Trust as a Fault Line

When stress rises, the first casualty is trust.

That's where we left the last piece in this series, where we talked about stress as a finite resource, burned unevenly until even the most capable minds start to misfire.

This is where the cracks in trust begin to appear.

We like to imagine trust as something we either have or don’t, as if it were a light switch. But in reality, it’s more like a spectrum… elastic, dynamic, and constantly shifting with context. On one end of the spectrum lies blind trust, where we hand over control and stop thinking critically altogether. On the other, utter frozen distrust, where no one moves until every little variable is known.

Neither extreme is useful.

Between them lies the hard work of calibration… of deciding, moment to moment, how much confidence to give to another human or machine.

Most discussions of "trust in automation" focus almost entirely on reliability. Can the thing do what it says it can? But reliability is only the outer layer.

The deeper question is: how do humans decide when to believe in it? That piece of judgment is far from rational. It's not a spreadsheet calculation that we can chuck into Excel to do the hard work for us; it's a story we tell ourselves about predictability and control.

Trust, in other words, is risk management disguised as emotion.

We weigh the cost of being wrong against the comfort of being sure. Just like in aviation, pilots develop trust in an iterative way by repetition and feedback loops; the instrument which never lies, until one day it does. In medicine, robotic assistants have now gained credibility through precision.

Still, we see time and time again, even in these contexts, surgeons instinctively double-check. The same pattern plays out across industries: trust grows with experience, breaks with failure, and never quite returns to neutral.

Psychologists call this trust asymmetry.

It builds slowly, collapses fast, and behaves differently for everyone. One person’s acceptable risk is another’s unacceptable gamble. A study in Human Factors Journal found that pilots given identical flight data made starkly different choices about when to engage or disengage automation.

Not because of skill, but because of prior exposure and cultural conditioning. We’ve heard the stories in Malcolm Gladwell’s books and countless other retellings.

That difference matters.

It means that "trust calibration" is not only individual; it's social, it's cultural, even national.

When two people - or two partner nations - approach technology with different baseline assumptions about control, the friction isn't technical. It's philosophical.

Imagine this in a team, or working across allied national boundaries, or different organisations.

One partner delegates quickly to a system that has proven itself reliable. Another keeps manual oversight, wary of giving up too much agency. Both are rational. Both are doing what their training and culture have taught them to do.

Yet their differences create invisible turbulence (to throw back to a little aviation language). Trust stops being a bridge and starts becoming a boundary.

This is where that trust spectrum becomes a practical tool. It lets us see those differences not as flaws, but as variables to manage.

High-trust modes, where delegation is fast and confidence runs high, work best in environments where speed matters more than certainty. Low-trust modes, where verification is constant, are vital when the cost of error is catastrophic.

Neither mode is wrong. The danger lies in not knowing which mode you're in, or when to shift.

Teams that understand this can move fluidly along the spectrum.

In steady-state operations, they operate in a low-active-trust mode: checks, balances, quiet confidence.

Under pressure, they slide into high-implicit-trust mode: swift delegation, intuitive coordination, minimal questioning.

When the moment passes, they step back down the curve. Trust, like a muscle, flexes and releases.

Across boundaries, this kind of flexibility could be transformative. Some cultures institutionalise high-trust delegation - faster decisions, higher risk tolerance.

Others prize deliberate consensus - slower tempo, lower volatility. Instead of trying to erase those differences, groups could design for them. The cautious partner catches what the confident one misses; the confident partner acts when hesitation would hurt.

This is what it means to leverage trust asymmetry. Not to standardise it, but to understand its rhythm.

Because low trust isn't always failure-it's sometimes discipline.

It forces transparency and documentation. It keeps the humans in the loop. It ensures that decisions get questioned before they cascade. The goal isn't maximal trust, it's appropriate trust.

The right level, at the right time, for the right context.

If we start seeing trust as a capability, measurable, trainable, adjustable, then it becomes something we can design for.

Teams could build “trust dashboards” the same way they model workload or cognitive load.

Exercises could measure not just reaction time, but the elasticity of trust under strain.

Ultimately, trust is less about connection than about perception. It’s not just a fault line between humans and machines,  it’s a lens through which we see both. And that lens distorts easily.

Once we decide who or what is trustworthy, we begin collecting evidence to defend that decision. We filter what aligns with our belief and dismiss what doesn’t. That’s confirmation bias in action, the first quiet echo of what comes next.

Because when trust wobbles, bias rushes in to fill the space. Bias in how we read data. Bias in how we interpret others’ intent. Bias in how we forgive a machine’s mistake but not a human’s. That’s the next fracture to map… the bias in the loop.

The ghosts in our judgment, which technology doesn’t erase but amplifies.