First and foremost: thank you, Mike, for highlighting my piece! Very honored.
"Fair use" in trademark comes up in a few different contexts. Courts are not crystal clear as to the boundaries here, but our legal guide highlights a few of them. These include nominative fair use (using "McDonalds" to talk about McDonalds the company and its products), descriptive fair use (use of a trademark for its non-trademark meaning, such as referring to a deal as the "best buy" without reference to Best Buy), and more general First Amendment defenses (best articulated by Rogers v. Grimaldi and progeny, where courts give the contents of creative works more latitude through more abstract balancing tests).
I could see this situation cast in the "nominative fair use" light as a few commentators suggested, but I think it actually better fits into a whole different line of cases: the "genuine goods" line, which addresses circumstances where you are using the trademark to refer to the exact same goods as the trademark owner. The goods here are those that come from the Nutella company.
The contours of both "genuine goods" and "nominative fair use" are similar: in both cases courts have allowed the use of another's trademark, provided that the use doesn't go further to suggest a level of sponsorship that isn't there. As I noted in the particular post, I could see a skilled argument pushing a court either way on this one (though personally I never for a minute thought that Boloco had any sort of "deal" with Nutella). But that's hardly the point, as I think Nutella loses much more than it gains in approaching the issue this way.
Thanks for the comments on the paper. I hope you did take time to read it, as I do go into some detail as to why the analogy to obscenity is appropriate, how even the procedural safeguards common in copyright cases would call for greater safeguards here. And the law is evolving in this area, as the Salinger case identifies and Lemley and Volokh noted in their influential article on point over ten years ago.
I would love to examine this issue purely under copyright, but the law is novel. A probable-cause, ex parte standard for the removal of expressive material in the name of copyright is, as far as I can tell, unprecedented. But the analogy to obscenity (I think) is a completely reasonable one. Both obscenity and infringement examine the legality of speech entirely on the basis of its content. Both deal with content that, if they are not found to be illegal, is fully protected under the First Amendment. And both deal in dim lines and uncertainties. There are easy cases in both, sure, but there are many where reasonable minds (and reasonable judges) will split. So the procedural safeguards we adopt in obscenity should apply with copyright infringement as well, even though neither obscene speech nor infringing speech will be saved by crying "First Amendment."
The closest thing copyright had up until this point is the Rule 65(b) TRO, and that requires more procedure than is followed in civil forfeiture proceedings. A magistrate judge's only concern in examining a forfeiture warrant is whether there is sufficient probable cause stated. In a TRO, a court has to be persuaded that the hardships command that the court not wait until communication is established with the defending party.
The idea that "obscenity works different" than copyright is fairly empty, as there is nothing in the opinions I cited that suggests that the concerns about taking the speech out of circulation or seizure of expressive material without a warrant is limited to allegedly obscene works. The court spoke of speech more broadly than that.
The whole Eldred argument (both raised for and against my claim) is not really appropriate. That was looking at substantive First Amendment issues and their scrutiny, not anything regarding the correlating procedural concerns.
This is a bit scattershot, and for that I apologize. I do encourage you to read the paper, as I address many of these concerns you're raising directly.