Countering

Learner

The short version

• A counter costs three things: money, time on the tech tree, and whatever you did not build instead. Most players price only the first.
• Counters arrive late by construction. The thing you are answering was already being built when you found out about it.
• Prepare in proportion to what the threat would cost you, not to how likely you think it is.
• Over answering is the more expensive mistake. The right amount of defence is enough to survive the window, not enough to win outright.
• A hard counter is a zero in the table from T8. A soft counter is just a bad exchange rate, and those are fixable with numbers.

The situation

Your scout sees a starport. You stop what you were doing, take the detection, build the anti air, and hold the position you were going to hold anyway.

Nothing comes. It was one starport, built partly because it is on the way to other things, and you spent six hundred minerals and ninety seconds of production on an answer to a question nobody asked. You are now behind, and the replay will show you losing a fight at fifteen minutes that was decided at nine.

That is not a scouting failure. It is a pricing failure, and it is the most common way a reasonable player loses a game.

The three costs of an answer

Money. The obvious one. The units and buildings you bought.

Tech time. The less obvious one. Most counters sit behind a building, and the building has a build time, and during that build time you have neither the counter nor whatever else that money would have been. Tech time is why counters arrive late and why the deadline matters more than the price.

What you did not build. The one that decides games. Every counter is also a non-expansion, a non-upgrade, or four fewer units of the thing you were already making. In a game where economy compounds, spending on a defensive answer at eight minutes is a choice you keep paying for at fourteen.

Write all three down when you make the decision and the decision changes. Not always toward building less: sometimes toward building the answer immediately, because tech time means later is much worse than now. The point is that you are pricing it rather than reacting to it.

Counters arrive late, always

The structural fact underneath everything here.

They decided at time zero. You found out at time one. You start building at time one and your answer exists at time two. The threat exists from somewhere around time one and a half.

So the interval where you are exposed is not a failure of your play. It is built into the shape of the information, and the only three levers on it are: scout earlier (T4), pick builds whose answers are cheap and fast, or hold ground that covers several answers at once.

That third one is why players who look conservative are often not. Holding a position that answers three things badly is frequently better than answering one thing well, precisely because the answer arrives on time.

How much is the right amount

Here is the result worth the whole guide, and it is not intuitive.

You should prepare for a threat in proportion to how much it would cost you, not in proportion to how likely you think it is.

Think about why. Your opponent is choosing what to do, and they can see what you prepared. If you always prepare for the cheap threat and never for the expensive one, they will simply always bring the expensive one. The only allocation they cannot exploit is the one that makes them indifferent, and the arithmetic below says that allocation weights each threat by the damage it would do.

Which means: the thing that would end the game on the spot gets defence even if you think it is unlikely. The thing that would cost you two workers can be ignored even if you think it is coming.

The working

Over answering

The opposite error, and the more expensive one at every level above beginner.

Three shapes it takes.

Answering a signal rather than a threat. T5 is the whole guide on this: the cheapest lies in the game are the ones where the liar's bill was paid by something they wanted anyway, and your bill is not. If the evidence for a threat is one building, the correct response is usually a partial answer plus another look, not a full answer.

Answering permanently to a temporary problem. A window is a window. Defence that covers the next ninety seconds and then becomes useless is a legitimate purchase. Defence that changes your whole army composition for a window is not.

Bringing the whole army home. The most expensive version, and the one an opponent can trigger on purpose for the cost of a few units. See the three kinds of answer in T5.

The rule that falls out: answer with the cheapest thing that survives the window, and get another look before you spend more. Defending is cheaper than attacking, so the cheap answer is usually enough, and the money you did not spend is still available if the second look says you were wrong.

What the lecture said

The class framed countering as a subject with costs: every answer you build is something else you did not build. The derivation above is ours, and it follows from that framing.

The lecture that goes furthest into this is lesson 3's section on imperfect information, from ►21:00. That is where the bounds live: what they can have is capped by their resources divided by the unit cost, zeroed by what the tech tree forbids at this moment on the clock, and weighted by everything else you know. A counter decision is a decision made against that bounded set, and it is far more tractable than it feels, because most of the set is zero.

The other half, which lesson 11 teaches and T5 carries, is that a competent opponent knows you are making this decision and will spend money to move it. Countering and deception are the same subject from two directions.

Still true today