Take the case of anonymity clients--http://en.wikipedia.org/wiki/Tor_(anonymity_network)--extreme perhaps, but illustrative of the basic point that the internet was consciously designed to enable the technical distancing of communication from attributes that would otherwise qualify identity.
Its worth protecting some anonymity, and people who really really need that can get it. For the average person, internet anonymity is a bald faced lie though--every site we visit embeds 100's of cookies, etc...
What the article urges, to my mind, is that any judicially formalistic approach to anonymity issues demand a probing factual analysis of the context in which the identifying information is sought.
The recent RIAA/ISP 'agreement' seems to stipulate that technical procedures at revealing anonymity in the context of copyright infringement, even if it is a violation of either first or first amendment rights for the accused, is justifiable to protect the established state interest in maintaining an incentive for the production of artistic material.
This is because the methodology utilized prior to the information request comes within a reasonable modicum of statistical certitude that such a violation has taken place. The utilization of traffic analytics on P2P servers is certainly sufficient justification for the issuance of a warrant.
Such an inquiry is clearly impossible in the context of the defamation suits without the ISP investigating the content of a users browsing history, e.g., what pages she has visited. This is certainly not a narrowly tailored means for the forcible disclosure of anonymity, unless the interest is just fucking amazingly compelling--SUPER SPECIALS or something like that (or terrorism evidentially).