Interesting paper about torque based ECUs but i got to this
...
Author may be smart but should eat a dick for this bullshit alone.
I am not sure I understand your concern about this statement.
I am yet to read this to see what the rumble is about, but from what I have seen by scanning it the message is that a badly calibrated control algorithm can go out of its tracks, duuh...
There is one point though, that bothered me for a while now. If there are maps that are meant to be inverses of each other, and for performance reasons they are not calculated on the fly, why does the ECU not check their correspondence (with some reasonable error margin) during initialization, along with the checksums, etc. Not only it would add robustness to the system, it would also eliminate about 30% of posts around here