Howard, Serge, Floyd, Chris, Brett (...forget anyone? if so apologies...) I'll be brief. (Yeah, right...)
Study Serge's diagram. The bellcrank plate is mounted parallel to the span and the long axis of the model, which creates much of the difference in response either way from neutral. (Which is small, and we've been been able to manage it, by feel, ever since...)
A way to reduce that geometric "error" is at the flap horn, where the line from the pushrod connect to the flap hinge, at neutral, should be perpendicular. This still has that forward and back discrepancy from the 'flat' bellcrank plane of rotation.
Now, what if you mounted the bellcrank plate tilted so it and the rod from bellcrank to flap horn were parallel at neutral? It won't stay that way over the full bellcrank rotation, of course. Perhaps put the rod parallel at about 5° to 10° each side of neutral? That should minimize ONE distortion...
Another view of things? View the model from the top. (Never considered this view? Hmmmm...?)
In the top view, and for the same reason we'd want the flap pushrod to make the least up and down deviaton, wouldn't we want the least 'off-axis' sideways motion? ...The flap hingeline is almost always perpendicular to the fuselage long axis, or for a swept hingeline, the angles away from pependicular are the same to each side.
If we must put the flap horn on the fuselage centerline, THAT determines where we should mount the bellcrank pivot (in the top view.) For the inboard bellcrank-to-flap-pushrod connection (UP line forward) we'd need to mount the bellcrank pivot further outboard. Or, if we MUST put the bellcrank pivot at a certain place, THAT determines where the flap horn goes, across the fuselage width. The same +/- 5° to 10° offset used for the side view can help keep the center (neutral) 'off-axis' motions small.
Summing up:
1) In the side view of the fuse, tilt the bellcrank mount so that it and the bellcrank-flap pushrod are at -or near- parallel at say +/ 5° to 10° from neutral?
2) In the top view, 'center' the bellcrank-flap pushrod's sideways swing the same way.
I have gunned up a spreadsheet documenting the deviations using this "aligned" bellcrank/flap horn idea. A monster which takes me several minutes to 'get back into' every time I consult it... Anyway, the deviations caused by different bellcrank-out and flap-in arm radii are very small from neutral to ~30° each way, then they start to diverge, gradually but increasingly. (The graphs of the deviations impress me!)
Final thought: If we keep the "response relationship" the same with a larger bellcrank arm radius, we have to increase the flap and elevator horn radii to preserve that relationship. It will 'feel' slower because we have to move the handle further to turn the bellcrank to comparable angles. Effort applied to the handle will decrease, because of the longer lever arms, but to turn the flaps and elevators to the same angles, the bellcrank must turn the angles we used with 3" bellcranks. Torque depends on force and arm length. Any change in one demands a compensating change in the other.