The problem in one sentence
The physical systems that civilization depends on are largely unable to describe their own state.
This is easy to overlook because the systems work so well. Power arrives. Water arrives. Goods are made and moved. The reliability is real, and it is purchased with scheduled maintenance, redundancy, inspection rounds and the experience of people who can tell from a sound that something is wrong. Those methods have served well. They are also expensive, unevenly applied and increasingly difficult to staff.
What observable means
A system is observable when its internal condition can be inferred from what it emits. Software engineers adopted the word for the practice of instrumenting applications so thoroughly that their behavior can be understood from the outside. We use it in the same sense, for hardware.
An observable machine is one whose health is known, continuously, without anyone opening it. An observable facility is one whose operators can see which assets are sound and which are changing. An observable infrastructure is one that can be managed on evidence.
Three commitments
Observe continuously
An inspection is a snapshot. Degradation is a process. Continuous observation is the only way to see a slow change begin.
Understand earlier
Data is not the goal. The goal is an earlier and more accurate understanding of what a machine is doing, delivered to someone who can use it.
Act sooner
The measure of success is a decision made in time: a repair scheduled before a failure, a load shifted before a trip, a replacement ordered before it becomes an emergency.
Why it matters now
Several long-running trends have converged. Sensors have become smaller and less expensive. Processors capable of running useful models now fit in devices that can be mounted on a motor. Machine learning methods for finding patterns in physical signals have matured.
The need has grown at the same time. Much of the infrastructure in service today is aging. Facilities that support computing are being built at great speed and run close to their limits. Many of the engineers who understood older equipment by instinct are retiring. The case for giving machines their own senses has not been stronger.
How we will know we are succeeding
By whether the people who operate equipment trust what AFFERENT tells them. Trust of that kind is built slowly, through findings that prove correct and through honesty about the ones that do not. It is the standard we intend to be held to.
