EPTMacromanagement in Starcraft II
![[image loading]](http://www.teamliquid.net/staff/SilverskY/MuleQueenObelisk.jpg)
Banner by SilverskY
Warning
This article (not by choice) contains incorrect information. As the reader, it is your job to read this in its entirety and comment, to yourself or publicly, on whether you agree or disagree with its conclusions.
It is important to remember that even if an idea is based on incorrect or incomplete information, the conclusion can still be accurate. For example, if the queen's build time is actually 50 seconds, instead of 45, it probably isn't going to change the disparity shown by this article.
Onward and Outward
Starcraft II is different than Starcraft. That's a good thing. Despite knowing this, many people at Blizzcon entered the game playing it like the original on day 1. Units from Starcraft II were substituted for their equivalent Starcraft units, and normal builds were tried. In my experience, the first person to break that mold was Hot_Bid, as shown in his Zerg Queen article. After I saw this, I quickly changed how I played, focusing ever more on macromanagement in its most raw form. This article looks to explain the operation of macromanagement in Starcraft II, as well as briefly touch on why it has changed the basic form of the game.
Less is More?
As you probably know by now, Starcraft II workers return 5 minerals per trip instead of 8. They cost the same as the original. They move the same as the original. They mine slightly more efficiently, but not enough to make up the disparity of 3 minerals per trip. Yet everyone who got a decent look at Starcraft II, and has a reasonable understanding of Starcraft, would agree that you could make units more quickly in Starcraft II. Why? Enter the new macro mechanics.
Get Money
Below are the basic macro mechanics for each race.
Zerg
No need to touch this. Read this. I agree with almost everything in that article.
Protoss
For 200 minerals, after having a gateway, Protoss can create the obelisk. The obelisk regains energy automatically, energy which can be used to recharge shields, give mana, or speed up mining. You will only use the latter - It casts an area of effect spell (about exactly the size of a standard arc of 8-10 mineral patches) which lets the targeted Probes mine 6 minerals instead of 5 for a short time. In that time you will build up enough energy to immediately cast the spell again.
Terran
For 150 minerals, after having a barracks, Terran can upgrade their command center into an orbital relay. The upgrade grants scan, upgrade supply depots, and MULE. You will again only use the latter - For 50 energy, it drops a "super SCV" that mines for around 60 seconds. It mines like a normal SCV but returns 30 minerals (I believe) per trip instead of 5. You can get 50 energy before the time is up, meaning it's possible to have 2 MULEs at a base for a brief time. Further, if you forget to use the ability, you can drop 4 MULEs at once. Obviously it is better to use them as you can, but you aren't losing too much potential until you hit 200 energy, unlike the other abilities. I also believe that the MULEs can mine patches that are currently being mined by SCVs.
Comparison
Undoubtedly, the Zerg mechanic is the best. It grants more overall production of anything for the entire game. More drones in the early game means more hatcheries in the midgame, yielding more queens and a dangerous compounding advantage. More larvae in the early game means more hydralisks threatening an attack, meaning more static defense needed for your opponent to live. Further, in a straight-up macro war, you will never out macro a Zerg player who opens 13 pool into queen. It's not possible. I will try to prove this later in this article.
The Protoss and Terran advantages do differ slightly. Because the MULE is basically giving you a set amount of minerals over the next 60 seconds, it is extremely good at the start of the game, when worker saturation is low. Once you have achieved full saturation, the effect is still the same, but the improvement won't affect you as much. Conversely, the Protoss ability is most powerful when you have a nice saturation of probes, since they are all affected by the AoE.
There Were... Other Abilities?
There has been a lot (A LOT) of talk on the forum about the lack of choice between the macro mechanics, and the devastating effects of Terran losing scan. These are unfounded. After playing Starcraft II for 10 games and understanding the mechanics, you will see there is actually no choice. It's an illusion of choice. Given the "choice" between using the MULE and scanning, you will always choose the MULE unless forced otherwise by immediately cloaked units. The return is just too great. Even if I were supply capped, I would bank the MULE minerals while building a supply depot before I used the ability to gain extra supply. You've got to figure the MULE can make a round-trip every 6-8 seconds. And it returns 25-30 minerals every trip (I believe 30 but I'm erring on the safe side). Assuming it stays for 45 seconds (I believe it's actually 60). This yields somewhere between 187 and 300 "extra" minerals per minute. Clearly something that can't be skipped for a convenient scouting scan. This also shows the importance of getting the MULE early, which is a theme common between all the mechanics. The MULE pays for itself in the first minute, after which you are generating around 200 minerals extra every minute. You can see how this compounds.
You may, rarely, use the obelisk's ability to charge mana in a nearby high templar; but at 2 energy for 1 mana, and with your obelisk rarely having more than 25 energy, the options look grim. So you can recharge 12.5 (let's hope they round up to 13!) mana, or boost your probes' mining in case you live the attack. The choice is obvious.
The Queen can... You will never use the Queen for anything other than pumping out a ridiculous amount of larvae.
Take the Red Pill
Believe this: There is no choice in the current state of macro mechanics. You get them as quickly as possible and you use them as often as possible. In the end, RTS strategies usually boil down to the simplest common factor - maximize resource production as quickly as possible. Starcraft II's macro mechanics do not open up more options, they seal your strategic fate in the early game.
Racism
Ideally, all the macro mechanics should provide similar advantages. It would still be possible for them to operate differently and remain balanced, although it would be unbelievably hard to accomplish. For example, Protoss could give a boost to mining and Zerg a boost to production, and if the costs were well-balanced, it is conceivable that the matchup could be balanced. It would just be very, very hard to achieve this balance.
Hot_Bid's Zerg Queen article has talked about the power of the queen reasonably in-depth. The conclusion is that the queen is a very powerful unit; however, this article looks to compare the queen to the other race's macro mecahnics to find if one is objectively better. Let's begin with some theoretical discussions applicable to Starcraft.
I am playing PvT. Gambling that Terran didn't wall in, I build 2 gateways and begin producing zealots. By the time I scout him, I have 3 zealots either in production or produced, and I see he has a tight wall. I've already commited to attacking, but my attack will be ineffective. Terran has given up map control to me, but has a far superior economy for it.
I am playing ZvP. I place a standard 12 hatchery, 11 spawning pool, 13 hatchery, only to find that Protoss is rushing with zealots from 2 gateways. Unfortunately, I've already invested in my economy and production, and thus can't support units to defend the attack. I conceded defense for economy, and paid the price.
Now, let's look at Zerg in Starcraft II.
I am playing Zerg. I make a queen at 13 and inject 4 larva. 25 seconds later I have 4 - 7 available larva, depending on my strategy. I can make any combination of attacking units and drones. I can commit to attacking, see it isn't viable, and make 4 drones when the next injection finishes.
Zerg only makes concessions to itself. By Zerg starting a game by taking map control against a player who stayed safe and focused on economy, Zerg will only be behind relative to where they could have been. The larva inject is just so good at this point that you can rush, take map control, and then make workers, and still end up far, far, ahead of a player who just safely focused on making workers.
Warning: ESPORTS Science Follows
So that is my theory. Until a week ago, I had nothing else than my red name to back up my words. That's not enough, I needed numbers and graphs.
So I wrote a program to model Starcraft II's economy.
+ Show Spoiler [Results] +
Disclaimer and Assumptions
This wasn't designed to be accurate. It's not accurate; however, the inaccuracies are carried among all three races, meaning if we compare the races directly, we should get a fairly nice picture of the situation.
In the interest of full-disclosure and open-sourceness, I am releasing my "model". You can find it as an excel spreadsheet and VB Macro here. The VB Macro is safe and will not harm your computer, because I don't even know how to do that if I tried. I have only taken one programming course in my life. The code is ragged and inefficient. I encourage people to modify what's there, make it better, or rewrite something from scratch. Note that if you run the macro in Excel 2002 you may get an error about updating the XValues. I've found if you just ignore it, the graph updates anyway. All that piece of code is doing is setting the limits on the graph so you don't get 300 extra, useless data points. If it doesn't work, simply set the data range manually, ending at the last row of the simulation.
+ Show Spoiler [Known Glitches] +All variables are declared at the start of each Sub and should be able to be changed freely. Where there are limitations, they are listed in the comments. Each race sends their workers to mine (2 seconds to reach minerals, 3 to mine, 2 to return). In the case that there are no free minerals, the worker will wait. I have allowed no time for moving back and forth looking for free minerals. The AI will build overlords, supply depots, and pylons, at times which are hardcoded into the code. After that, they will build a spawning pool, barracks, gateway, queen, obselisk, and upgrade to orbital relay when it is "logical" to do so.
This simulation shows how long it takes for each race to reach saturation, which I've arbitrarily defined to be 30 workers, while using their race-specific macro mechanics. Essentially, this is a race to 30 workers, and nothing else.
![[image loading]](http://www.teamliquid.net/staff/Daigomi/Other__Terran_graph.jpg)
![[image loading]](http://www.teamliquid.net/staff/Daigomi/Other__Protoss_graph.jpg)
![[image loading]](http://www.teamliquid.net/staff/Daigomi/Other__Zerg_graph.jpg)
![[image loading]](http://www.teamliquid.net/staff/Daigomi/Other__Comparitive_graph.jpg)
As you can see, it's not even close. Once the queen has an effect on the battlefield (which is actually 30 seconds after the other mechanics in this model, indicated by the coloured vertical lines), the Zerg drone count heads to the sky and never looks back. Also note that the Terran and Protoss macro mechanics complete within 1 second of each other, which is why there appears to be only two lines. Zerg reaches 30 workers just under three minutes before Protoss and just over three minutes before Terran. Even if we assume the Zerg portion of the model is out by a full minute, the results are still obscene, and will only compound as time goes on.
True to Your Ancestry
An interesting result comes from this model, which I thought I'd mention in passing. Although the macro mechanic allows Zerg to get more of everything, basically, it doesn't allow them to get more minerals as they approach saturation. Because the Terran and Protoss mechanics are immune to being affected by saturation, Zerg are required to expand earlier to keep up with their income. Although that isn't the focus of this article, it is nice to see that Zerg will probably still be trying to stay one base up on his opponent in Starcraft II.
This Proves Nothing
By itself, posting about who can get to 30 workers first proves nothing. I'm sure that unchecked, Zerg can get 30 workers before the other two races in Starcraft. The problem is that the queen lets Zerg seamlessly slip from one strategy to another.
Again, let's imagine Starcraft. Your overlord arrives at Protoss' base and you see 2 gateways. It is natural to assume that he is going for an early attack, or else he wouldn't have invested in those. Even if he is playing mindgames and really focusing on economy, you will be ahead if you don't over-invest in defense because his gateways are useless to boosting his economy.
Similarly, if your probe arrives fairly late to Zerg's base and you see 2 hatcheries and a hydralisk den, you can be confident he is coming to break your expansion. Again, even if he doesn't, the fact that he stayed on 2 hatcheries has hurt his economy badly.
In Starcraft II, your probe arrives at Zerg's base to find 2 hatcheries and a hydralisk den. Two minutes after the probe dies, Zerg could theoretically have 20 hydralisks or 30 drones, and it's impossible for you to know.
This wasn't designed to be accurate. It's not accurate; however, the inaccuracies are carried among all three races, meaning if we compare the races directly, we should get a fairly nice picture of the situation.
In the interest of full-disclosure and open-sourceness, I am releasing my "model". You can find it as an excel spreadsheet and VB Macro here. The VB Macro is safe and will not harm your computer, because I don't even know how to do that if I tried. I have only taken one programming course in my life. The code is ragged and inefficient. I encourage people to modify what's there, make it better, or rewrite something from scratch. Note that if you run the macro in Excel 2002 you may get an error about updating the XValues. I've found if you just ignore it, the graph updates anyway. All that piece of code is doing is setting the limits on the graph so you don't get 300 extra, useless data points. If it doesn't work, simply set the data range manually, ending at the last row of the simulation.
+ Show Spoiler [Known Glitches] +
- Once the queen injects, all 7 larvae will eventually come back naturally without a second injection, if you wait long enough. This is irrelevant to me, since I'm always reinjecting every 25 seconds, long before all the larvae can return.
- There is an error updating the XValues of the graphs in Excel 2002
This simulation shows how long it takes for each race to reach saturation, which I've arbitrarily defined to be 30 workers, while using their race-specific macro mechanics. Essentially, this is a race to 30 workers, and nothing else.
The Results:
As you can see, it's not even close. Once the queen has an effect on the battlefield (which is actually 30 seconds after the other mechanics in this model, indicated by the coloured vertical lines), the Zerg drone count heads to the sky and never looks back. Also note that the Terran and Protoss macro mechanics complete within 1 second of each other, which is why there appears to be only two lines. Zerg reaches 30 workers just under three minutes before Protoss and just over three minutes before Terran. Even if we assume the Zerg portion of the model is out by a full minute, the results are still obscene, and will only compound as time goes on.
True to Your Ancestry
An interesting result comes from this model, which I thought I'd mention in passing. Although the macro mechanic allows Zerg to get more of everything, basically, it doesn't allow them to get more minerals as they approach saturation. Because the Terran and Protoss mechanics are immune to being affected by saturation, Zerg are required to expand earlier to keep up with their income. Although that isn't the focus of this article, it is nice to see that Zerg will probably still be trying to stay one base up on his opponent in Starcraft II.
This Proves Nothing
By itself, posting about who can get to 30 workers first proves nothing. I'm sure that unchecked, Zerg can get 30 workers before the other two races in Starcraft. The problem is that the queen lets Zerg seamlessly slip from one strategy to another.
Again, let's imagine Starcraft. Your overlord arrives at Protoss' base and you see 2 gateways. It is natural to assume that he is going for an early attack, or else he wouldn't have invested in those. Even if he is playing mindgames and really focusing on economy, you will be ahead if you don't over-invest in defense because his gateways are useless to boosting his economy.
Similarly, if your probe arrives fairly late to Zerg's base and you see 2 hatcheries and a hydralisk den, you can be confident he is coming to break your expansion. Again, even if he doesn't, the fact that he stayed on 2 hatcheries has hurt his economy badly.
In Starcraft II, your probe arrives at Zerg's base to find 2 hatcheries and a hydralisk den. Two minutes after the probe dies, Zerg could theoretically have 20 hydralisks or 30 drones, and it's impossible for you to know.
a >> Δv/t
If you accept what I've written above, you should expect to see the majority of games opening with a macro mechanic rush off one base. After this, the game reverts back to "standard" play with an accelerated income.
The more I've thought about it, the more I think this economic acceleration won't matter for any race but Zerg. You can support a ridiculous army off of one base in Starcraft II. For example, you can support 4 barracks with reactors off one base if you get the MULE quickly enough. Think about that. That's 8 barracks in constant marine production. That's a lot of marines. However, if you are able to scout 3 barracks with reactor addons, you know what's coming. If you see marines being rallied 8 at at time, you know what's happening.
The problem, as mentioned, is that Zerg don't have to show anything. Two hatcheries and a queen - That's enough to go anywhere in 7 minute from an all-in attack with 30 hydralisks, to pushing 60 supply in drones. When Zerglings are coming 28 at a time, and suddenly they stop, you can predict that Zerg has switched over to drone production again, but it's already too late for you to adapt.
We Can Rebuild; We Have the Technology
How does Blizzard fix the balance of the macro mechanics? Is it a problem? Does it need to be fixed?
Step 1: Make the mechanics equal. Assume everyone is going to rush for the mechanics, because that's what they're going to do in the current form. Find the number of larvae that balances the game. Make the Queen have to choose what unit it is going to inject, rather than having a ridiculous amount of choice in uncommitted larvae.
Step 2: Make the mechanics cost something. There are two options: lower the benefit of the mechanics or increase the cost. I would prefer to see the latter but both are viable. It needs some way to open up vulnerability in the early game to either timing attacks or tech rushes, so that the player is reacting to the opponent, as opposed to the current situation of rushing for the macro mechanic.
Step 3: Open up real choice. If I am giving up 200 minerals for not using my MULE, it had damn well better be doing something other than blinking some lights around for 5 seconds. Players are going to be giving up real income, and a lot of it, if they choose to use something other than the macro mechanics. They need to be given something that makes them think before they reactively slam another 8 larvae into their 2 hatcheries.
Step 4: Test it under the premise that good players are going to abuse it. Return to Step 1.
This being said, I really did like all three of the macro mechanics. A lot. They fit well and seem like they will eventually open up new timing choices when the game draws closer to balance. I really hope they are slightly tweaked and allowed to remain a fundamental part of the Starcraft II economy.
I am confident Blizzard will understand this issue. If they haven't already understood the issue by now, they surely will in the beta. I'm not concerned that it will slip by unnoticed, and you shouldn't be either. I do, however, still think it is an interesting problem in its current form.
