That's quite different to how I've been doing it. If I'm understanding correctly you seem to be using the reduction value as a percentage by which to reduce rotation delta. 26% - 49% to bring it down between your linear lines.
A couple of problems I see with that -
The scoop function in game has no concept of wheel deltas and is working purely with forces. There isn't a linear 1:1 relationship between the two. ie. Reducing forces by 50% won't reduce the wheel delta by the same 50%. As your graph shows, The force required to hit 80 is 0.38. To hit 40 requires a force of 0.28.
I think that's the main difference to our approach. I'm using reduction as the literal amount that forces are reduced by directly below the knee. Another thing - I don't think forces are 1:1 past the knee. They do however scale up so that an input force of 1 will output as 1.

I'm not sure that forces are simply lerped from the knee out to the extents (0-1) as this graph implies though.
Will do some more testing today.