I'd say the major factor in what most people think of as DRM is the "restrictive" aspect. Hypothetically, there might be some stuff left over after stripping which might allow identification of the original source, but the potential tracking aspect is not what most people are probably thinking about when they mention how it affects their e-book usage.
And there's a spectrum of restriction and identification between ADE/Amazon's DRM and watermarking, with B&N/eReader DRM lying somewhere between, with its use of your name and credit card # as a password to unlock the
- single file you can download direct from their website
- without having to pass it through some other software which may not work on your chosen OS and
- can put on as many of your supporting devices after
- without having to authorize through a central server which may one day go offline
- nor download a separately encoded file for each device and have to re-download them all if something messes up your app authorization and you have to re-install (or just go straight to keeping the stripped files around for the inevitable eventuality of trying to transfer your ~2000+ DRMed freebie ebooks back to your K4Mac library).
For the record, if there has to be some kind of restrictive DRM on books which I pay actual money for, I would much prefer it to be the B&N/eReader form of DRM, and I hope support for it ends up a default turned-on option in the next few iterations of readers using the Adobe SDK.
Long story short (too late!): I vote for the "restrictive and non-restrictive DRM".
ETA: I'm perfectly fine with watermarking, and would take that over even B&N DRM, as long as it was unobtrusive and did not interfere with the reading experience/quality of included images and so forth.