☰
02: Engineering Failure - Hindsight is 20/20. What About Foresight?
What engineering failures can teach us about anticipating problems
10 September 2026
How did I miss that?
If you are an engineer, you have probably been there. You designed a mechanism, specified a process or wrote some code. You followed the appropriate design, review and testing procedures — and yet something failed. Then comes the investigation and, eventually, the frustrating eureka moment. The failure makes perfect sense. Worse still, in retrospect it can seem almost obvious. Hindsight may be 20/20, but perhaps with sufficient discipline we can make our engineering foresight a little sharper. So how can we become better at seeing failures before they happen?
Why Good Engineering Still Fails
Failures can arise for many different reasons.
One common problem is that our model does not adequately represent reality. Mathematical models, FEA and CFD are enormously powerful, but a convincing result is not necessarily a correct one. Boundary conditions, assumptions, material behaviour, physical models and interactions all have to represent the real system sufficiently well.
Then there are assumptions. Engineering would be impossible without them; we simplify reality to make problems solvable. The danger arises when assumptions become inherited truths rather than propositions that still need to be challenged.
Precedent can create similar problems. Engineers rarely start from zero. We inherit successful designs, standards, software, test procedures and rules of thumb. But something changes — a material, supplier, lubricant, duty cycle, operating temperature or manufacturing process. Each change may appear harmless, while collectively the evidence that established the original design as safe may no longer apply.
Failures can also arise through unexpected combinations. A component may be safe. A subsystem may be safe. A procedure may be reasonable. Yet their interaction creates a failure mode nobody considered.
And perhaps the simplest problem is that we naturally think about normal operation. Yet failures often occur at the edges: startup, shutdown, maintenance, installation, unusual environments, accumulated tolerances, degradation or an operator doing something unexpected.
Takoma Narrows Bridge (1940) - Physical behaviour not adequately captured by contemporary design approaches.
Foresight Review
There are many formal methodologies for interrogating engineering designs, and they are valuable. But ultimately foresight also depends upon the discipline of the engineer. Before finalising an important design, deliberately step outside the normal design process. Instead of repeatedly asking: “Will this work?” ask: “How could this fail?”
Then keep going:
*Under what assumptions does it work?
*How could those assumptions be wrong?
*What happens when several adverse conditions occur simultaneously?
*What evidence would convince me that my model is wrong?
*If this fails in six months, what will I wish I had checked today?
In other words, earnestly try to prove yourself wrong.
Thought experiments are not confined to theoretical physics. Mentally operate the design outside its intended conditions. Start it. Stop it. Abuse it. Age it. Change the environment. Remove power. Combine unlikely events. Ask what the user might do that you never intended.
And be humble enough to invite others to challenge the design without being defensive — particularly people who have not travelled the same intellectual path as the project team. They may question something everyone else has unconsciously accepted.
Hyatt Regency Walkway Collapse (1981) - A seemingly straightforward design change altered a critical load path.
From Hindsight to Foresight
We will never achieve 20/20 foresight. Engineering involves uncertainty, and some failures will always surprise us. But hindsight has one useful characteristic: it reveals the questions we wish we had asked earlier. Perhaps the trick is to ask some of those questions before we have the benefit of hindsight.
ffdfdfd
dfdff
02: Engineering Failure - Hindsight is 20/20. What About Foresight?
What engineering failures can teach us about anticipating problems
10 September 2026
How did I miss that?
If you are an engineer, you have probably been there. You designed a mechanism, specified a process or wrote some code. You followed the appropriate design, review and testing procedures — and yet something failed. Then comes the investigation and, eventually, the frustrating eureka moment. The failure makes perfect sense. Worse still, in retrospect it can seem almost obvious. Hindsight may be 20/20, but perhaps with sufficient discipline we can make our engineering foresight a little sharper. So how can we become better at seeing failures before they happen?
Why Good Engineering Still Fails
Failures can arise for many different reasons.
One common problem is that our model does not adequately represent reality. Mathematical models, FEA and CFD are enormously powerful, but a convincing result is not necessarily a correct one. Boundary conditions, assumptions, material behaviour, physical models and interactions all have to represent the real system sufficiently well.
Then there are assumptions. Engineering would be impossible without them; we simplify reality to make problems solvable. The danger arises when assumptions become inherited truths rather than propositions that still need to be challenged.
Precedent can create similar problems. Engineers rarely start from zero. We inherit successful designs, standards, software, test procedures and rules of thumb. But something changes — a material, supplier, lubricant, duty cycle, operating temperature or manufacturing process. Each change may appear harmless, while collectively the evidence that established the original design as safe may no longer apply.
Failures can also arise through unexpected combinations. A component may be safe. A subsystem may be safe. A procedure may be reasonable. Yet their interaction creates a failure mode nobody considered.
And perhaps the simplest problem is that we naturally think about normal operation. Yet failures often occur at the edges: startup, shutdown, maintenance, installation, unusual environments, accumulated tolerances, degradation or an operator doing something unexpected.
Takoma Narrows Bridge (1940) - Physical behaviour not adequately captured by contemporary design approaches.
Foresight Review
There are many formal methodologies for interrogating engineering designs, and they are valuable. But ultimately foresight also depends upon the discipline of the engineer. Before finalising an important design, deliberately step outside the normal design process. Instead of repeatedly asking: “Will this work?” ask: “How could this fail?”
Then keep going:
*Under what assumptions does it work?
*How could those assumptions be wrong?
*What happens when several adverse conditions occur simultaneously?
*What evidence would convince me that my model is wrong?
*If this fails in six months, what will I wish I had checked today?
In other words, earnestly try to prove yourself wrong.
Thought experiments are not confined to theoretical physics. Mentally operate the design outside its intended conditions. Start it. Stop it. Abuse it. Age it. Change the environment. Remove power. Combine unlikely events. Ask what the user might do that you never intended.
And be humble enough to invite others to challenge the design without being defensive — particularly people who have not travelled the same intellectual path as the project team. They may question something everyone else has unconsciously accepted.
Hyatt Regency Walkway Collapse (1981) - A seemingly straightforward design change altered a critical load path.
From Hindsight to Foresight
We will never achieve 20/20 foresight. Engineering involves uncertainty, and some failures will always surprise us. But hindsight has one useful characteristic: it reveals the questions we wish we had asked earlier. Perhaps the trick is to ask some of those questions before we have the benefit of hindsight.
grgrgrgr