Student pilots absorb a rich folklore: the instructor who says to always trust the checklist, the old hand who swears you can feel a stall coming, the friend who insists the most experienced person should handle any emergency. Some of it is gold and some of it is nonsense. The same split exists in the world of online operations, where IncidentOps, the practice of coordinating a response to unplanned service disruptions, has collected its own myths. If you are learning to fly and also dabble in running a club site or a booking tool, these six are worth testing against the evidence.
Myth 1: The most senior person should always take charge
It sounds right, and it is how many organisations behave by reflex. The better-documented rule is different. One openly published runbook gives the commander’s seat to the top-ranking participant who is physically on the bridge, regardless of the rank they hold back at their desk, and urges teams to pass the seat around among those trained for it. The reasoning is practical: the best coordinator is the one who has practised coordinating, not the one with the longest tenure.
Crew resource management in aviation made the same point decades ago. Rank does not stop a junior crew member from spotting the problem, and it does not guarantee that a senior one will be calm. In incident work, roles are described as hats rather than titles, which is a handy reminder for a busy volunteer club.
Myth 2: Calling an incident is an admission of failure
Many teams privately believe that a declared incident will be counted against them. The result is silence, followed by a bigger mess. The Ops Atlas, which sets this out in its piece on the IncidentOps discipline, argues that punishing the number of declarations suppresses them. On the cost side, a declaration that proves small takes minutes, while a late one can take hours. A safety culture rewards the early report of a near miss; this is the same principle.
Myth 3: The cause has to be known before anyone touches anything
A pilot who diverts for weather does not wait to understand the front. The operational equivalent is to take a generic, reversible step first. Roll back a recent change, drain traffic from a struggling server group, disable a new feature, fail over to a spare, or shed some load. All of these require knowing roughly where the fault lies, not why it appeared.
The Atlas retells a managed-Kubernetes outage in which new clusters could not be created for roughly 400 minutes. Forty-one people crowded the chat and seven task forces sprang up. The post-mortem later judged that simply reverting the suspect images, once the fault’s location was clear, would have restored service by mid-morning instead of around noon. Understanding the origin comes later, in the review.
Myth 4: If a person made the mistake, the cause is human error
Aviation insiders hear this in accident-report summaries all the time, and the better investigators treat it as the opening of the inquiry, never its conclusion.
This one deserves special scepticism, because aviation has already learned to distrust it. One safety researcher, cited in the Atlas, noted that hindsight colours the way we grade human performance after something goes wrong. Knowing the outcome makes every earlier choice look obvious. A review that ends at “the engineer made a typo” has stopped asking the productive question, which is what allowed the typo to cause an outage.
One public example the Atlas recounts is a 2017 outage in which a mistyped input to a routine playbook removed far more server capacity than intended. The better lesson was about safeguards: limits on how much capacity a single command can remove. If you want to see how a system-level mindset applies to learning itself, the site’s practical advice on trimming flight-training costs frames mistakes the same way: look at the structure of the plan, not only at the person.
Myth 5: A longer list of action items means a better review
Every instructor has left a debrief with a page of notes and forgotten most of them by the next lesson. Reviews suffer from the same overload.
It feels productive to leave a meeting with twenty tasks. Evidence from public review practices points the opposite way. Strong reviews end with a short list; every line has a single owner and a deadline, and the team tracks the proportion completed punctually. The Atlas warns about three failure patterns: actions that are mostly mitigations, actions so vague that nobody can tell when they are done, and actions treated as equally important when they are not.
It also flags a sneaky habit, closing an action by downgrading its priority instead of completing it. A big software firm goes further and sets a time limit on priority follow-ups (a month or two, depending on how critical the service is), so follow-through can actually be audited. Think of an airworthiness directive with a compliance date: without the date, it is just advice.
Myth 6: Everybody should join the call
This myth is the friendliest of the lot, which is why it survives. Nobody wants to be the person who left a colleague out, and a full room looks like diligence.
The instinct is to invite the whole company so nobody is left out. In practice, crowds slow decisions. Field research the Atlas references found that when hundreds of people join a single incident room, the core responders retreat to private channels, which splits the record and defeats the purpose. A related guideline from emergency management is one supervisor for roughly five people; beyond that, create sub-teams.
The fix is two rooms. A working channel holds the commander, the hands-on group, the communicator and the scribe. An updates channel holds everyone else. Observers get a steady flow of information without adding noise. This matches the way a busy airfield separates the tower frequency from the ATIS broadcast.
A bonus myth about the numbers
Picture a school that quotes the average time its students take to solo. A single outlier who needed a long layoff pulls the figure upward and hides what the typical student experienced. Incident data behaves the same way.
Many teams proudly report an average time to resolve incidents and assume that a falling number proves improvement. According to the Atlas the acronym stands for four or more different phases. Separately, a simulation built on real incident histories showed that a genuine ten percent gain could register as a decline in a sizeable fraction of runs. Outage lengths are skewed, so one long event can dominate. Reporting the spread of times for detecting, acknowledging, mitigating and resolving is a more honest approach, as is tracking how many follow-up actions land on time.
Putting the myths to work
If you are drawn to the operations side of aviation technology, a structured study path helps. The article on building a site reliability learning plan step by step pairs naturally with the habits above. For this week, pick one myth you have believed and replace it with a written rule. Start with the trigger for declaring, because everything else depends on someone saying out loud that the problem has begun.

