Scouting: work out what they have without seeing it

Lesson

7 of 14

Watch

The lecture, 37 minutes

Practise

A checkpoint i a plan for your next game

Most players scout to see what the other player has. Good players scout to find out what they cannot have, and where the rest of their money went. After this lesson you will know what to look for on each scout, how long a sighting stays true, and how to turn "they spent about 500 minerals I did not see" into a decision.

What you will learn

• The one equation behind every scout: what they have is what you saw, plus what you did not see.
• Why the size of what you missed is easy to work out, and the direction is the part that beats you.
• How to draw your opponent's options as a budget line and read your safest build off it.
• Why every sighting starts to go stale the moment your scout leaves.
• When to scout, and what each thing you see rules out.

Why pay for a scout at all

A scout is not free. A worker out on the map is a worker not mining: at about 42 minerals a minute, a one minute trip costs you 42 minerals, plus the worker if it dies. An overlord in the wrong place can cost 100 minerals and a supply block.

You pay because the alternative is guessing, and a wrong guess costs far more. Cannons nobody attacks, an expansion that dies to zerglings, mutalisks in your mineral line with no turret up. Everything in this lesson is about getting more out of each trip.

Two kinds of choice

Every decision your opponent makes is one of two kinds ►35:00 , and your scout looks for them differently.

Discrete choices pay in full or not at all. There is no eighty percent of a spire. So your scout is asking a yes or no question: is the spire there. One look answers it.

Continuous choices pay by how long they have been running ►4:00 . An expansion taken at 3:30 has mined for a minute longer than one taken at 4:30. So for a continuous choice, the useful thing to scout is not whether, but when. The same idea drives timing attacks in lesson 12.

Their future is a cone

Write everything a player can count as one list: minerals, gas, supply, workers, army, upgrades ►8:00 . That list is their position.

You know where they were when you last looked. From there they can move a distance $n$ in any direction, and $n$ grows every second. That spreading set of possibilities is a cone. A scout cuts it back to a point: your overlord arrives, you count nine drones and one hatchery, and you now know their zergling count is zero.

The one equation behind every scout

What your opponent has is what you saw, plus what you did not see ►12:00 . It sounds obvious. Written down, it splits scouting into two separate jobs.

$R_{\text{obs}}$ is what your scout wrote down. More scouting makes it bigger.

$|R_u|\,\hat{u}$ is everything else, and it has two parts.

• The size, $|R_u|$. You can work this out without seeing anything. You know roughly how many workers they have and what a worker mines in a minute. A clock and some arithmetic give you the total.
• The direction, $\hat{u}$. Which way they spent it. You cannot compute this. You have to look, and it is the part that decides the game. Two hundred unseen minerals is eight zerglings, or it is the minerals for a lair. Same size, opposite games.

So a scout's real job is not "find out how much they have". It is "find out which way they spent it".

Draw it as a budget line

Put drones on one axis and zerglings on the other ►16:00 . Everything they could have built with the unseen money sits on one line: every point on it costs the same.

A worked example with real prices

Here is the budget line with numbers on it. The figures are rounded on purpose: you need to do this in your head during a game.

Your overlord looks into their main at 3:00. It sees nine drones, one hatchery, a spawning pool and nothing else. You do not look again until 5:00.

1. The size. Nine drones at about 42 minerals a minute, for two minutes, is about 750 minerals.
2. Take off what you saw. The spawning pool cost 200. That leaves about 550 you did not see.
3. The directions. Now list what 550 minerals can be.

Every row costs the same. That is the easy half, and you did it with a clock. The rows are the hard half, and only looking tells you which one you are in.

Now read the table for your own build. Three of the four rows are held by the same handful of units at your ramp. The fourth, the lair, needs a completely different answer. So that is the one your next scout must check for: look at their gas.

Pick your build off the line

Put their economy on one axis and army on the other, and take a decision every Protoss player knows: nexus first against Zerg, or forge and cannons first ►23:00 .

They are not two builds. They are two points on the same line: roughly the same 400 minerals, spent in different directions.

Now put the Zerg on their line. Say their unseen money is anything from zero to twelve zerglings. For each point on their line you can read off the smallest defence that keeps you alive. That reading tells you three things: what you can safely do, what you can only do if you guess right, and what loses on the spot. Your scout's only job is to say where on the line they are.

Information goes stale

A sighting is true at the moment you make it, and only then ►23:00 . After that the gap reopens at the rate they produce.

A rule you can use mid game: count their mining workers and multiply by about 42. That is how many minerals a minute are going somewhere you cannot see. Twenty workers means about 840 a minute. Three minutes without a look means about 2,500 minerals of army or tech you are guessing at.

There is a bright side. A strong player wastes nothing, so they are always on the line and your scout only has to find the direction. A weaker player often floats money and sits inside the line, so on the ladder the thing you are scouting for is usually smaller than the worst case. Plan for the worst case, but do not panic about it.

What to look for, and what it rules out

Every look should answer a question. These are the ones worth asking.

Common mistakes

• Scouting to look, not to decide. Before you send a scout, name the question it answers. If no answer would change your build, save the worker.
• Trusting an old look. A sighting from two minutes ago is two minutes of production out of date. Do the workers times 42 sum.
• Counting only what is there. The most useful thing a scout finds is what is missing: the empty gas, the missing natural, the army that is not at home.
• Scouting the size and not the direction. You rarely need an exact count. You need to know tech or no tech, bases or army.
• Preparing for everything. You cannot afford the answer to every row. Scout for the row that kills you, and play greedy against the rest.

Watch the lecture

The full lecture with chapters; the default order plays it from the start of the class, and the short cut skips the announcements.

Practise

• The Scouting Lab. Set when you last looked and what you saw, and watch the budget line grow until you look again. Try the 3:00 example from this page. Open the Scouting Lab.
• Replay drill. Open a replay of one of your losses. Pause at your last scout before their attack. Write down what you saw, then do the sum: workers times 42 times minutes until the attack. Compare it with what hit you.
• Timed scouts. Play three games where you scout at fixed times, for example on your first overlord or worker and again just after 4:00. Each time, say out loud what you saw and what it rules out.
• Guided reading. The theory guide Scouting is subtraction goes deeper on the same idea.

Checkpoint