For more than a decade, a New Jersey appellate court’s opinion guided courts nationwide as they determined whether to enforce subpoenas for the identities of anonymous online speakers.

But in 2013, courts increasingly moved away from the New Jersey court’s test, leading to further divergence in the methods that courts use to determine whether a request for an anonymous Internet user’s identity violates the First Amendment or discovery rules.

In Dendrite International v. Doe No. 3, the New Jersey Superior Court Appellate Division held in 2001 that before enforcing a subpoena for the identity of an online speaker, the issuer of the subpoena must: (1) notify the anonymous speaker of the subpoena and provide the speaker a reasonable opportunity to respond; (2) specifically identify the speech at issue; (3) allege a prima facie cause of action; (4) present sufficient evidence in support of each element of the claim. If the issuer has satisfied these procedural requirements, the trial court must balance the anonymous speaker’s First Amendment rights against the strength of the plaintiff’s prima facie case and the necessity of disclosure. Publishers and web sites generally support the use of Dendrite because the test shifts the burden from the site to the commenter and enables the commenter to decide whether to fight the subpoena.

Although the Dendrite ruling came from a New Jersey intermediate appeals court, state and federal courts nationwide relied on the ruling over the next decade. Among the courts to rely on Dendrite are the Maryland Court of Appeals, the New Hampshire Supreme Court, and the U.S. District Court for the District of Connecticut.

Over the past year, however, some courts have moved away from Dendrite. Most notably, last April, the New Jersey Superior Court Appellate Division reversed a trial judge’s decision to quash a subpoena for the identity of individuals who allegedly hacked a hospital’s intranet to circulate defamatory messages about hospital employees. In Warren Hospital v. Does 1-10, the appellate court held that Dendrite did not apply because the defendants’ actions were “no different than if they had broken into the hospital and spray painted their messages on the hospital’s walls.”

Also in April, the Court of Appeals of Michigan reversed a trial court’s application of Dendrite to a subpoena issued by a law school, seeking the identity of an individual that allegedly defamed the school anonymously online. In Thomas M. Cooley Law School v. Doe, the appellate court held that “Michigan rules of civil procedure adequately protect Doe 1’s constitutional interests.”

Similarly, in June, the U.S. District Court for the Southern District of New York rejected the application of Dendrite to a lawsuit against defendants who anonymously published blogs that infringe trademarks owned by Bloomberg, L.P. Instead, the court relied on a less stringent standard that courts have used in online copyright infringement cases.

And in July, the U.S. District Court for the District of Colorado denied a blog administrator’s motion to quash subpoenas for the identities of 10 John Doe defendants who had allegedly defamed the plaintiff company on the blog. The blog administrator relied largely on Independent Newspapers, Inc. v. Brodie, a 2009 Maryland Court of Appeals opinion that had adopted the Dendrite test. The Colorado federal court held that Brodie is not binding. Acknowledging that “I cannot be certain that each IP address corresponds to an offensive comment,” Judge Ellen L. Hollander wrote that “absolute certainty is not required.”

The recent decisions demonstrate that courts have yet to settle on a uniform First Amendment test for the right to anonymous online speech, and increase the likelihood that federal appellate courts or the U.S. Supreme Court will have to confront the issue.