A written paper usually treats every point the same. Ten points for a hard item, ten for an easy one.
Difficulty-weighted scoring changes that. Each item contributes in proportion to the difficulty its author set. A hard question is worth more of the paper than an easy one.
The model is off until you enable it. Enable it and two things change. Points are assigned a weight, and the pass rule receives a second test.
The difficulty here is the author's own 0-10 label, chosen while writing the item. The eleven question types cover that scale and its four named bands. The measured p-value is an entirely different number, and it is determined afterward in item analysis.