After you load a new init performance but BEFORE selecting Part 1 for edit if you rotate knob 1 counter clockwise you can see the value go to 0.

Before selecting Part 1 I imagine the knobs are set to common (Performance-level) assignable knobs. Not Part level. So rotating a knob at the common level doesn't necessarily affect the Part level knob unless the two are linked. What you described they shouldn't be linked.

Assuming Part 1 - you would look at Part 1's assignable knob position and this position will be the input value for the Control Assign curve.

The procedure for setting Part 1's Assign 1 would be:

1) [PERFORMANCE] (HOME) - just a sanity check to be sure you're on the home screen. Pressing this button will also de-select any Parts. On the home screen the 8 assignable knob positions are showing the Common Assignable knobs.

2) Touch Part 1. A popup menu shows up on the far left. You don't have to select anything here. You can press [EXIT] (the button) to go back to the home screen or just live with the popup in the far left. Either way - now these 8 knobs on the home screen (now that Part 1 is selected with a white box around it) will be showing Part 1's Assignable knob positions.

3) ... and after Part 1 is selected - now you can rotate Knob 1 since it will represent Part 1's Assignable Knob 1 (assuming the [ASSIGN] button is lit in the general case).

Init Normal (AWM2) defaults the Part and Common level Assignable Knobs to 64.

Off should mean OFF.

Jason never really got a good answer to this question he ask:

So again - why does the curve affect the pitch when I turn [MOTION SEQ ON/OFF] to off?

1. Init Normal (AWM2)

2. Control Assign - MS Lane 4 as source, Part Pitch as destination, Polarity = Bi

3. Pitch is affected

Note that the master sequence switch was OFF (default), the part sequence switch was OFF (default) and all 4 MS  Lane switches were OFF (default.

No sequences were EVER used, or harmed, during the tests.

The above will affect the pitch by the simple assignment of Lane1 (2,3,4) as the source. If you assign Lane 4 as the source then step 1 of the sequence for Lane 4 will be taken into account in affecting the pitch.

So an updated version of Jason's question might be: why is step 1 of a motion sequence being used when the motion sequence functionality has NEVER been enabled or used?

You get the same result no matter which of the 8 motion sequence selections you select. 

They seem to be little known facts but:

1. all Motion Sequences, for ANY part, are already defined and populated by default with default sequence curves, steps and parameters. A bulk export of the INIT NORMAL (AWM2) confirms this. That means that each part has 32 sequences predefined: 8 sequences of 4 lanes each.

2. users NEVER actually create a sequence - they simply modify one of the predefined ones (see #1 above)

3. users can NOT delete a sequence nor can they turn it ON or OFF.

4. users can NOT 'reset' a sequence or otherwise restore it to a set of default values.

5. some presets have modifications to the 32 sequences of their parts even though the sequences aren't used in the performance and with some of the LANE switches set to OFF. A sound designer creates a performance, tests some motion sequences and then abandons some of them (see #4). They have no way to actually delete or clear them. So when such a performance is used as a starting point any such left-over sequence 'garbage' will still be part of it. Turn on the LANE and you can't know for sure what the sequence contents will be.

Users can, of course, turn the Master Sequence switch ON or OFF, turn the part Sequence switch ON or OFF and turn 1-4 Lane switches ON/OFF. None of those settings will change the results of what Jason posted or the simple test above.

So why should a sequence affect a control assignment destination if NONE of the 4 lanes of the 'selected sequence' are active?

That is in addition to the issue Jason raised about sequences affecting things if the master or part sequence switches are OFF.

To reiterate what Jason said earlier: the above is merely to document 'how it works' and isn't meant to be a mandate for change.

Back in 2016 when I asked this there wasn't a vehicle for suggesting improvements -- and I wasn't necessarily suggesting an improvement in my thread.  What I was expressing was, as an analogy, when I walked into the "Yamaha" room the light switch was on.  And I looked up at the ceiling and saw lights that were filling the room with soft white illumination.   I found a light switch in the room and expected when turning off the switch the room would go dark with exception of a few things in the room that had their own small illumination (my phone, etc).  But when I switched the light switch off the lights in the ceiling now illuminated the room with a green color.  I looked at the switch again and with the switch up it was labeled "ON" and when it was down it was labeled "OFF".   There is a reasonable amount of missed expectations here -- and, likewise, the labeling and language around motion sequence for this button could set up the unknowing to think one thing is going to happen (a dark room) rather than what does (a green room).  

How the system does work has been sufficiently documented - which was my main goal.  To document for myself and everyone else. 

If we want an "off" that really disconnects the lanes from the destination parameters then that's fair game for an ideascale.  In the meantime, I have just made sure that t=0 in my pulses result in a zero offset if I really want that to happen with MS OFF.  Which seems like a kludge - but you've got to work with what you have.  In the end, I just avoided using MS OFF altogether and found different ways to arrive at these goals.