Also, just to be absolutely clear on this: this approach will technically not guarantee a true k-shortest path solve, so we must be careful not to incorrectly label it as such. As I said previously, it is merely an approximation for this concept. A more appropriate name might be to call this a k-alternate path approach (as opposed to shortest). The alternate paths you get will ultimately depend on the scale factors you use. The good thing, however, is that it is therefore a parameterizable approach, which can be tuned to the user's needs.