Eight Black Swans in Eight Years

At some point, a black swan stops being a black swan.

The term comes from Nassim Taleb’s work on rare, high-impact events that fall entirely outside the range of normal expectations—things so statistically improbable that planning for them feels like planning for nothing in particular. The rarity is the point. Remove the rarity, and you just have swans.

Eight consecutive years. Eight material disruptions to travel demand that no commercial forecast anticipated:

Eight years. Eight times the toolkit failed to keep pace with the environment.

I am not sure we get to keep calling these black swans. At some point, volatility is just the baseline—and the commercial infrastructure most airlines are operating was not built for the baseline we are actually in.

What Normal Already Costs

telos has written in previous posts about two structural problems that exist in airline commercial operations before any disruption arrives.

The first is that the feedback loop is open. Decisions get made, actions get taken, and then nothing—no structured record of what was done, no tracking of what happened next, no calibration signal feeding back into the system or the analyst. The loop stops rather than closes, and a new one starts on the next queue item with no memory of the last.

The second is that the orientation phase of the decision cycle is already compromised. Assembling the context required to make a good decision—across booking data, competitive intelligence, historical patterns, and RMS outputs living in separate systems—takes more time than the volume of the queue allows. Analysts are deciding from an incomplete vantage point not because they are careless, but because full orientation is not operationally feasible at the required pace.

These are the conditions on a calm Tuesday in February with no external disruption and a booking curve behaving more or less as expected.

Now add a demand shock.

What Disruption Actually Does to a Broken System

A demand shock does not create new structural problems in commercial operations. It amplifies the ones that already exist, at a speed that makes them impossible to manage around.

The orientation problem does not get incrementally worse under disruption—it gets worse by an order of magnitude. The context you need to make a good decision is now changing faster than you can pull it together. The competitive landscape you oriented to an hour ago is no longer accurate. The historical patterns you would normally reference aren’t comparable to a situation with no real precedent in your data. The RMS is recalculating on inputs that are themselves in flux. And the queue, which was already longer than the team could fully cover, has just doubled.

The feedback loop problem gets worse differently. You are now making more decisions, faster, under more pressure, with less context—and none of those decisions are being tracked in a way that will inform how you respond next time. Every disruption is, in effect, the first one the system has ever seen. The eighth consecutive year of disruption finds the commercial infrastructure in essentially the same position as the first one: observing without memory, deciding without calibration, acting without a closed loop.

That is not a technology failure. It is an infrastructure design problem. And it is expensive in ways that most commercial operations do not calculate, because calculating it would require the decision quality infrastructure that does not exist.

The Cost Nobody Is Counting

Every disruption produces a set of commercial responses: inventory tightened in markets that subsequently went soft, fares filed too aggressively or too conservatively given how demand actually shifted, capacity left unprotected in markets where the disruption drove unexpected demand spikes. Some of those responses were correct. Some compounded the problem. Without a closed loop, you cannot tell which was which.

What most commercial operations do instead is look at total revenue performance against the disrupted period and draw broad conclusions about how the team managed it. That aggregate number averages together the good calls, the bad ones, and the many decisions that were made in markets nobody had time to review at all. It is not a measure of decision quality. It is a measure of aggregate outcome, and those are not the same thing.

The airlines building genuine resilience are not just investing in better disruption forecasting, though that matters. They are investing in the infrastructure that tells them, after each event, what they did, what they should have done, and where the gap was largest. That is the learning that compounds across disruptions. That is what makes the ninth event less costly than the eighth.

This Is the Baseline Now

I think the airline industry is still, in some corner of its planning consciousness, waiting for conditions to stabilize. For the environment to return to something more like 2015—predictable seasonality, manageable competitive dynamics, demand curves that behave the way the models expect.

I don’t think that environment is coming back. The 2026 US-Israel-Iran War has made that clearer than ever. Jet fuel doubled in a matter of weeks following the closure of the Strait of Hormuz. Airlines that had exited fuel hedging programs in 2024 and 2025 found themselves fully exposed at the worst possible moment. More than 20,000 flights were grounded in the conflict’s opening days. The Europe-Asia and North America-Asia corridors that run through Middle Eastern airspace are still being rerouted months later. Commercial teams that had just stabilized after the tariff-driven demand collapse of 2025 were immediately asked to rebuild their orientation from scratch—against a backdrop with no comparable precedent in their data.

The last eight years have not been a streak of bad luck. They have been a demonstration that the range of possible operating conditions is wider than the toolkit was designed for, and that the width of that range is not shrinking.

Building commercial decision infrastructure for that environment is not a response to volatility. It is a response to the recognition that volatility is structural—that the planning environment has permanently changed and the tools need to change with it. The OODA loop described in prior posts was designed for exactly this kind of high-tempo, high-uncertainty environment. Boyd’s insight was that the competitor who can cycle through Observe-Orient-Decide-Act faster than the adversary does not just win individual engagements—they compound that advantage over time in a way the slower operator cannot replicate.

Eight years in, the question is not whether the next disruption will test the commercial team’s infrastructure. It will.

The question is whether the infrastructure will be different when it does.

Author’s note: telos builds decision infrastructure for airline commercial teams. The closed loop we keep writing about is not a product feature—it’s the reason we started the company.