Every mine runs plenty of safety checks: pre-start cards, permits, inspections, critical control checks, monitoring systems. Most of them are done on time. Their limit is built in. A check looks for what its author expected to find, at the place and time they expected to look.
Meanwhile a crew walks past hundreds of things on a shift. A few of them end up on a permit or a take-5 card. Most stay in someone’s head, and the rest sit in separate systems that nobody reads side by side. That gap, between what a site checks and what it notices, is what these notes are about.
Take a handrail on a mill walkway. Watch the footprints people leave: grey for the scheduled check, yellow for the traces everyday use leaves, green for the checks people do before they rely on something. Then watch the next crew follow them.
Three kinds of looking touched that rail. The scheduled check did its job: it tested a handrail for leaning. Everyday use, from the leans to the drill to the photo on the board, showed the rail carrying a load nobody had planned for. And one hard check, an anchor proven in person, settled it. None of the three would have been enough on its own, and the next crew did not have to ask again: they followed the footprints.
A list sees what we expected
A list is a good thing. It makes sure the known hazards get looked at, and the law requires it. But it looks only at what someone wrote down, only when the schedule says. Between visits and outside the list, nobody is looking. Over the years the checks keep happening while the reasons behind them fade, and adding lines makes a list longer, not wider. The evidence points the same way: the injury rate barely tracks what kills [1], most fatal control failures happen in doing the control rather than designing it [2], and a ticked box is often not a done one [3].
Everyday use keeps looking
Every use of a place leaves a footprint: a door swipe, a permit, a work order, a radio call, a photo on the board, a remark at the top of the stairs. Nobody has to log anything extra. Put those footprints together by place, and they show what no list can:
Something used in a way nobody planned for. A rescue rope on a handrail.
Fewer visits to an isolation point, while work on that conveyor keeps growing.
A place in daily use that nobody has checked in a long time.
Catching these is the clearest sign the system is working. Most will turn out to be nothing, and finding out costs a short walk.
On a real site
The idea is not another app. It is to bring these footprints into one place, so that every object and every activity keeps a persistent memory: how it is used, what was checked, what was found, and why. The handrail would remember the rope. The next crew would not have to rediscover it, and the knowledge stays when people move on. When you work in someone else’s area, that memory tells you what to watch for, and what you notice goes back into it. Sensors and cameras leave footprints too.
We are building this system. A first version is in our demo portal →
Picture the start-of-shift line-up. The TV in the crib room shows what the site’s memory picked up last week, next to a short screen on whether its systems are working: microseismic, ventilation, blast clearance, gas. Everyone sees the same record, and anyone can ask about the line that looks wrong.
Badges, tagging people and a word of thanks are only tools to keep people adding to that memory. Better teams report more, not less [4], so reward what was found, never the number of posts or ticks.
What comes next
Visibility and peer-to-peer noticing matter, but they are not enough on their own. People drift toward easy activity, a chat or a like, and away from the hard checks: proving zero energy, testing the isolation, rating the anchor. How many hard checks an area needs, and how often, should be set by physics and statistics, from the energy stored there and how much it is used, not by what feels busy. That is the subject of our next post.
For the wider case for keeping the reason with every record, see our CIM Magazine article, Data rich but knowledge poor. If you are working on the same questions at your site, book a call.
References
- Hallowell, M., Quashne, M., Salas, R., Jones, M., MacLean, B. and Quinn, E. (2021). The statistical invalidity of TRIR as a measure of safety performance. Professional Safety 66(4): 28–34.
- ICMM (2025). Safety performance: benchmarking progress of ICMM company members in 2024.
- Recorded against observed checklist completion: Giles et al. 2017; Flinders QI audit; Ballarat 2013; video study, Surgical Endoscopy 2025.
- Edmondson, A.C. (1996). Learning from mistakes is easier said than done. Journal of Applied Behavioral Science 32(1): 5–28.