Member-only story
Is a One-Person Development Team Optimal?
Research, anecdotal, and personal experience claims teams of 3–8 are the best — but here is what I found
The ideal number of programmers on a project is nearly always one. — liberal interpretation of (Fred) Brook’s Law
I am not sure if he really said that, but he did say:
“If there are n workers on a project, there are (n²-n)/2 interfaces across which there may be communication, and there are potentially almost 2^n teams within which coordination must occur.”
Visually the communication problem is like below (Metcalf’s Law):
You spend/waste a lot of time communicating as teams grow — some say as much as 40% of your day. If you’re alone, you may waste time in other ways but you have a lot more time to code. Does it mean one is the optimal size? More on this later.
Research Says < 9
In a paper analyzing the International Software Benchmarking Standards Group (ISBSG) repository of software projects, researchers summarized three factors impacting software productivity:
- Programming Language
- Development Platform
- Team Size
The data showed a mean team size of 8 and median of 5 from a whopping range of 1–40. Their finding, specific to team size, is quite general despite the paper’s name, in summary: teams of < 9 are clearly more productive. It further breaks down productivity by language and platforms, and cites other observations such as new projects being significantly less productive than enhancement work to existing systems.
Miller’s Magic 7±2, Amdahl’s Law, etc.
Most Scrum and Agile guides justify numbers like 7±2 citing Miller’s famous research that this number is the typical limit of human cognition (later research suggests four chunks in terms of short-term memory capacity). This is interesting, but it…
