A Journey Through the CAST Handbook: Learning From Tragedy

Previous: How Not to Cry Over Spilt Coffee

An accident where innocent people are killed is tragic, but not nearly as tragic as not learning from it. - Dr. Leveson

My own troubles in the kitchen are trivial compared to the accidents and mishaps resulting in fatalities or catastrophic losses that happen all the time, such as the oil rig fire on the cover of the CAST Handbook (Leveson 2019) or the 1984 B-1 accident during a flight test.

When such losses occur, it is natural to investigate, to try and figure out what happened. Accident investigations, in the United States at least, even have several agencies devoted to accomplishing them.

The handbook begins with the most natural of questions.

Why do we need a new accident analysis tool?

Accident investigations have a long and storied history, especially in the transportation industry, whether land, marine, or air, due to the significant losses in life or property that can occur. Investigation techniques have evolved and there are many to choose from. So why a new one?

Dr. Leveson states simply that we are not learning enough from the costly events surrounding an accident. Conclusions and recommendations are oversimplified and tend to focus on "operator error" or one or two "root causes," without exploring the systemic contributions to the accident.

Spilt Coffee

In the case of my coffee kerfuffle, a comment on my LinkedIn post sums up what a typical accident investigation would conclude

Operator error. Closed case.

The commenter is absolutely right. It was operator error. There was no malfunction or component failure within the machine. Everything worked according to design. It was my fault.

But have we learned everything about how we could avoid losses in the future? Is there more to the story than finding who or what to blame?

CAST, a systems approach, starts with the premise that operator, or human, error is a symptom of a system that needs to be redesigned.

Human error is a symptom, not a cause.

The operator error in my case was simply: "failed to remove old grounds."

The root cause was: "old grounds still in filter basket caused overflow."

But what to do about it? How can future accidents be prevented? How can the results of our analysis be shared with others?

We can start by moving away from a focus on blame, accusation, and identifying who failed or is at fault.

Blame is a legal or moral concept, not an engineering one

Instead, we can move towards a focus on learning, explanation, and identifying systemic flaws.

Reframing From Blame to Learning

So let's reframe from "operator failed to remove old grounds."

Starting with a description of what happened.

WHAT: The previous grounds, combined with the chute droppings and the fresh grounds, reduced the volume in the filter basket available for the water, resulting in the water and grinds overtopping the filter and the basket

WHY: Reasons for the previous grounds and chute droppings remaining in the filter basket

1. During chute cleaning, previous grounds were kept in the filter basket to catch the chute dropping
2. During chute cleaning, filter basket cover was removed to allow chute droppings to fall directly into filter basket and not clog the filter basket cover
3. During chute cleaning, filter door was closed to allow chute droppings to fall directly into filter basket
4. After chute cleaning, filter door remained closed, water and beans were added, and the "Go" button was pushed
5. After "Go" button was pushed, the "filter basket cover missing" alarm sounded, prompting the replacement of the filter basket cover, but not the removal of the old grounds, chute droppings, and used filter
6. After the filter door was closed again, the "Go" button was pushed again with no further alarms
7. Overflow continued for the duration of the brewing cycle

This reframing sets the stage for a few fundamental definitions, then we can dive into an actual CAST analysis...

Next: Welcome to Haz(z)ard County