One of the things that may need to be looked into on this is how the 767 control system responds if the pilot pushes the column while the autopilot is engaged. More recent models recognize pilot intervention via significant controller displacement as time to disconnect the autopilot. On older models the autopilot does not immediately disengage in response to pilot input. There may have been a period of time where both Hal and Row 0 were providing significant control inputs. Learning that the autopilot remained engaged would not necessarily mean that Hal was at fault.

This event was going on for 18 seconds. I would assume that one or both pilots had hold of the yoke.

I can't imagine that either did dot push the A/P disconnect on the yoke.

Note: I am assuming there is an A/P disconnect on the yoke as on previous Boeing's.

There is an autopilot disconnect button on the yoke and a disconnect bar on the glareshield Mode Control Panel on the classic B-763.

Also, the NTSB release said the throttles went to full power before the pitch over.

I also can't imagine not pulling the throttles and speedbrake lever back with the nose pointed down for whatever reason.

No, I'm sure they were trying everything humanly possible, exactly the way we all would have been. But if the darn automation wouldn't relinquish control, that would be a horrible feeling.

When you"re 49 degrees pitch down and going thru redline?

Also read that their speed increased from their assigned 230 kts to 430 kts in the descent, and that at impact it was close to 500 kts.

Deliberate act possible. Examination of pilots' histories needed.

This possible cause should be evaluated in the same way as a technical issue. Methodically and objectively.

Well, in fact, almost anything is possible until we hear differently!

They move the lift distribution outboard, so if you're pulling 2g with the speedbrakes out the root bending moment will be higher compared to a 2g manouevre with a clean wing.
In any case I doubt you'd still be accelerating if you're pulling 2g+ as the drag would be significantly higher

Yes, there is. The 767 doesn't have "high tech" computer "augmentations." It's a straight-forward, easy airliner to fly. A/T easy to disconnect or override. Autoflight easy to turn off.

Cornering Speed
 

Written from a fighter pilot viewpoint. When applying to a transport, there are going to be additional factors to consider such as control hinge moments, wing bending moment, and control system design. (For Airbus FBW, all such decisions have been already made for you.)

With the nose buried downward, and with concern about hitting the ground, you would like to achieve a minimum radius turn, but without tearing the aircraft apart.
The minimum radius turn is generally achieved where max AOA and max allowable G meet on the performance curves.

You can consider using up some of your safety margin when planning what G level to attain. Then look up the performance curves and determine a cornering velocity based upon your most probable gross weight.
Below cornering velocity, you can slightly improve performance by accelerating slightly and pulling into buffet. Above cornering velocity, you must slow down to improve turn performance and meanwhile avoid going beyond your max acceptable G (without a G meter).

Knowing your cornering speed is strictly emergency knowledge and would be used solely to ensure you are not completely out of the ballpark in your pull out efforts.
No doubt, the test pilots and aerodynamics guys can flesh this out better. As I recall, the F-4 cornering velocity was 420 knots. You definitely will not want to be anywhere near that speed when you pull out in your transport.

Correct. Less lift less load. That is how gust load relief via the ailerons works on the bus.

The interesting question would be, how the speed brakes influence the cornering radius pulling out of the dive. However, even here they won't make things worse I would guess.