STUDY GUIDE · 3 INTERACTIVE LABS · ABOUT 12 MIN

Adding people to a late project makes it later.

Fred Brooks learned that the hard way managing IBM's OS/360, and wrote it down in 1975. Software engineering is the discipline of building large programs with teams: planning, designing, testing and changing code without breaking it. This guide lets you try the key ideas.

RELATIVE COST TO FIX A BUG, BY WHEN IT'S FOUND
Requirements1×
Design5×
Coding10×
Testing20×
In production100×
Commonly cited ratios; real multipliers vary widely by project. The lesson holds: catch problems early.

What is software engineering?

Software engineering is the systematic design, building, testing and maintenance of software, especially large systems built by teams over years. Writing code is only part of it; most effort goes into requirements, design, reviews, testing and changing existing code.

The term was popularised at a 1968 NATO conference called to tackle the "software crisis": projects running late, over budget and full of bugs. Many of its lessons are about people and communication as much as technology.

RequirementsWhat should it do, and for whom?
DesignArchitecture, modules and interfaces.
ImplementationWriting and reviewing code.
TestingUnit, integration and user tests.
DeploymentReleasing to real users.
MaintenanceFixes and changes: often most of the lifetime cost.
LAB 1 · BROOKS'S LAW

Every new person adds a conversation with everyone else

A team of n people has n(n−1)/2 possible communication lines. Each one costs time in meetings, messages and misunderstandings. Grow the team and watch the lines multiply faster than the people.

6people
15communication lines
15%time spent coordinating (model)
TEAM OUTPUT (ILLUSTRATIVE MODEL)
2team size30
Small teams communicate cheaply: this is why many companies cap teams at around 'two pizzas' of people. Brooks also noted that newcomers need training from existing staff, which slows a late project down even further.
LAB 2 · REFACTORING

Clean up code without changing what it does

Refactoring improves a program's structure while keeping its behaviour identical, ideally with tests proving nothing broke. Apply each step to this checkout function and watch the metrics.

function checkout(c, i) { let t = 0; for (let x = 0; x < i.length; x++) { if (i[x].type == "book") { t = t + i[x].p * i[x].q * 0.95; } else if (i[x].type == "food") { t = t + i[x].p * i[x].q * 1.0; } else { t = t + i[x].p * i[x].q * 1.2; } } if (c.member == true) { if (t > 100) { t = t * 0.9; } } if (t > 50) { return t; } else { return t + 4.99; }}
Lines of code16
Cyclomatic complexity of checkout()6
Duplicated expressions3
Original:it works, but cryptic names (c, i, t, p, q), repeated arithmetic and nested ifs make it hard to read and risky to change.All 6 tests still pass
LAB 3 · PROCESS MODELS

Plan it all, or ship in small slices?

week 1week 12
Plan
Build
Test
Review
Users get value
STRENGTHSWorking software every two weeks. Feedback arrives early, so wrong turns are cheap to fix.
RISKSScope can drift, and long-term architecture may be neglected without discipline.
BEST FORProducts whose requirements are uncertain or changing: most consumer and web software.

Core ideas at a glance

Version controlGit records every change, so teams can work in parallel on branches and merge.
Code reviewAnother engineer reads each change before it's merged; one of the most effective ways to catch bugs.
Unit testAn automated check of one small piece of code in isolation.
Technical debtShortcuts that speed you up now but cost extra later, like interest on a loan.
RefactoringRestructuring code without changing its behaviour.
CI/CDContinuous integration and delivery: every change is automatically built, tested and ready to ship.
Design patternsReusable solutions to common design problems, such as Observer or Factory.
Separation of concernsEach module does one job, so changes stay local.

From the software crisis to DevOps

1968The software crisis.A NATO conference in Garmisch popularised the term 'software engineering'.
1970Waterfall described.Winston Royce's paper sketched the sequential model, while warning about its risks.
1975The Mythical Man-Month.Fred Brooks's essays on why adding people to late projects backfires.
1994Design Patterns.The 'Gang of Four' catalogued 23 reusable object-oriented designs.
2001Agile Manifesto.Seventeen developers in Utah valued working software and responding to change over rigid plans.
2005Git.Linus Torvalds wrote it in about two weeks to manage Linux development.

Further reading

The Mythical Man-MonthFrederick P. Brooks Jr. · classic essays on teams and schedules
RefactoringMartin Fowler · the catalogue of code improvements
The Pragmatic ProgrammerHunt & Thomas · habits of effective developers
Test yourself: 20 software engineering questionsAlgorithms guide

Good design is clear thinking. Measure yours.

Our full IQ test covers logical, numerical, verbal and spatial reasoning: 60 questions, about 30 minutes.

Take the full IQ test