Understanding gains

So I finally got my 6 DOF built and have been doing some very brief trials between racing and flying to see how it reacts and working on figuring out how to tune it. In general the whole system seems like it's probably a little too "exuberant". In a car I don't think it's quite as bad or as noticeable (maybe also just because I'm more used to driving a car vs flying a plane), but my quick test in MSFS last night in a little Cessna 152 had the rig throwing me around at just a light touch of the rudder pedals.

What I'm not sure I understand is why the base gain settings are set the way they are. The linear movements (sway, surge, and heave) all have a base input gain of 10, while the rotational movements (roll, pitch, yaw, and the "blended" axes) are all set to 1. Could anyone explain why? Does this just make the filters work better (like if they didn't deal with decimals well or had rounding issues) or is there some study that says in testing those forces needed to be larger since they are shorter stroke or something like that?

I'm only guessing here. I also have to learn a lot more of how the gains, filters etc. work in FlyPT Mover. But one possible explanation of the different gains are the different units of inputs and outputs.

The telemetry data from the sim software usually comes in metric units, for example m/s for speeds and m/s² for acceleration. AFAIKS, outputs are in mm for linear motion and degrees for angles. Some motion controllers also need "number of motor turns" instead of mm and °.

So a lateral acceleration of 1g together with gravity should result in the rig tilting to a 45° position to generate the same cue. This means you have to scale the 9.81m/s² lateral acceleration roughly by a factor of 4.5 to get the effective neutral gain of one.

IMHO, the default gains in the Mover settings are a bit too high. We reduced most of the output gains to something around 0.3 to 0.5 to get realistic results.

I did wonder if the input value had something to do with it. I know the linear moves are all mostly calculated straight from the acceleration values, while the rotational ones are often based more on the change in rotation. So you get the onset effect and initial motion, but not the sustained tilt, like when you first roll a plane and enter a turn you would feel the tilt, but it would go away as you sustain the turn.

I had another thought the other day about how I might go about tuning things. I'm debating turning off all but one axis at a time and working on the tune for that axis, then slowly adding the others back in. I'm not sure whether to do the linear and rotation at the same time (like surge and pitch for instance), or whether to just do surge first, then pitch after. To some degree I think I'd want to do them together as ultimately the blended outputs rely on each other. I think getting rid of all the extra forces going on would help to get things dialed in better. I'm not sure what vestibular system is going to make of it, but it seems like a good idea on paper at least.