Somehow, I feel that there is much discussion about the safety of the FBW-systems but not from control theory point of view. From control theory point of view the functionality is more important that the means the functionality is implemented. The same functionality can often be implemented with computers or analog devices. As far as I have understand, computers are often more reliable.
With computers you can implement features that can not be implemented with analog devices. However, the basic questions remains the same. You have to solve at least stability, reliability, controlability and observability problems.
I have a feeling that that the pilot aids are not the core problem but the practices. I don't know if the sales men, who promote the new protections, or technical persons who don't reorganize the potential problems, should be blamed.
When you have deep enough understanding on the control theory, you should understand that every time a inequality constraint is changing the status, the risk level is increased. When there are too many inequality constraints in the system, it becomes practically impossible to analyze it completely.
The common sense says that when system is driven against too many constraints, the control system capabilities are misused. It corresponds the reasons of the financial crisis: “insure” instruments were used as money making instruments.
I feel that every time a control law is driven against “hard” protection constraints, someone has made, at leas a small, mistake in the chain form design to practice. If it is a common practice to solve problems by driven the system against protections, there is something fundamental wrong in the culture.
I feel that every time protection is activated the situation should be studied afterward and practice to avoid the activation of protection should be learned. If it is acceptable to use protections in a normal operation, the safety gains of the protections are destroyed. Most likely the safety is decreased because the skills of the pilot are not gained.
As far as I have understood right, the cause of this accident is pilot error. It is hard to deny it. The real question is: what lead to the errors. We might found out that they acted relatively reasonable...
As a control engineer I feel that would be relatively easy to substitute the pilot in the “normal” conditions. I feel that the pilots are in on board to solve unexpected situations. Humans are better than computers in those conditions. However, human has to learn how to act. If in the training everything goes too smoothly, I have to ask what in hell the pilots do in the cockpit? The problem solvers have to trained to solve problems.