Brooks's law states that adding people to a software project that is already late delays it even further. Frederick P. Brooks formulated it in 1975 in "The Mythical Man-Month" ("Le Mythe du mois-homme" in French), a classic of software engineering (Wikipedia (opens a new window)). Three effects explain it: training newcomers, the growing volume of communication and tasks that cannot be shared out.
The basic principle: a difficulty that grows with team size
The bigger a team gets, the more time everyone spends coordinating and the less time is left to produce. Brooks takes aim at the "man-month", the unit that makes it seem that work taking one person n months would take n people a single month. Three mechanisms contradict that idea:
- Training: every newcomer has to learn the project, and that time is taken from the people already on the team.
- Communication: for n people, there are n(n-1)/2 possible communication channels. With 50 developers, that makes 1,225 channels (Wikipedia (opens a new window)).
- Indivisible tasks: some tasks cannot be shared out, whatever the headcount. Brooks cites a proverb that has remained famous: "Nine women cannot make a baby in one month."
Productivity per person therefore falls as the team grows. Total output can still rise on a task that is easy to divide, until everyone gets in everyone else's way.
Delays in IT projects: why late reinforcements slow things down
Late reinforcements slow a project down at first, because they tie up the people who are moving it forward. The team already in place has to train the newcomers and correct their mistakes just as the deadline approaches.
That said, the law does not forbid hiring. As Welcome to the Jungle (opens a new window) points out, integrating new people has a cost in time and resources, and that cost has to be accepted. Better to plan for it early than to discover it at the finish line. When the delay is already there, narrowing the scope or pushing back the date is better than looking for last-minute reinforcements.
The limits of Brooks's law: evolving working methods and technology
Modern methods and tools soften Brooks's law without abolishing it: they act on team size, on how work is divided and on how information flows.
Modern working methods
Agile and Scrum approaches partly address the problems Brooks describes. The Scrum Guide (2020 version) aims for teams of 10 people or fewer as a general rule (Scrum Guide (opens a new window)). Its authors note that small teams communicate better and are more productive. Three practices point the same way:
- Splitting work into small batches: each person works on a clear part, which limits dependencies.
- Short iterations: a slip against the schedule shows up early, when correcting it costs little.
- Short daily check-ins (stand-ups): information flows without endless meetings.
The role of technology in collaboration and productivity
Collaborative tools (shared code repositories, messaging tools, project management software) circulate information in real time, including remotely. They make communication easier, but do not replace the time needed to understand a project.
AI revives the question: a coding assistant or agent is like an extra pair of hands added without a hire. Google Cloud's 2025 DORA report shows that AI acts primarily as an amplifier, and that the best returns on investment come less from the tools than from the organisation around them (DORA (opens a new window)).
In an essay republished by O'Reilly in March 2026, Wes McKinney observes that coordination does not disappear with agents: it changes form, because several agent sessions can produce contradictory plans that humans have to reconcile (O'Reilly (opens a new window)).
Team size and composition therefore remain core decisions that neither methods nor tools erase.
The importance of structuring and organising work well
A project that is well scoped from the outset has less need of emergency reinforcements. Three steps help:
- Draw up a precise requirements document that lists the objectives and the specifications.
- Determine the resources needed (human, technical, financial) and check feasibility.
- Plan each stage with margins for the unexpected.
Also check that the delay is real: an overly optimistic initial schedule gives the impression of a delay where there is none. Some constraints linked to a team's growth remain unavoidable, but rigorous organisation limits the risk of delays and budget overruns, and keeps the project under control.
To strengthen a tech team at the right time, hand the role to a headhunter who specialises in your field: at Muzzo, the signature comes 26 days after the brief on average. Launch a mission (opens a new window).
FAQ
What is the formula for Brooks's law?
Brooks's law is a rule of thumb, not an equation. Yet Brooks attaches a coordination formula to it: for n people, there are n(n-1)/2 possible communication channels. That number grows faster than headcount, which is why communication ends up taking precedence over the work.
Does Brooks's law still apply with agile and AI?
Yes, in part. Agile limits the damage with small teams and short cycles, and AI takes over part of the coding. Neither removes training, coordination or indivisible tasks. The law remains a good reflex before strengthening a team.
What should you do when an IT project is late?
Start by checking that the delay is real and that the original schedule was realistic. Then revisit the scope or the date rather than the headcount. If extra help is unavoidable, choose someone who already has experience with the project's technology and give them a well-defined task, such as testing or documentation, to limit the training and communication involved.


