I only quickly glanced over the paper, but I don't think the authors argued that the donkey/carrot model is a general mechanism for cell migration. From what I know, it seems that it would probably be a fairly specific mechanism, given that most/all cell migration and differentiation occurs through signal gradients from "non-fleeing" sources. Mind you, that's not at all a bad thing, because NC cells are fairly special in that they undergo a transition from epithelial to mesenchymal localization, which is also what all metastatic cancers do. However, whether the parallel holds meaningful when comparing each's molecular mechanisms is another issue entirely. I'm still skeptical because the mechanism requires two coordinating cell populations, the "donkey" AND the "carrot", which also happen to derive from different cell lines. I would guess that most non-NC cancers probably lack the "carrot" population either because the tissue type never had or needed one in development, or because the "carrot" population already got chased far away during normal development. In both these cases, any invasive tumors would need to arise through a separate pathway. If that's the case, then the research is not nearly as directly applicable as something that gives answers to all forms of metastasis. Even still, understanding potential mechanisms behind cancers that arise from NC (neuroblastoma, etc) is still really damn useful, so I'm glad this was done, but not so glad if people think the applications are more awesome than the data warrants.
