Should Starliner be given another chance?

[KilroySmith]

So when Wilmore says "now we can't go forward", how do you read this?

I like that question.
I read it as "I commanded forward movement and" one of two things happened:
1. The movement was not forward, but in some other direction. Maybe there was a forward component, along with an x movement and a rotation. This would happen if the flight software was designed to do SOMETHING if a movement was commanded in such a degraded state.
2. Nothing happened.  This would happen if the flight software was designed to do NOTHING if it couldn't do what was requested due to the degraded state.

As an outsider (but a professional SW engineer), this is a very interesting part of the design constraints. If you run the math on a commanded motion and there's no solution given the current state of the vehicle, do you do nothing, or do you do something that you know isn't quite right?  In the latter case, you're relying on the more flexible programming of the meatsack to maybe resolve several different motions into a desired result.  In the former, you eliminate some of that flexibility but provide consistent responses to commanded motion reducing the likelihood of bumping into something you didn't really want to bump into.  This is a fascinating part of system design to me.

[JayWee]

As an outsider (but a professional SW engineer), this is a very interesting part of the design constraints. If you run the math on a commanded motion and there's no solution given the current state of the vehicle, do you do nothing, or do you do something that you know isn't quite right?  In the latter case, you're relying on the more flexible programming of the meatsack to maybe resolve several different motions into a desired result.  In the former, you eliminate some of that flexibility but provide consistent responses to commanded motion reducing the likelihood of bumping into something you didn't really want to bump into.  This is a fascinating part of system design to me.

Reminds me somewhat of the Airbus (FBW constrained by envelope) vs Boeing (well, you wanted it) control debate.
A commanded system should never ever do anything "random", especially something which might change with firmware updates.
I'd probably:
a) fully manual: have a display showing the 6DOF axis with red/yellow/green arrows for the pilot to evaluate.
b) semi-auto: have a trajectory planner "command: translate 10m forward" -> "ok, we'll spin 180deg and burn backwards" -> "okay, do it"
c) fully auto: (say, cargo mission), either faults and waits for ground operator do b) or is fully aware of surroundings and is sure it's safe (including fuel reserves) to proceed with it.