Quantification is a genuine achievement. A well-run risk function can put a credible number on a risk's probability and severity, with a defensible range around it. The danger is not that the number is wrong. It is that a precise number invites you to stop there — to treat the figure as the whole of what is known.
Position is not in the number
A risk that scores as moderate in individual assessment might sit where four pathways converge — several streams of consequence flowing into one point, each assessed in isolation, none of them showing the convergence. A risk that scores as critical might function in isolation. The score measures size. It is silent on position. And in a connected system, position is often what decides whether a moderate risk becomes the one that matters.
The threshold nobody wrote down
Every risk report has a cut-off somewhere, whether or not it was ever stated. Call it five percent of the original shock: a system that absorbs more is “significantly affected” and gets reported; a system that absorbs less is reached but stays silent. That single number does an enormous amount of work — it is the line between a finding and a footnote. The systems sitting between one and five percent felt the failure and passed it on, and by the rule none of them mattered enough to mention.
When two percent is not noise
A two percent effect on a regulatory reporting system is nothing on an ordinary Tuesday — comfortably inside the range of things you would never escalate or log. Put the same two percent on the same system during an unrelated incident that is already consuming the team, and the margin that would have absorbed it is gone. The effect the threshold told you to ignore becomes the one that tips the system over. The importance of a second-order effect is not fixed; it depends on what else is happening at the time — and a threshold can only ever judge it in isolation.
So what?
Keep the numbers. They are real information, and the network does not replace them. But hold them with the knowledge that precision and significance are not the same thing, and that a cut-off applied in silence is a judgement, not a measurement. The point is to choose what you ignore, rather than to let one number choose for you.