Whenever a major aviation accident occurs, a good deal of time later, a detailed report is assembled by investigators analyzing the causes, risks and suggesting corrections. These are usually amazingly rich in information. I have written of some previously.

Today, the Canadian Transportation Safety Board released a report on the 2011 crash of a 737 in the far north. As usual, it is excellent reading about a tragic event. However, it seems to me to miss one important element - an elephant in the cockpit so to speak.

That is the pilots' over-dependence on automation. In this case, the pilot flying appears to have become confused with some instrument errors (compensating for compass weirdness around the north pole), and with interaction modes with the autopilot. The autopilots in these larger jets have many parameters & modes (and sometimes different modes in different control axes!). This particular B737 one appears to accept gentle yoke inputs to momentarily override it, to let a pilot nudge the airplane a little. Nudge it too far, and the autopilot mode quietly changes, so instead of following a navigational beam down to the runway, it starts dead-reckoning along a heading, or something like that.

In this case, the pilot appears to have thought the autopilot was in one mode, when it was actually in a different one; the pilot also started to nudge the controls because of a growing inconsistency because of the compass/heading indicators and the ground-based navigational transmitters. That led to stress, ignoring a first-officer's increasingly alarmed (but not sufficiently assertive) request to go around and try again. Instead they crashed, killing most aboard.

So, what stands out in this to me, is that these pilots kept the autopilot active - at all - when they started getting confused. Instead of reverting to basic stick & rudder flying, they retain automation despite its failing to improve the situation. This has happened again and again and again and ...

We have an autopilot too in C-GXRP. It's not a complicated model with many modes, but I still treat it with guarded contempt. If anything unexpected happens while the machine is running the show, it gets turned off immediately. (There is a thumb button right on the yoke just for this (plus circuit breakers and switches elsewhere to kill it harder if needed.) Simple stick & rudder flying can save the day by letting a pilot focus on the essentials of aviating - management of the airframe as a whole, with reference to basic controls & instruments, not management of all the equipment on board. I am a believer of the classic Children of the Magenta Line talk that advises the same thing to jet jocks. Nor am I the only one.

I'm just surprised that the TSB neglected to mention it.