Before anyone writes code, they think about the problem. Exam papers name three techniques.

• Abstraction: remove the details that do not matter, so you can focus on the ones that do. A map of a railway shows which stations are on each line and where lines meet. It leaves out the bends in the track, because a passenger does not need them.
• Decomposition: break a big problem into smaller problems that are each easier to solve. A quiz app becomes: store the questions, ask them, check the answers, keep the score.
• Algorithmic thinking: work out the exact steps, in order, that solve the problem. "Check the password is at least 8 characters, then check it has a digit, then..."

Abstraction leaves things out. Decomposition splits things up. Algorithmic thinking puts steps in order.

A structure diagram draws a decomposition: the whole problem in the top box, its subsections in a row below, and the smaller tasks of each subsection below that. Each box can be planned, written and tested on its own, often as its own subroutine.

An algorithm is a sequence of steps that solves a problem. You can write one in three ways: numbered steps in plain English, a flowchart, or code (pseudocode or a real programming language).

Flowcharts use five symbols. You need to recognise all of them.

• A decision has two arrows out of it, labelled Yes and No.
• A process is an action or a calculation, such as "total = total + price".
• A subroutine box runs another part of the program, then comes back.

Most programs take something in, do something with it and give something back.

• Input: the data that goes in (typed, read from a sensor, loaded from a file).
• Process: what the program does to it (calculations, decisions, sorting).
• Output: what comes back (shown on screen, printed, sent, saved, a sound).

A program that asks for a room's length and width, multiplies them and shows the area: the inputs are the length and width, the process is the multiplication, the output is the area.