What is the theoretical basis here? By increasing ISO (ie giving less exposure), you are presumably shunting shadow values down into the toe of the characteristic curve, where they will be more compressed. At the same time, you fit all or most of the mid and light tones onto the straight-line section, whose slope/contrast you increase by longer development. Is that right?
I wonder whether
@aparat could oblige us with comparable curves of Tri-X and HP5+ in the same developer, so we might see what the difference is?