Let’s Be Reasonable About Geofence Warrants (Part 3) – New Jersey v. Van Salter

I have previously posted against categorical prohibition of geofence warrants, and worked through the dizzying array of opinions in the Fourth Circuit’s Chatrie en banc.  Here, let’s consider the Superior Court of New Jersey Appellate Division’s May 20 split decision permitting a geofence warrant as supported by probable cause in State v. Van Salter, No. A-3963-23 (N.J. Super. Ct. App. Div. May 20, 2025).

So as not to bury the lede, the upshot is this: There is much to like in the majority opinion, as it correctly asserts that probable cause can be established by selective revelation, meaning “geofence warrants are not unconstitutional per se.”  Yet there is also blood drawn by the dissent, as the proffered suspicion is very minimal on these facts, relying upon a string of ‘hopes’ and ‘maybes.’  And this is because the case presents an instance of geofencing in investigation of a single crime, thereby limiting the screening power of selective revelation.  Thus, as I argued in 2016 (and before), the best way to approach the problem is for the decision-maker—ideally a legislature in the first-instance—to first decide how private are various durations of location information.  For example, duration at a single instant might be not private or minimally private, whereas duration for over twenty-four hours might be highly private.  That privacy determination in turn ought to trigger different access restrictions.  And that is where this case understandably goes astray: because the New Jersey legislature has not done that work, the state’s courts have been left to look for probable cause in order to permit a warrant, even when a lesser standard of suspicion is almost surely the more reasonable one.

The facts of Van Salter are these: On March 16, 2022, a male entered a gas station store at 8:59 pm; he was facially masked and seemed to be speaking on an unseen mobile phone, as witnessed both by the station attendant and by a customer.  After that customer witness left, the man assaulted the station attendant at gunpoint, not only stealing the money from the registers but then punching the attendant and throwing her to the ground.  A surveillance camera at the station captured the perpetrator entering the store at 8:59 pm and leaving at 9:13 pm, making for a most relevant period of fourteen minutes.  Police therefore obtained a geofence warrant directed to Google and covering the fourteen minute window, hoping the perpetrator was one of those persons who had opted in to that company’s location logging.  Google responded that there was only a single ‘hit,’ meaning Google had information on only a single phone recording location at that place within that time window.  So, police followed up with a second warrant requesting identifying information for that phone, which belonged to defendant Van Salter.  (There is a discrepancy in the case records, such that perhaps there were four hits rather than one, a discrepancy I will set aside here.)

Thus, there occurred a violent robbery in which there was reason to believe the perpetrator spoke on his cellular phone at the location of the crime, within a fourteen minute window.  Given the significant government need and the relatively minor privacy intrusion involved in a person’s location over a fifteen minute period, I’d permit police to obtain lists of everyone within the gas station during that time.  Now, the available location information from a particular provider will be more or less granular, which is again why legislative rules—which can categorically approach such complexity—are superior to court necessarily-ad-hoc considerations.  But, again, I would permit police access to such information.  Indeed, I tend to think this is the reasonable solution whether or not the perpetrator was observed using a phone; it is reasonable given the realities of the pervasiveness of mobile telephony, the magnitude of the privacy intrusion of location over a short duration, and the seriousness of the crime at issue.

And this is why the ABA Criminal Justice Standards for Law Enforcement Access to Third Party Records (for which I was Reporter) counsel the decision maker to first determine not the reasonableness of particular access, but instead how private is the type of information.  The Standards recognize four levels of privacy: large (highly private), medium (moderately private), small (minimally private), and not at all (not private).  And each privacy level then defaults to a given magnitude of legal protection: Highly private information, for example, defaults to being highly protected, which typically requires a probable cause warrant for access.  And moderately private information defaults to being moderately protected, where there is a menu of access options.  Etcetera.  The crux is that it is not reasonable—indeed, it’s perhaps absurd—to think there ought to be a single criterion of access restraint (probable cause) for the myriad types and amounts of information police might wish to access.  And the law ought to be reasonable… including, of course, because such is the federal constitutional command as a floor, but also because it is dimwitted to deny the State reasonable access necessary to solve dangerous crime.

So, let’s bring it back to Van Salter.  The reason the dissent makes good points is because it is quite strange to hinge law enforcement access upon the perpetrator having been heard seemingly talking on an unseen phone—so the State couldn’t learn who all was present without that happenstance, even were the crime a murder?!—and for law enforcement to request data from only a single service provider (Google) for whom only one-third of its then-customers opted in to the relevant location tracking.  In the words of the dissent, “The bridge between the suspect and Google is built on a series of assumptions: defendant had a Google account capable of recording his location to Google’s server; he had opted in to permit Google to record the estimates of his location; the user agreement between Google and defendant, which is not in the record, provides the location ‘data belongs to Google’ and authorized Google to ‘share the results of [a] search or even sell it’; his phone was a smartphone; his smartphone had GPS functionality; his smartphone had Google’s operating system or a downloaded Google application with enabled location-tracking services; he had that particular phone with him at the time of the robbery; and he was logged into his Google account on that smartphone at the time of the robbery.”  Surely police access to location information—however it ought to be regulated—ought not rest on anything like that list of ‘maybes.’

And is there probable cause—or even reasonable suspicion—for most geofence warrants?  As I stated the matter in 2016,

[H]ow could there be reasonable suspicion in the case of the … cell-tower dump?  Only, it would seem, if very few people were talking on their cell phones.  Think of it this way: if police were physically present after the commission of the crime, could they temporarily detain—and thereby constitutionally seize—everyone standing within the relevant cell-site sector?  If there are many people present, of course not.  In other words, reasonable suspicion requires a respectable hit to no-hit ratio.  It can be constitutionally permissible to temporarily detain—though not, I think, to arrest—two persons when only one can have committed the crime.  But, as that number of potential suspects increases, at some point—and rather quickly, I think—reasonable suspicion dissipates.

Thus, again, it is incumbent upon decision makers—ideally legislatures in the first instance—to decide how private is given information, and then upon reasonable access restraints, and those restraints may be probable cause, reasonable suspicion, or something else.  When several crimes occur with a similar modus operandi, selective revelation may alone provide probable cause—e.g., there is only a single phone present at three robberies.  But for other crimes there will not be that probable cause… but maybe there doesn’t need to be.  Perhaps probable cause ought not be required, depending upon the particular law enforcement ask.

For example, the following are two tables from my 2016 article showing how a jurisdiction applying the ABA Standards might approach access to location information like that in Van Salter—the Standards not providing a legal straightjacket to any circumstance, but rather providing a framework or algorithm through which a decision maker can arrive at consistent, reasonable restraints:

Unfortunately, legislatures haven’t done their work in the past decade, so courts are still having to confront these not-at-all-new issues as matters of first impression with only two tools at their disposal: the federal and respective state constitution.  So, while there is much to like in the appellate decision in Van Salter, we won’t get this fully right until legislatures begin doing that job.  As I’ve expressed it, “When it comes to regulating law enforcement access to information, my preference is for legislative differential regulation, by which I mean a hierarchy of regulation proportional to privacy, yet responsive to law enforcement needs, subject to a constitutional backstop.”

It’s been over a decade… we are still waiting.