Safety scientist James Reason saw near misses as \u201cimmunizations\u201d<\/a> for a safety system, chances to detect and fix underlying vulnerabilities before real harm occurs. But too often, we waste these chances. We get lucky, and instead of investigating or analyzing what went wrong, we move on.<\/p>\n My interest in near-misses comes from practising medicine and from my research into the history of disasters and system failures, work that informed my book Written in Blood<\/a>. Studying accidents across fields, from fires to transportation to health care, shows that warning signs are often visible long before catastrophe strikes.<\/p>\n Apple\u2019s iOS 26.1 software update patched multiple critical vulnerabilities that could have allowed attackers to seize control of iPhones. Luck is not a strategy<\/p>\n Take something as mundane as your phone. In late 2025, Apple released iOS 26.1, a routine software update. Except it wasn\u2019t routine. It patched multiple critical vulnerabilities<\/a> that could have allowed attackers to seize control of iPhones. Had hackers succeeded, millions of users\u2019 data and privacy could have been compromised. And while some phones probably had been hacked, for most people, the crisis was avoided.<\/p>\n In health care, near-misses are common: A medication nearly given to the wrong patient but caught in time, or a surgical tool counted incorrectly but found before the patient\u2019s incision is closed. These are serious signals, but too often they go unreported. The majority of health-care workers fail to report near misses<\/a> due to fear of blame, lack of feedback or the false belief that no harm means no problem.<\/p>\n Often, staff in health care don\u2019t even realize a near-miss has occurred. If we\u2019re not looking for near-misses, we are nearly guaranteed not to learn from them<\/a>. <\/p>\n A recent Canadian Automobile Association study found that at just 20 monitored intersections, more than 610,000 \u2018near-miss\u2019 incidents were recorded from September 2024 to February 2025. Transportation shows the same pattern<\/a>. Near-collisions on icy highways. Trains braking just before overshooting a signal. Aircraft diverting after onboard systems detect a mechanical fault mid-flight. In aviation and rail, these close calls are treated as data. In many other sectors, they are dismissed as background noise. But the data is there. <\/p>\n A recent Canadian Automobile Association (CAA) study<\/a> found that at just 20 monitored intersections, more than 610,000 \u201cnear-miss\u201d incidents \u2014 close calls between vehicles and pedestrians or cyclists<\/a> \u2014 were recorded from September 2024 to February 2025. <\/p>\n Our systems are sending signals. Every time we get lucky is a chance to learn \u2014 a chance to build better layers of defence; a chance to prevent the next tragedy. Near-misses aren\u2019t false alarms. They\u2019re the most honest feedback a system gives: The future, whispering in the present.<\/p>\n Our brains aren\u2019t wired for prevention<\/p>\n So why don\u2019t we learn from close calls?<\/p>\n Psychologists have long understood the human brain is terrible at processing invisible risks<\/a>. We overreact to dramatic events but underreact to near-misses. We confuse luck with safety. And we discount what \u201calmost\u201d happened.<\/p>\n Three psychological traps are especially pernicious:<\/p>\n Availability bias<\/a>: We remember big disasters, but not the hundreds of times catastrophe was narrowly averted. This skews our risk radar. Reason\u2019s \u201cSwiss cheese\u201d model<\/a> describes how disasters happen when weaknesses in multiple layers of defence align. A near miss is when they almost line up and something, often by chance, blocks the path. But unless we plug those holes, the next time, we might not be so lucky. <\/p>\n Aviation, nuclear energy and air traffic control are so-called \u2018high-reliability organizations\u2019 that treat close calls as data points. There are exceptions. Aviation, nuclear energy and air traffic control, so-called \u201chigh-reliability organizations<\/a>,\u201d understand this. Ideally, they treat every close call as a data point. They institutionalize reporting. They never forget to be afraid. <\/p>\n These organizations cultivate a chronic unease, a kind of productive paranoia. It\u2019s not pessimism; it\u2019s realism. They know that systems often drift toward failure unless they\u2019re constantly corrected. That mindset is why they\u2019re among the safest sectors in the world. <\/p>\n![\"a](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/03\/file-20260303-57-mrt0nu.jpg\")
<\/p>\n
\n (Unsplash+\/Getty Images)<\/p>\n![\"Blurred](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/03\/file-20260303-57-wygi15.jpg\")
<\/p>\n
\n THE CANADIAN PRESS\/John Woods<\/p>\n
\n Confirmation bias<\/a>: We assume a system is safe because it didn\u2019t fail. But many systems survive not because they\u2019re strong, but because nothing has lined up to break them \u2014 yet.
\n Optimism bias<\/a>: We know bad things happen to other people but assume our skill or luck will protect us.<\/p>\n![\"An](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/03\/file-20260303-57-ne5p88.jpg\")
<\/p>\n
\n THE CANADIAN PRESS\/Darryl Dyck<\/p>\n