Emerging risk is hard precisely because it has not happened yet. There is no clean history to extrapolate from, and the thing you most need to spot is the thing your register has not thought to name.
Horizon-scanning is not enough
Scanning the environment produces a longer list of things that might matter. It rarely tells you which weak signal deserves attention. Without a sense of where a new disruption would actually land, every signal looks equally plausible and equally ignorable.
Structure points to where it would hurt
A network view changes the question. It shows the junctions, the tightly-coupled clusters and the points with no alternative path — the places where a new, unfamiliar disruption would do the most damage if it arrived. That is where scarce attention should go first.
From speculation to grounded judgement
Read this way, “what could go wrong?” stops being open-ended speculation and becomes structurally grounded. The network shows where the system is exposed; expert judgement decides what, specifically, might exploit that exposure. The two together find things a scan alone would miss.