Circle of Competence
Know the boundary of what you genuinely understand, and act differently inside and outside it.
“The size of that circle is not very important; knowing its boundaries, however, is vital.”
To an executive
Confident advice from people outside their expertise is a hidden risk. Knowing who is expert and who is guessing lets you spend scrutiny where it matters.
To an engineer
Mark which of your design views are expertise and which are guesses. At the edge, bring in a specialist, run a small trial or state your uncertainty openly.
The idea
Warren Buffett used the phrase in his 1996 letter to Berkshire Hathaway shareholders: an investor only needs to be able to evaluate the businesses inside their circle of competence. "The size of that circle is not very important," he wrote; "knowing its boundaries, however, is vital." Charlie Munger made the same point often.
Competence, in other words, is less about how big your circle is than about knowing exactly where its edge lies. Inside it, you can decide with confidence. Outside it, confidence is dangerous.
Architects are pushed to have opinions on security, data, networking, finance, law and more. The credible ones are explicit about which views are expertise and which are informed guesses, and they change how they work at the edge.
Why it works
Errors cluster outside the circle. Most expensive mistakes are not made in an architect's speciality; they are made in an adjacent area where they felt competent enough not to check.
Labelled confidence helps others weigh your advice. When you say "this is my field" or "this is a guess", stakeholders know when to rely on you and when to seek a second opinion.
A known edge can be grown on purpose. If you can name what you do not know, you can plan to learn it, rather than discovering the gap in production.
Where it applies
- Specialist domains such as cryptography, identity, low-latency trading, safety-critical systems and regulatory compliance.
- Vendor selection, where a confident demo can pull you well outside what you can judge.
- Design reviews that span many disciplines, where no single reviewer covers the whole design.
- Stakeholder conversations, where you are asked for a firm view on something you only partly understand.
How to apply
- Map the decision to areas of expertise and mark which ones you own, which you know partly and which you do not know.
- Bring in a specialist for each area outside your circle, and give them the authority to block, not just to comment.
- Label assumptions by confidence in architecture decision records, so later readers know which parts were guesses.
- Grow the circle deliberately through spikes, pairing with specialists and small, reversible experiments.
Example scenario
An enterprise architect with deep experience of integration patterns signed off on a home-grown token scheme for a new partner API. It looked like a small piece of plumbing. A security specialist brought in later found it open to replay attacks, and the scheme had to be replaced after partners had built against it.
The team changed its review checklist: any design that touched authentication, authorisation or cryptography now needed sign-off from the security architect before approval. The next design used a standard OpenID Connect flow, which had been available from the start.
Questions to ask
- Which parts of this decision sit inside my real expertise, and which are informed guesses?
- Who knows this area better than anyone in the room, and have they reviewed it?
- What would I need to learn, or who would I need to involve, to be confident here?
When not to use it
- Small, reversible decisions, where the cost of getting a specialist exceeds the cost of being wrong and fixing it.
- Genuinely new territory, where nobody has a circle that covers it yet; there you need careful experiments, not a hunt for an expert.
How it misleads you
The circle can become an excuse never to stretch. Architects have to work at the edge of their competence much of the time; the point is to do it with humility, specialist support and checks, not to stay safely in the middle. Also beware the opposite trap: deferring to anyone who claims expertise without asking how they know.
Further reading
Last reviewed · Suggest an improvement