What micro is in StarCraft Brood War

Learner Competitor Advanced

The short version

Micro is firing the most weapons for the longest period of time. That is the definition the class chose, and it is better than the one you already have.
It is behavioural on purpose. It says nothing about clicking, and everything about output.
Every micro technique you have ever seen is a way of raising one of those two factors.
Retreating is micro. Target selection is micro. Standing still is sometimes micro.
Micro's payoff grows with the square of the length of a fight. Economy compounds over the length of a game. That comparison is how you decide where your attention goes.

The situation

Someone tells you your micro is bad. You go and practise clicking faster.

Six weeks later your clicks are faster and your fights go the same way, because speed was never the thing that was wrong. What was wrong was that four of your units spent the fight walking, two of them were shooting a full health unit instead of the one on two hit points, and you pulled the whole group back at the moment it was winning.

None of those are speed problems. All of them are definition problems.

The definition, and the one it beat

The class polled the room first. Students offered the answers everybody offers: unit control, maximising the effect of your units, controlling units manually instead of letting the game's AI do it.

That last one is the definition most players carry, and the lecturer did not take it.

Micro is firing the most weapons for the longest period of time.

Alan Feng Lesson 3 lecture 30:20

Why this one is better: it is about output, not about input. "Manually controlling your units" is a description of what your hands are doing. It cannot tell you whether what your hands did was right. "Firing the most weapons for the longest time" can, because it names a quantity that goes up or down, and every decision in a fight can be scored against it.

The lecturer invited the room to argue, and meant it. A student objected that target selection, picking off the weakest unit, must also be micro and is not obviously covered. The answer is that it is covered, through the second factor: killing their units faster means fewer of their weapons firing, which is the same term seen from the other side. A second student pointed out that microing a worker out to scout is micro too, and that one the lecturer conceded outright and bracketed as a different subject for a later lesson.

Both objections are good, and a definition that answers its best objections is one worth keeping.

The five corollaries

Five rules follow from the definition. In the lecturer's order:

Keep things alive as long as you can. A unit that survives keeps firing, and that is the whole of it.
Retreat, for future battles. A unit pulled out at thirty hit points is a whole unit in the next fight. A unit that stays is a corpse and about four more seconds of damage.
Maximise regeneration time, and minimise your opponent's. Shields, healing and Zerg regeneration are free hit points paid for by the clock.
Maximise splash damage, and minimise splash taken. Both directions, and the second is the one players forget.
Dodge attacks where you can. Anything that lands on empty ground was free.

Read them as five ways of moving the same two numbers. One, two and three are about the length of time. Four and five are about how much of the enemy's output lands.

Three things professionals do, and why

The second half of that lecture was three professional replays, watched against the five rules. What each one actually demonstrates is more useful than the clips themselves.

Retreat when the local trade is bad. In the first clip, a Protoss facing a bigger gateway count retreats a zealot whenever that particular zealot is about to lose its particular fight. The lecturer's framing is the good one: he is not avoiding damage, he is reducing the opponent's effectiveness. Two or three enemy units chasing a retreating unit are two or three units not killing anything, and the chase is free damage taken on the way.

Put the units that deal damage where they take it last. In the second clip, a player under a dark templar attack with no production left sends his probes in first. Not as a sacrifice: as an ordering. Whatever hits hardest should be the last thing to take a hit, so it keeps firing for longer, which is corollary one serving the axiom directly. Anything that shifts incoming damage onto the part of your army with the lowest damage output is micro, including deliberately losing cheap things.

Know what the fight is buying. The same clip has the best economics in the lecture. Protoss shields regenerate, so stretching a fight out means more effective hit points, which sounds like a free win for the defender. But every second spent not attacking is a second the opponent spends mining, and mining buys units. So regeneration is real and it is bought with the enemy's economy. That is lesson 1's "every conversion costs something" applied to a decision you make with your mouse.

Go where they are not. The third clip is marines against lurkers, and it is mobility against immobility. Lurkers must burrow to fight, so the answer is simply to be somewhere else. The lecturer kept calling this running away, including when the movement was toward the opponent's base, and he was right to: the name is about the damage you are refusing to take, not about the direction you are walking.

The working

Show the maths

The axiom, written down. Let $n(t)$ be how many of your units are firing at time $t$ . Micro maximises

W = \int_{0}^{T} n(t)\,dt

(1)

Weapons fired, integrated over the length of the engagement.

Every corollary is a term in that integral. Keeping units alive and retreating raise $T$ . Flux, from T3, sets the ceiling on $n$ . Dodging and splash control lower the enemy's $W$ without lowering yours. There is no sixth kind of micro hiding anywhere: the definition is exhaustive by construction, which is what makes it a definition rather than a list.

Why target selection is already in there. Your losses depend on the enemy's firing count, and the enemy's firing count falls only when enemy units die, not when they are wounded. Spreading damage across five units at eighty percent health lowers nothing. Concentrating it kills one and lowers their $n$ by one immediately. The volley model in T2 is this argument with numbers: the gap between concentrating and spreading grows as

\text{gap} \;\sim\; \frac{N_1 D}{2\,HP}\,M^{2}

(2)

Quadratic in the number of volleys, which is to say quadratic in how long the fight lasts.

Now price it against macro. A fight of length $M$ volleys pays micro something like $k_1 M^2$ , and it is over when the fight is over. The same attention spent on macro pays $k_2 e^{\gamma t}$ and keeps paying for the rest of the game. An exponential beats any polynomial eventually, so:

In a short game, or in a fight that ends the game, micro wins the comparison outright.
In a long game, the macro term runs away and no amount of micro catches it.
The crossover is not a number you compute mid game. It is a habit: spend attention on micro in fights that decide something, and on macro in every other second.

The lesson 4 lecture makes exactly this comparison, with growth rates rather than with opinions. It is the best single argument in the course for where your attention belongs.

What to actually do

The definition earns its keep by generating habits. Five that cost nothing and are worth a rank.

Stop before you engage. Units still resolving their pathing are not firing. Arriving stopped, in the shape you want, is worth more than any click you could make afterwards.

Kill things, do not damage things. Pick a target, finish it, move on. Splitting fire across a line feels efficient and lowers nothing.

Pull the wounded one out, not the group. The group retreating loses the fight. One unit stepping back keeps the fight and saves the unit.

Count what is shooting, not what is there. If the answer is less than most of your army, the problem is a T3 problem and no amount of clicking fixes it.

Do nothing, on purpose, sometimes. Micro has a cost, and the cost is the macro you were not doing. A fight you were going to win anyway does not need you.

What the lecture said

Micro is defined from 27:00, with the axiom going on the board at 30:20 and the five corollaries read off the handout immediately afterwards. The three clips run from 36:00 to the end of the lecture.

The bridge to lesson 4 is at 34:00, and it is the sharpest thing said in either lecture: the mathematics is not there to make your micro better. It is there to price it. If you spend three hundred actions a minute on micro alone, what is that worth against spending the same attention on having a bigger army. The model that answers it is the heart of lesson 4.

Still true today

Checkpoint

Which of these is not micro, under the class's definition?
Why does spreading your damage across five wounded units lower nothing?
The game has gone forty minutes and you are on four bases. Where does the next ten seconds of attention belong?

Sources for this page

Lesson 3 lecture, on YouTube, channel VioleTAK Micro is defined from 00:27 and the three professional clips run from 00:36
Lesson 4 lecture, on YouTube, channel VioleTAK The volley model at the start of the recording is where the quadratic comes from