To me if that gradient field from 0-100% pledge and 0-100% delegation is accurate then it could benefit from revision. Clearly using this field the k-parameter is not driving real network decentralization.
How can we revise it though? Exactly what numerical value should we propose and why? We should both work on that.