Why Expertise Has a Shelf Life

There is a physicist called Samuel Arbesman who has written about what he calls the half-life of facts — the rate at which established knowledge in a given field becomes outdated or superseded. In clinical medicine, this half-life is somewhere around five years. In mathematics it is much longer. In certain applied technology fields it is shorter than three.

The number is less important than the principle. What you know has a use-by date, and in most fields that date arrives faster than most people expect.

The competence trap

The pattern plays out in organisations in a specific way. Someone becomes genuinely expert in a domain over a decade of work. They are promoted into a leadership role that relies on that expertise — their judgment, their ability to evaluate what others bring them, their capacity to ask the right questions. And then, gradually, the domain shifts under them while they are busy managing, and they do not notice until the gap is already significant.

The problem is not laziness. It is that maintenance learning — keeping current with a field you already know — requires active effort and competes for time with everything else that comes with seniority. The senior person who stops attending conferences, stops reading primary sources, stops spending time with people who are learning the current state of a field — that person’s expertise is depreciating in real time, invisibly.

The worst version of this is when their confidence does not depreciate with it. A twenty-year expert who stopped actively learning five years ago often feels more confident in their knowledge than someone who is more current, because they have decades of successful pattern-matching to draw on. The patterns were real. They may no longer apply.

This is especially visible in technology

The half-life problem is acute in IT and cybersecurity. The threat landscape, the tools, the attack surfaces, the regulatory environment — all of it moves faster than most professional development programs track. A CISO who built their career on perimeter security thinking needs to have genuinely updated their mental model for the way modern attacks work, or their strategic advice will be wrong in ways that are hard to diagnose from inside the organisation.

This is one reason the fractional advisory model has value. An advisor who works across multiple organisations in a domain has more exposure to current patterns than someone inside a single organisation, because the problems come to them from more directions. Current expertise is easier to maintain when the work requires staying current.

What active maintenance looks like

Reading the current literature in a field is the obvious answer and not wrong, but it is insufficient on its own. Connecting regularly with people who are doing the primary work — practitioners, not commentators — is more valuable. Being willing to have your existing assumptions challenged, including by people with less seniority, is essential.

The hardest part is the willingness to find out that something you have believed confidently for years is no longer quite right. That experience is uncomfortable and it is how expertise stays alive.

The alternative is confident obsolescence, which is a worse problem than acknowledged uncertainty. An expert who says "I’m not sure how current my knowledge is here, let me check" is more useful than one who answers with authority from a model that expired several years ago.

IdeaSpace exists partly to think out loud about problems like this — the structural features of how organisations and individuals get things wrong, not from bad intentions but from the normal operation of human cognition and institutional inertia. There is more in the archive if this is the kind of thinking you find useful.

The Cost of Confident Obsolescence

The most dangerous expert is not the one who knows they are out of date. It is the one who does not know. Confidence that has not been recalibrated against current reality produces bad decisions that are delivered with authority, which means they are harder to challenge. A junior team member who suspects the approach is wrong will hesitate to push back against a senior person who sounds certain. The wrong decision gets made with full conviction.

This is especially dangerous in fields where the rate of change is high and the consequences of being wrong are serious. Cybersecurity, regulatory compliance, and technology architecture all have this property. A security strategy based on a threat model from five years ago is not just outdated — it is actively dangerous, because it creates confidence in a defence that no longer works. The same pattern explains why digital transformation projects fail — the expertise that designed the old system is often the expertise that designs the new one, and if that expertise has not been updated, the new system will replicate the old assumptions.

How to Stay Current Without Burning Out

Staying current does not require reading everything. It requires a system. A small set of trusted sources that you read consistently. A habit of testing your assumptions against new information. A network of people who are working at the edge of the field and will tell you when your thinking is stale. The people who maintain their expertise over decades are not the ones who read the most. They are the ones who have built the most effective feedback loops.

The most underrated feedback loop is teaching. Explaining your field to people who are new to it forces you to articulate assumptions you had stopped examining. A question from someone who does not share your mental model will often reveal the gaps in it faster than reading the latest research. The expert who teaches stays current because teaching is a form of testing.

The Organisational Dimension

Organisations have a role in this too. If the only people whose expertise is valued are the ones who have been there the longest, the organisation is systematically overvaluing stale knowledge. The twenty-year veteran who stopped learning five years ago is given more weight than the five-year veteran who is current, because seniority is easier to measure than expertise. The organisation gets the decision quality it measures for.

The organisations that avoid this trap create space for challenge. They expect senior people to be challenged by junior ones. They treat a good question from a new hire as a sign of health, not a threat to authority. This is hard culture to build and easy one to lose, but it is the only reliable defence against the slow decay of institutional expertise. The same cultural muscle that allows a junior developer to question a CTO’s technical assumption is the one that allows a project manager to report bad news early — which we explored in the dashboard is green, the project is red.