Macro in StarCraft Brood War explained

Newcomer Learner

The short version

Macro is not how much you mine. It is how much you spend, and how quickly.
A bank is not power. Minerals in the bank are minerals that have not become anything yet, and they never will in time for the fight that is happening now.
When you are floating money, the cause is usually not enough production, not not enough workers.
The test of an economy is whether it produced units that were in the right place at the right time. Supply that arrived late is supply you paid for and did not use.
Growth compounds, which is why an expansion taken while you are safe is worth several times the same expansion taken while you are scared.

The situation

Twelve minutes in, you look at the top of your screen and there are fourteen hundred minerals sitting there.

You know it is bad. What you probably do not know is which mistake it is, and there are three completely different ones that produce the same number. You might not have enough production buildings, in which case more workers makes it worse. You might have production buildings and not be using them, in which case it is an attention problem, not an economy problem. Or you might be saving on purpose for something, in which case the question is whether the thing you are saving for arrives before the thing they are building.

Those need three different fixes, and the score screen calls all three of them "resources collected".

Macro's purpose, in one line

Lesson 6's answer, and the best line in either macro lesson: macro's purpose is units you can actually use.

Every word in that does work.

Units, not minerals. A mineral is potential. A unit is a thing on the map.

You can implies control. Forty supply you cannot look at is not forty supply. This is the bridge to T10: production and attention are the same budget seen from different ends.

Use implies place and time. An army that finishes building ninety seconds after the fight is an army that was not in the game. T2 has the arithmetic for why arriving in pieces is worse than not arriving.

So a good macro game is not one where you mined a lot. It is one where the things you bought were present, together, in the place that mattered.

Spending is the metric, not income

The number worth watching is not what you collected. It is what you had in the bank, and for how long.

Think of every mineral as being paid interest, in the form of whatever it would have been doing if you had spent it. A mineral sitting in the bank for sixty seconds has lost a minute of being a zealot, or a minute of being a worker mining more minerals. Add that up over a game and it is usually the largest single loss on your ledger, larger than any fight you lost.

So the question to ask about your own macro is not "am I mining enough". It is "how long does a mineral stay in my bank after I earn it". If the answer is more than a few seconds, you have found the thing to fix, and the fix is almost never more workers.

The three causes of a float, and their three fixes

Not enough production. Your income is larger than your production buildings can consume. This is the most common one by a wide margin, and the giveaway is that the float grows steadily rather than in jumps. The fix is more gateways, more barracks, more larvae. It is not more workers, which makes the float grow faster.

Not enough attention. You have the buildings and they are idle. The giveaway is a float that jumps: flat while you are at home, climbing every time you look at the map. The fix belongs in T10, and it usually starts with production hotkeys rather than with trying harder.

Deliberate saving. You are banking for an expansion, a tech switch or a reinforcement wave. This one is legitimate and it is still a cost. The question is whether the thing you are saving for arrives before the thing they are building, which is a T4 question.

Naming which of the three you are doing, out loud, during the game, is worth more than any build order refinement available to you.

The working

Show the maths

Why production capacity is usually the binding constraint. Let income be $I$ minerals a minute, let your production buildings be able to start $P$ units a minute, and let the average unit cost $\bar{c}$ . Your bank grows at

\frac{dB}{dt} = I - \min\!\left(I,\; P\,\bar{c}\right)

(1)

You can only spend as fast as you can start things.

So the bank is flat while $I \le P\bar{c}$ and grows at $I - P\bar{c}$ after that. Adding workers raises $I$ and does nothing to $P$ , which is the formal version of "more workers makes it worse". The break even count of production buildings is

P^{*} = \frac{I}{\bar{c}}

(2)

Income divided by the price of what you are making. Cheap units need more production, not fewer.

Note the direction of that last one, because it surprises people: the cheaper your army, the more production buildings the same income needs. A zergling economy and a carrier economy at the same income are not the same macro problem.

What an expansion is worth, honestly. A base costs $C$ , takes $\tau$ to build, and eventually adds $\Delta I$ of income. The naive payback is

T_{\text{payback}} = \tau + \frac{C}{\Delta I}

(3)

but that is not the real bill. During $\tau$ and for the payback after it, the same money could have been an army, so the true cost of expanding is the army you did not have in exactly that window. Which gives the rule every strong player follows without deriving it: expand when the window is safe, not when you are rich. Being rich tells you that you can afford it. It tells you nothing about whether you survive the next ninety seconds.

Support units divide the death rate. This is the lesson 5 worked example, and it is the prettiest small result the course produced. Take a unit with $H$ hit points taking damage at $\beta$ a second. It dies every

t = \frac{H}{\beta}

(4)

so your losses run at $\beta/H$ bodies a second. Now add healing at $\alpha$ a second. The loss rate becomes

\frac{\beta - \alpha}{H} \quad\text{and the effective health becomes}\quad H_{\text{eff}} = H\,\frac{\beta}{\beta - \alpha}

(5)

A medic does not add hit points. It divides the rate at which you lose them.

Read the second expression. Healing that keeps up with half the incoming damage doubles your effective health. Healing that keeps up with three quarters of it quadruples it. And when $\alpha \ge \beta$ the denominator goes to zero and nothing dies at all, which is why a small amount of incoming damage against a healed army is not a small amount of progress, it is none.

This is also why support units break the efficiency matrix in T8. A medic's own row is zero kills. Its real contribution is a multiplier on everybody else's, and a linear model has nowhere to put that.

Growth compounds, and that is the whole argument for economy

T1 has the derivation and it is worth restating here in words, because this is the guide where the conclusion lives.

Money spent on economy buys more money, which buys more money. Money spent on army buys a fixed amount of army. Over a long enough game the first one wins by an amount that is not close.

But the compounding is capped, and the cap is production capacity, which is why on one base the exponential phase is over in the first minute and everything after that is linear. The only way to buy back the compounding is another base, which adds mineral patches and production room at the same time.

That is the entire economic argument for expanding, and it is also the reason a player who never expands can feel fine for ten minutes and then lose without ever being outplayed in a fight.

What is not macro

Three things get filed under macro and are not.

Mining. Workers mine on their own. Nothing about mining is a skill after the first twenty seconds of the game.

Having a lot of stuff. Stuff you cannot use is not an advantage. See the definition above.

Never fighting. An economy that is never converted into pressure gives the opponent a free choice about when the game is decided, which is worth more than the economy is.

Where this comes from in the course

Lesson 5 teaches macro in practice, starting from the marine and medic problem that is the healing derivation above. Lesson 6 teaches what macro is for, and the phrase at the top of this page is its summary.

Common mistakes, in the order most players make them. Building workers and forgetting production, so the bank grows. Building production and forgetting to look at it, so the bank grows anyway. Saving for a tech building and not noticing that the army you skipped would have won the fight that decided the game. Each one shows up as a float, and each has a different fix.

The frame the whole thing hangs from is the resource taxonomy in the lesson 1 lecture at 1:06:00, where macro is named as a transformation resource, which is to say a machine for turning one thing into another rather than a thing in itself.

Still true today

Checkpoint

You are floating 1,200 minerals and your gateways are all busy. What is the fix?
A medic heals at half the rate your marines are taking damage. What does that do?
When is the right time to take an expansion?

Sources for this page

David Sirlin's write ups of weeks 5 and 6 His notes on lessons 5 and 6.
Lesson 1 lecture, on YouTube, channel VioleTAK The resource taxonomy, where macro is named as a transformation resource