The disco can pivot slightly in its vertical plane, remember the disco is kept in alignment by the frames lower shelf (bottom of the disco) and the sear cages lower bear surface area (top of the disco). In a perfect world there would be no vertical dispersion of the disco.
From looking at the discos engagement face after modification, it appears the difference we remove is keeping the trigger bar and disco face perpendicular to each other and thus compensating for the lack of correct initial alignment. If the disco pivots on it's plane even slightly (and it has to), then the engagement surfaces are no longer parallel to each other. This will create a high spot at the trigger bar/disco interface.
Cutting the disco to compensate for the mis-alignment distributes the load over a wider bearing surface area, thus making the DA more linear and smoother. The opposite is what's been occurring, i.e. a smaller engagement surface can create more drag and increase the DA pull.
I feel all we have done is compensate for the lack of proper alignment.