Thread by @jessitron

Some tweets brought to you from a business magazine that someone left on the airplane.

There’s an interview about why out systems fail, “the Paradox of Progress”

András Tilcsik says there are two properties of modern systems that lead to surprising + high-consequence failures.

One is tight coupling, or lack of slack. “If something goes wrong, you won’t have much time to figure out what’s happening and make adjustments.”

Another is complexity: the system is a web, not a line. “Much of what goes on in these systems is invisible to the naked eye.”

Parts “interact in hidden and unexpected ways.”

He ties complexity closely to out-of-view.
Old dams, people could go see what the gate was doing. Now operators in control rooms are “removed from having direct access to the system.”

All software has this quality, because it runs in digital space.

Software is “much harder to peek into and much more tightly coupled” than mechanical systems.
It operates at out of view and at faster time scales than humans naturally work.

And yet, we can make it observable!
If we try super hard.

Lessons to learn from aviation:
“A simple punitive approach is an impediment to learning.”

Separate work on safety from rule enforcement. Ya gotta “consider how the rules themselves contribute to accidents.”

It doesn’t make sense to stop building complex, tightly-coupled systems.
But keep an eye on it. Maybe sacrifice some efficiency to get some slack, slow down the rollout to loosen coupling while you learn.

“Navigating a simple system, diversity might actually be a net cost.” Coordination is more efficient with sameness.

But in complexity, “diversity helps because it makes everyone on a team slightly less comfortable - and as a result, more skeptical.”
#disfluency

“Homogeneity has the advantage of reducing friction and speeding up decision making
- but often at the cost of creative solutions.”

Jeanne Liedtka et al, in the next article.

Modern innovation starts with “asking some very different questions.”

“The definition of the problem is a hypothesis to be tested.”