Like the entire DOI system, this is a bit breezy. The missing qualification is that DOIs are not unique, and thus “content object” doesn’t mean anything as operationally defined. Suppose one issues a commemorative issue of a journal consisting mainly of reprints, and recall that DOIs were originally touted as the solution to deep linking (something that never even came close to occurring). Does the reprinted version of Figure 1 of Article 1 have the same DOI as the original? Does Article 1 itself have the same DOI as the original? By necessity, no. The framework is just a label printer. As such, “because of the DOI system’s technology, DOI links are persistent” obscures the fact that there is *no new technology*–it’s merely a subscriber-based contractual arrangement in which one pays $35,000 a year for the privilege of adding an extra step to solving the original problem, which is keeping track of where you’ve put the data, and getting in return an uninformative user name and somebody else’s branding.