Related question (but not quite duplicate): How do I learn to effectively transfer to 2v2 hard AI as Zerg?

We're playing Zerg and Protoss and using co-op vs AI as a training ground to practice the macro cycle before experimenting on the ladder. We're having a problem defending against terran during the initial rush in 2v2 vs the AI.

We can both beat the hard and very hard AIs individually, and we can easily beat 2v2 hard AI by cheesing (the AI especially doesn't like being cannon rushed). However, as we're trying to practice macro, cheesing seems counterproductive.

The problem is, the hard AI predictably moves out around 6:30 with an aggressive timing push which is quite difficult to hold with "normal" openings. If we go up against two Zergs or Protoss, we can hold reasonably effectively. If we have one terran opponent it's a bit scrappy - if one of us fast-expanded we usually lose the expo, but we can usually kill the AI's expansions in return and go on to win. If we have two terran opponents we can't hold.

I've had a quick look over our replays and we're not doing anything differently in the matches we're struggling with. The causal factor seems to be the terran.

The timing push consists of around 10 marauders and 12 marines. It hits around 7 minutes. We'll usually have 8 gateway units (3 zealots, a stalker, a sentry, and one round of warpins from 3 gates), an immortal, and around 30 zerglings.

  • Do we need more (or different) units?
  • Is there an obvious counter I'm missing?
  • Is there anything you can think of that we're likely to be doing wrong in the opening?
  • Is 2v2 vs AI just not good for this kind of practice?

As I say, we're trying to practice macro, so we'd rather avoid doing anything obtuse - e.g. it's pretty easy to 1v4 hard AI if you cheese + exploit the AI's weaknesses (the AI hates being cannon rushed...), but that's not very good macro-cycle practice.

Thanks in advance!

  • 2
    This is why, when practicing a build, you always choose the very easy AI. It's naive to execute a build without any regard to what your opponent is doing. If you scout early aggression, react accordingly. Your problem is you are not reacting to your opponents. Maybe if you upload a replay someone can take a look at what you should do more specifically. If you just want to practice macro and not worry about your opponent, then choose Very Easy AI. If you want to play a reactive game, scout and react better.
    – tenfour
    Commented Oct 16, 2012 at 9:32
  • We're happy to react. If we dial it down to medium, it's so easy that we usually win accidentally before we've saturated two bases. We are scouting, and we always catch their push leaving their base at 6:30. We don't want to "not worry" about what the opponent is doing, we're just looking for a way to hold vs Terran without robbing the game of any macro-practice value by cheesing. Commented Oct 16, 2012 at 9:57
  • Isn't the reaction @tenfour is talking about some kind of turtling. If I understand well, you try to macro, scout an early agression, then react by turtling in order to be able to defend. So what about canons and spine crawlers to defend your expos?
    – tharibo
    Commented Oct 16, 2012 at 10:08
  • 1
    What builds are you executing? Standard 2v2 builds would probably be a gas/pool around 10-14 for Zerg and a 3gate robo expand for Protoss.
    – Decency
    Commented Oct 16, 2012 at 12:38
  • 1
    I just played the AI and I think I see the push you're talking about. Zerg can probably hold the entire thing with pure ling if you cut enough drones. Marauders are worthless when surrounded speedlings. My best guess is that your protoss ally is over-making stalkers. He should only make 1-2 of them; for the rest make zealot sentry immortal. For Z, pure ling with speed, maybe some banes (you should have the baneling nest anyway just in case). Don't bother with static defense unless for example you don't think you can hold the push so you make a couple spines by your main buildings.
    – tenfour
    Commented Oct 16, 2012 at 13:04

3 Answers 3


As you are expanding and the opponent (AI) is not you have to use static defenses (Spine crawlers / Cannons / Bunkers etc.), because you will usually have a unit disadvantage.

The logic works like this:

  • You expand.
  • You scout -> opponent does not expand
  • This means: Cheese / Timing Push (with or without tech) / All-In
  • Your task: Deduce the strategy of your opponent
  • Your response: build static defenses accordingly (cannons and spore crawlers vs air or cannons and spine crawlers vs ground)

Hold the attack, wait for your better economy to kick in and attack. Expand behind your push.

In your special scenario it might be enough to build 1 or 2 spine crawlers, transfuse with Queens and pull your workers.


Analyzing these forces:

5 marauder, 6 marine = 800 min, 125 gas, 16 supply
3 zealot, 4 stalker, 1 sentry, 1 immortal = 1100 min, 350 gas, 20 supply
"around" 30 zerglings = 750 min, 15 supply

The zerg player isn't carrying his weight. He needs to find a way to spend more resources on forces. If he is larva capped, then switch from zergling to roaches.

Yes, marauders do extra damage against roaches. It doesn't matter. Get more meat into the fight. Roaches give you nearly 2 health per mineral. Zerglings give you ~1.2 health per mineral. If you go exclusively roach - try to get at least 12 by the 7 minute mark.

Have your ranged attackers focus down the maruaders, while your meleers surround.

@tenfour, you are correct this answer is too theoretical. Here's some applied numbers from a couple of rusty 7 minute trial runs from this silver level player:

Lair + Queen + saturated gas + Zergling speed + 36 zerglings + 4 zerglings hatching


Lair + Queen + saturated gas + 11 roaches out + 3 roaches hatching

The zerg player needs to improve their macro. Zerglings or roaches don't matter, unit spending matters. As Day9 would say: sometimes the counter is "you just kill it".

Zerg macro - time your overlord hatching so you never have 3 larva waiting (3 larva waiting = no new larva from hatchery). Use the queen's ability on cooldown. Sneeze and hatch, sneeze and hatch.

  • 1
    It's not really a fair assessment I think. You need to consider the build, the map, the strategy for defense, your preferred transition. For example if Zerg has enough larvae, forget roaches - speedlings will beat this every time. Or if zerg wants to go for an early tech and save gas, again favor speedlings. If your strategy is to mass roach then by all means though... Looking at the resource cost is too theoretical I think. Speedlings basically counter marauders and are the fundamental zerg unit. This is why nobody rushes mass marauder against Zerg.
    – tenfour
    Commented Oct 16, 2012 at 19:45

Turns out Tenfour's advice in the comments under the question was correct.

I'm sure some static defenses at our natural(s) (or at the a choke ahead of our naturals) would also work, but cutting stalkers in favour of zealots, making better use of force fields, and chronoboosting out the second immortal works a charm with either roaches or speedlings.

I haven't tested it, but I suspect just cutting stalkers in favour of zealouts would do the job on its own. We can hold the initial push now even if one of us screws something up.

Once you've held off that push, the hard AI is basically beaten.

We're working now on both the very hard AI, and on getting the confidence to start on the ladder.


@tenfour: If you post your comment as an answer, I'll delete this one and accept yours instead.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .