Accordingly, the evidence we have on nicotine being "non-addictive", which comes from studies in which patches are used needs to be viewed with a degree of skepticism.
Yes, it does, especially since there's such a paucity of data regarding nicotine use in tobacco-naive individuals. "Smoking" and "nicotine" have been conflated for so long in the public consciousness (among scientists and lay people alike) that we still know very little about what happens when you administer the latter without the former; all we really do know is that when you administer nicotine orally or transdermally in tobacco-dependent individuals, it very rarely results in tobacco cessation.
It seems to me as though this is a case where an RCT using existing smokers would be the optimal means of investigation. Supply one group with their regular brand of cigarettes and have them smoke as they normally would (control group), and give the other group cigarettes made of zero-nic tobacco, or as close to zero-nic as possible, but which are packaged to appear indistinguishable from their regular brand (placebo group). Encourage the members of both groups to try quitting if they're so inclined, and if they demonstrate such inclination, supply them with the same materials and resources you normally would. If, after 12-18 months, the placebo group shows an appreciably higher quit rate than the control group, then we can state with some degree of confidence that nicotine is the primary factor, or at least one of the primary factors, that cause smoking dependence.
Obviously, there would be some rather sticky ethical considerations involved in such a trial, and maintaining proper controls for the duration of the study period would be a major challenge, but it these obstacles could be overcome, we would probably glean much more instructive data on this question than we currently have.