 13/13 will get speed when your pool finishes, 14/14 will give you a slight mineral boost. 14/13 will give you a mineral boost, lings out a bit faster, but speed won't start for a few seconds after your pool completes. On a larger map, 14/14, or 14/13 might be a good idea. If you're worried about very early aggression, 13/12 might work better.
Plus I can add my own touches to it. I know if I go 13/12 I can get baneling nest right away instead of ling speed, this could possibly be useful in certain circumstances. If I go 13/14 I can get ling speed and baneling nest when the pool completes and pull 2 drones off gas, and have enough gas trickling in to make a reasonable number of banelings. If I go 13/13 and pull 2 off gas, I can get roaches to supplement. If I go 14/14 and pull 2 off gas, I can get roaches more economically when my pool completes.
But because of the previous research I also know about options I have if I somehow get forced into 1-base play, and I can play off 1-base effectively without a serious hit to economy assuming I can get an expansion out before my opponent. I never aim to play a one-base game because I need to be able to recover from this early deficit that zerg tech puts you in. But if scouting I notice a weakness that can best be exploited by a 1-base strat, then I'm well equipped to do it. For instance, if a protoss player tries to sneak an fast expansion without adequate defense, or a terran neglects his wall-off, I can one base with an amount of confidence because I know I'll be able to inflict enough damage to compensate for the later expansion, or outright win the game without having to risk my nat to a counterattack.
The purpose of this post is to explain how my original unorthodox opening worked in theory, and has a lot of good points to it. In practice, certain openings exploited weaknesses of the build that I didn't consider when I was first analyzing it. This is just fine, because we can work on those weaknesses to find solutions to them. By working through those weaknesses I was able to come up with a solid opener that has served me very well lately.
As it is, the opener that I came up with is identical, or at least incredibly similar to a well-known opener, and that's great, because it means that this process has yielded results that have been tested and proved effective. But as well as that, it means I understand the implications of many variations on that opener. It also means that if I start to notice weaknesses in this opening, I can use the same process to evolve it further into something that is even more potent, rather than waiting for another player to do that for me, and make it popular. While this process takes a lot more work and thought than simply looking up a build on the wiki, I think in the long run I am growing to be a better player because of it. And while I might have won more games up until this point by simply doing what is known to work, I will probably win more games now because I know how to deal with all sorts of edge cases relating to this build. Or I can use even that proto-build in situations where it's weaknesses aren't going to manifest.
