In re Search of Info. that is Stored at the Premises Controlled by Google LLC
In re Search of Info. that is Stored at the Premises Controlled by Google LLC
579 F.Supp.3d 62 (D.D.C. 2021)
December 30, 2021

Harvey, G. Michael,  United States Magistrate Judge

Internet of Things
GPS
Scope of Warrant
Mobile Device
Criminal
Cloud Computing
Privacy
Download PDF
To Cite List
Summary
The government sought a geofence warrant to collect ESI from a [Redacted] center in order to identify potential witnesses to alleged criminal activity. The Court approved a two-step process, whereby Google would first furnish the government with an anonymized list of devices found within the geofence, and then the government could, at its discretion, request from Google “account identifier and subscriber information” of any device found within the geofence. The Court noted that the disclosure of anonymized location data to the government at step one does not implicate significant privacy concerns, but that the deanonymization of the data at step two may trigger Fourth Amendment protection.
IN THE MATTER OF THE SEARCH OF INFORMATION THAT IS STORED AT THE PREMISES CONTROLLED BY GOOGLE LLC
Case No. 21-SC-3217 (GMH)
United States District Court, District of Columbia
Signed December 30, 2021

Counsel

Brooks H. Spears, McGuireWoods LLP, Tysons, VA, George J. Terwilliger, III, McGuireWoods LLP, Washington, DC, for Mark Meadows.
Douglas N. Letter, U.S. House of Representatives Office of General Counsel, Washington, DC, for Nancy Pelosi, Bennie G. Thompson, Elizabeth L. Cheney, Adam B. Schiff, Jamie B. Raskin, Susan E. Lofgren, Elaine G. Luria, Peter R. Aguilar, Stephanie Murphy, Adam D. Kinzinger, Select Committee to Investigate the January 6th Attack on the United States Capitol.
Harvey, G. Michael, United States Magistrate Judge

MEMORANDUM OPINION

Imagine a homicide in an alley caught on a nearby surveillance camera. The video is clear enough to see the attack, but too grainy to identify who did it. It is obvious from the video, however, that the perpetrator is checking his cell phone as he walks out of the alley. Having determined the location and time of the killing from the video, would it be lawful for the police to obtain a warrant leveraging the tracking capability of mobile devices to identify whose cell phone was in the area of the crime when it occurred? On October 6, 2021, the Court was presented with a warrant that asked that question, albeit not in a homicide case.[1] Commonly referred to as a “geofence” warrant, the government's application asked the Court to direct technology company Google to identify, through a multi-step process, the cell phone users that crossed into a defined geographic area around where the criminal activity under investigation occurred.
Though geofence warrants raise a number of important constitutional questions, there is not much federal caselaw discussing their legality. As of the date of this decision, the Court could identify only four federal decisions, three of which denied the geofence warrant under consideration.[2] That seeming reluctance to grant such warrants does not appear to have slowed their acceptance by other courts and their use by law enforcement. According to a *68 recent report, Google received over 11,554 geofence warrants in 2020, up from 982 in 2018; as of August 2021, they comprised nearly a quarter of all warrants served on Google.[3] Each of those warrants was authorized by a judge. So, it would appear that many more geofence warrants are being granted by courts than denied.
On October 13, 2021, the Court granted the geofence warrant application at issue in this case. This memorandum opinion explains why it was granted and, in the process, adds to the limited federal caselaw discussing the legality of such warrants.
I. BACKGROUND
A. Geofence Warrants
Soon after the advent of smart phones with the capability to track the location of their users, law enforcement sought warrants, or other legal process, to obtain the Global Positioning System (“GPS”) data such cell phones collect to track their users who were known to be engaging in criminal activity. See, e.g., In re Application of U.S. for an Order Authorizing Disclosure of Location Information of a Specified Wireless Telephone, 849 F. Supp. 2d 526, 530 (D. Md. 2011). Often referred to as “GPS warrants,” such warrants are, at this point, routine. They are frequently authorized, for example, to permit law enforcement to track a known or suspected drug dealer's cell phone to assist in locating the dealer's points of sale, drug stash houses, suppliers or co-conspirators. See, e.g., United States v. Burnett, No. 12-CR-00042 2013 WL 12430549, at *2 n.4 (D.D.C. Mar. 21, 2013) (noting that the government *69 sought search “warrants for GPS tracking data for [a suspect's] cellular telephone” as part of its investigation into drug trafficking), aff'd, 827 F.3d 1108 (D.C. Cir. 2016).
The warrant before the Court is different. It is what has been termed a “reverse-location” warrant: the perpetrator of the crime being unknown to law enforcement, the warrant identifies the geographic location where criminal activity happened and seeks to identify cell phone users at that location when the crime occurred. The “geofence” is the boundary of the area where the criminal activity occurred, and is drawn by the government using geolocation coordinates on a map attached to the warrant.[4] See Google I, 2020 WL 5491763, at *1 (explaining that a geofence warrant “is a warrant to obtain cellular phone data generated in a designated geographic area”); see also United States v. Asghedom, 992 F. Supp. 2d 1167, 1169 (N.D. Ala. 2014) (“A ‘geofence’ is a perimeter set up around certain locations based on the [ ] coordinates of the locations.”). The fence's boundary is quite flexible. Virtually any shape of any size that can be drawn using geographic coordinates can be used, including rectangles, triangles, or other irregular shapes, like the perimeter of a building or the length of a street. See Google III, 497 F. Supp. 3d at 351–52 (describing geofences constructed by the government as a triangle, an “ ‘L’ shape,” and a square). Further, circles can be constructed by instructing Google or other companies that store location information to collect data from a certain radius around a fixed point. The geofence is also bound by a time window dictated by when the crime is believed to have occurred (for example, between 1:00 p.m. and 1:15 p.m. on February 15, 2021). The time periods are also flexible; they can be as narrow or as broad as the facts of the criminal activity require and the law permits.
Understanding how Google obtains geolocation information from cell phones and the identities of their users is more complicated. Cell phones are basically “minicomputers that also happen to have the capacity to be used as a telephone.” Riley v. California, 573 U.S. 373, 393, 134 S.Ct. 2473, 189 L.Ed.2d 430 (2014). And like computers, cell phones—including phones with Google's Android operating system (“OS”)—need to “connect” with a variety of other devices in order to function. See Affidavit in Support of an Application for a Search Warrant (“Warrant Aff.”), ¶ 7. For example, Google OS phones must connect with wireless internet (“wi-fi”) devices in order to access the Internet. See id. Likewise, phones must use Bluetooth technology to connect to wireless headphones and other devices. See id., ¶ 8 (explaining that “Bluetooth uses radio waves to allow the devices to exchange information”).[5] Similarly, *70 when using the “map” function, Google OS phones use an internal GPS to ascertain the phones’ exact location. See id., ¶ 9. When Google OS phones connect to the Internet via wi-fi or scan their environments for Bluetooth devices or use GPS, they send information on the phone's location to Google.[6] See id., ¶ 16. “Google uses this information to calculate the device's estimated latitude and longitude,” and makes this information available for viewing by the phone's user in a service called “Location History” (the location information itself is stored on Google's servers). Id., ¶¶ 14, 16. Phone users must opt in to the “Location History” service, and Google has said that, as of 2019, about one-third of Google users had done so.[7] See id., ¶ 17. But even when users opt out of Location History or other location tracking, Google can still track and store their device's location data.[8] See id., ¶ 19; see also Google II, 481 F. Supp. 3d at 737 n.3 (“Published reports have indicated that many Google services on Android and Apple devices store the device users’ location data even if the users seek to opt out of being tracked by activating a privacy setting that says it will prevent Google from storing the location data.”).
Google also collects location information from phones that do not use Google OS (for example, iPhones) but nevertheless use Google applications, such as Gmail; YouTube; Google Maps; and Google's Internet browser, Chrome.[9] See Warrant *71 Aff., ¶ 19; see also Google III, 497 F. Supp. 3d at 350–51 (explaining that “Google apps exist for, and can be downloaded to, phones that do not run the Android operating system, such as Apple devices”). Google's Android OS is used by nearly 74 percent of the world's smartphone market, with a market share of approximately 46 percent in the U.S. See Warrant Aff., ¶ 19. But because iPhone users (who represent 23 percent of the worldwide market and 54 percent of the domestic market) also utilize Google applications on their devices—including Google Maps, YouTube, and Google's search engine—Google's cache of location information is even greater than its already-substantial market share suggests.[10] See id.
The cell phone location information Google collects is usually quite accurate—to within 20 meters, according to Google. See Google Amicus Brief at 15. This is because it is derived from multiple sources of location information including not only GPS data, but also from cell sites or towers, wi-fi networks, and Bluetooth devices. See Warrant Aff., ¶ 16. However, the location information Google collects is not perfectly precise, and includes a margin of error. See id., ¶ 15. Although the margin of error for each device depends on the quantity and quality of the location information the device transmitted to Google, Google says it aims to accurately capture the location of “at least 68% of users.” Id. Simply put, when Google searches its servers for the devices within a defined location boundary, a device that is outside of the boundaries of the geofence may be listed as within the boundaries of the fence due to imprecision in calculating the device's exact location. Conversely, a device that was within the geofence may not be recorded as being there. Google aims to correctly locate about two-thirds or more of its users.
The final piece of the puzzle is connecting the location data Google collects back to a particular user. According to the warrant affidavit, Google does this by collecting a phone user's information when the user registers for a Google account. See id., ¶ 20. In the registration process, the user can provide Google with their name, physical address, email address, and bank information, among other identifying information. See id. The account registration process is critical because many of Google's key application and features “are accessible only to users who have signed into their Google accounts”; others, including YouTube and Google Maps, are of “limited functionality” if the user is not signed into their Google account. Id., ¶ 11. When Location History is enabled, the location information Google collects from a device is associated with the account being used on the device—and that account can be traced back to a particular user with the information provided during the registration process. Id., ¶¶ 14, 17.
Thus, the geofence warrant works like this: Phones with Google OS and Google applications communicate location information to Google through use of common phone applications (e.g., YouTube and Google Maps) and other functions, including use of wi-fi, Bluetooth, and GPS. That location information is then associated with the device, and the device can be traced back to a particular account and accountholder who has “signed in” on the device. The search warrant at issue ultimately seeks, through a multiple-step process, information that can be used to identify the *72 accountholder associated with devices that Google has located within a certain geographic area, and during a specific time window, set by the government.
B. The Government's Requested Geofence Warrant
To avoid exposing the details of an ongoing criminal investigation, the Court will only generally describe the government's geofence warrant application. The geofence data sought in this case covers a [Redacted] center, where the government alleges federal crimes were committed. The [Redacted] center is located in an industrial area and shares a building with another business. See Warrant Aff., ¶ 38. The building premises abuts a road on two sides; between the road and the building is a parking lot. See id. According to the government, the [Redacted] center at issue “has a small customer service area” accommodating five or six customers “and provides limited services.” Id. To illustrate the geographic area where it seeks Google's location data, the government has drawn a triangle on a satellite map included as an attachment to the application. See Warrant Application, Attach. A, at 4. The triangular area is scalene and appears to cover a portion of the front-half of the [Redacted] center, plus its parking lot. See id. No other structures are included in the geofence area, and the part of the building the [Redacted] center shares with the other business is excluded. See id. Although the government does not provide the approximate area of the triangle, the shortest side is approximately 35 meters long. See id. at 3. The other sides appear to be between 45 and 50 meters in length, and the height of the triangle appears to be between 30 and 35 meters, suggesting an area of up to 875 square meters.[11] One edge of the geofence runs alongside the road abutting the [Redacted] center premises, that is, the road is not in the area covered by the geofence.[12] See id. at 4.
The government seeks a total of 185 minutes of geofence data for the geofence area. See Warrant Aff., ¶ 36. The 185 minutes are split into segments ranging from 2 to 27 minutes on 8 specified days over a roughly five-and-a-half month period, corresponding to the criminal activity under investigation. See id. Most of the time segments cover the early to mid-afternoon. See id. The government can be relatively precise about the data it requests because it has obtained (via subpoena) CCTV footage from inside the [Redacted] center showing the criminal activity as it occurs. See id., ¶ 29. The government represents that the footage shows suspects using cell phones—and in some cases more than one phone—when they engaged in that activity.[13] See id., ¶¶ 35–36. The footage *73 also shows the suspects “were either by themselves” when inside the [Redacted] center, or, if there were other customers, there were “on average” no more than 2 or 3 others. Id., ¶ 38.
The government proposed a multi-step process for obtaining the geofence data from Google. In its initial application, the government's proposed the following process:
a. Using Location History data, Google will identify those devices that it calculated were or could have been (based on the associated margin of error for the estimated latitude/longitude point) within the [geofence area] during the time period described .... For each device, Google will provide a anonymized identifier, known as a Reverse Location Obfuscation Identifier (“RLOI”), that Google creates and assigns to device for purposes of responding to this search warrant; Google will also provide each device's location coordinates along with the associated timestamp(s), margin(s) of error for the coordinates (i.e., “maps display radius”), and source(s) from which the location data was derived (e.g., GPS, wi-fi, bluetooth), if available. Google will not, in this step, provide the Google account identifiers (e.g., example@gmail.com) associated with the devices or basic subscriber information for those accounts to the government.
b. The government will identify to Google the devices appearing on the list produced in step 1 for which it seeks the Google account identifier and basic subscriber information. The government may, at its discretion, identify a subset of the devices.
c. Google will then disclose to the government the Google account identifier associated with the devices identified by the government, along with basic subscriber information for those accounts.
The Court had concerns about this protocol, namely the fact that government could, “at its discretion,” order Google to disclose the identifying information for certain accounts without any guardrails on the exercise of that discretion or further review by the Court. After discussions with the government regarding this issue, it submitted a revised warrant application (the one currently before the Court). In the revised application, the multi-step process is described as follows:
a. Using Location History data, Google will identify those devices that it calculated were within the [geofence area] during the course of the time periods laid forth in [the warrant].
b. For each device: Google will provide an anonymized identifier that Google creates and assigns to device for purposes of responding to this search warrant; Google will also provide each device's location coordinates along with the associated timestamp(s), margin(s) of error for the coordinates (i.e., “maps display radius”), and source(s) from which the location data was derived (e.g., GPS, Wi-Fi, Bluetooth), if available. Google will not, in this step, provide the Google account identifiers (e.g., example@gmail.com) associated with the devices or basic subscriber information for those accounts to the government.
c. The government will then review this list to identify devices, if any, that it can determine are not likely to be relevant to the investigation (for example, devices moving through the Target Location(s) in a manner inconsistent with the facts of the underlying case).
d. The government must then, in additional legal process to the Court, identify the devices appearing on the list produced by Google for which the government *74 seeks the Google account identifier and basic subscriber information.
e. In response to this additional legal process, the Court may then order Google to disclose to the government the Google account identifier associated with the devices identified by the government to the Court, along with basic subscriber information for those accounts.
See Warrant Aff., ¶ 37 (emphasis added). Thus, in the revised protocol, the discretion as to what devices falling within the geofence to deanonymize no longer rests with the government, but with the Court.
II. LEGAL STANDARD
Because the government applied for a search warrant, the Court assumes (but does not decide) that the Fourth Amendment's restrictions on searches and seizures apply to the collection of cell phone location history information via a geofence.[14]
The Fourth Amendment enshrines the “right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures.” U.S. CONST. amend. IV. As such, “no warrants shall issue, but upon probable cause, supported by oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized.” Id. Accordingly, two foundational issues that must be addressed in assessing any warrant are probable cause and particularity. See Kentucky v. King, 563 U.S. 452, 459, 131 S.Ct. 1849, 179 L.Ed.2d 865 (2011); see also Matter of Search of Info. Associated with [redacted]@mac.com that is Stored at Premises Controlled by Apple, Inc., 13 F. Supp. 3d 157, 163 (D.D.C. 2014) (“The Supreme Court has interpreted the Fourth Amendment to require: (1) that the warrant be issued by a neutral magistrate; (2) that the neutral magistrate find that there is probable cause to believe that the evidence sought will ‘ “aid in a particular apprehension or conviction” for a particular offense’; and (3) that the warrant describes with specificity the ‘ “things to be seized,” as well as the place to be searched.’ ” (quoting Dalia v. United States, 441 U.S. 238, 255, 99 S.Ct. 1682, 60 L.Ed.2d 177 (1979))).
Probable cause is a “practical, nontechnical conception” drawn from “common-sense conclusions about human behavior[.]” Illinois v. Gates, 462 U.S. 213, 231, 103 S.Ct. 2317, 76 L.Ed.2d 527 (1983) (first quoting Brinegar v. United States, 338 U.S. 160, 176, 69 S.Ct. 1302, 93 L.Ed. 1879 (1949); and then quoting *75 United States v. Cortez, 449 U.S. 411, 418, 101 S.Ct. 690, 66 L.Ed.2d 621 (1981)). It “deals with probabilities and depends on the totality of the circumstances,” Maryland v. Pringle, 540 U.S. 366, 371, 124 S.Ct. 795, 157 L.Ed.2d 769 (2003), and is “a fluid concept ... not readily, or even usefully, reduced to a neat set of legal rules,” Gates, 462 U.S. at 232, 103 S.Ct. 2317. Thus, the test for probable cause is not reducible to “precise definition or quantification.” Pringle, 540 U.S. at 371, 124 S.Ct. 795. That said, “[p]robable cause is synonymous with ‘fair probability.’ ” United States v. Jackson, 415 F.3d 88, 91 (D.C. Cir. 2005) (quoting Gates, 462 U.S. at 238, 103 S.Ct. 2317); see also United States v. Cardoza, 713 F.3d 656, 660 (D.C. Cir. 2013) (“[P]robable cause does not require certainty, or proof beyond a reasonable doubt, or proof by a preponderance of the evidence.”). At bottom, probable cause “is not a high bar.” Kaley v. United States, 571 U.S. 320, 338, 134 S.Ct. 1090, 188 L.Ed.2d 46 (2014).
For search warrants, probable cause requires (i) a “fair probability” that a crime has been committed and (ii) “a fair probability that contraband or evidence of [that] crime will be found in a particular place.” Gates, 462 U.S. at 238, 103 S.Ct. 2317; see also Andreen v. Lanier, 573 F. Supp. 2d 1, 5 n.3 (D.D.C. 2008) (“Probable cause requires only a probability or substantial chance that evidence may be found; it does not, by contrast, require absolute certainty.” (quoting United States v. Sidwell, 440 F.3d 865, 869 (7th Cir. 2006))). In other words, the core inquiry is whether the warrant application provides “a ‘substantial basis’ for concluding that ‘a search would uncover evidence of wrongdoing’ ” by “demonstrat[ing] cause to believe that ‘evidence is likely to be found at the place to be searched’ ” and “ ‘a nexus ... between the item to be seized and criminal behavior.’ ” United States v. Griffith, 867 F.3d 1265, 1271 (D.C. Cir. 2017) (first quoting Gates, 462 U.S. at 236, 103 S.Ct. 2317; then quoting Groh v. Ramirez, 540 U.S. 551, 568, 124 S.Ct. 1284, 157 L.Ed.2d 1068 (2004); and then quoting Warden, Md. Penitentiary v. Hayden, 387 U.S. 294, 307, 87 S.Ct. 1642, 18 L.Ed.2d 782 (1967)).
In addition, “the scope of the authorized search [must be] set out with particularity.” King, 563 U.S. at 459, 131 S.Ct. 1849. “The manifest purpose of th[e] particularity requirement was to prevent general searches.” Maryland v. Garrison, 480 U.S. 79, 84, 107 S.Ct. 1013, 94 L.Ed.2d 72 (1987). General warrants, widely reviled in colonial America, “le[ave] to the discretion of the executing officials the decision as to which persons should be arrested and which places should be searched.... [These warrants] provide[ ] no judicial check on the determination of the executing officials that the evidence available justified an intrusion into any particular [place].” Steagald v. United States, 451 U.S. 204, 220, 101 S.Ct. 1642, 68 L.Ed.2d 38 (1981). “By limiting the authorization to search to the specific areas and things for which there is probable cause to search,” the Supreme Court has said, “the [particularity] requirement ensures that the search will be carefully tailored to its justifications, and will not take on the character of the wide-ranging exploratory searches the Framers intended to prohibit.” Garrison, 480 U.S. at 84, 107 S.Ct. 1013.
Therefore, “[s]earch warrants must be specific.” United States v. Manafort, 313 F. Supp. 3d 213, 231 (D.D.C. 2018) (alteration in original) (quoting United States v. Hill, 459 F.3d 966, 973 (9th Cir. 2006)). There are two prongs of specificity: Particularity and breadth. “Particularity is the requirement that the warrant must clearly state what is sought. Breadth deals with the requirement that the scope *76 of the warrant be limited by the probable cause on which the warrant is based.” Id. at 231 (quoting Hill, 459 F.3d at 973). A warrant is not constitutionally overbroad so long as the time, location, and overall scope of the search are consistent with the probable cause set forth in the warrant application. See Google III, 497 F. Supp. 3d at 353. As to particularity, the warrant must “allow[ ] officers ‘to seize only evidence of a particular crime.’ ” Manafort, 313 F. Supp. 3d at 232 (quoting United States v. Young, 260 F. Supp. 3d 530, 546 (E.D. Va. 2017)). More, the Constitution “is not so exacting” as to require the “eliminat[ion of] all discretion of the officers executing the warrant.” United States v. Riley, 906 F.2d 841, 844–45 (2d Cir. 1990) (concluding that “the Fourth Amendment is not violated because the officers executing the warrant must exercise some minimal judgment as to whether a particular document falls within the described category”); see also United States v. Cobb, 970 F.3d 319, 332 (4th Cir. 2020) (noting that officers are allowed “some discretion” in “executing a search warrant, so long as the warrant at least minimally ‘confines the executing officers’ discretion by allowing them to seize only evidence of a particular crime’ ” (quoting United States v. Dickerson, 166 F.3d 667, 694 (4th Cir. 1999), rev'd on other grounds, 530 U.S. 428, 120 S.Ct. 2326, 147 L.Ed.2d 405 (2000))). “[E]xecuting agents must have as a practical matter ‘some discretion ... to interpret the words of every warrant no matter how particularly the items to be seized are described.’ ” United States v. Regan, 706 F. Supp. 1102, 1110 (S.D.N.Y. 1989) (second alteration in original) (quoting United States v. Marti, 421 F.2d 1263, 1268 (2d Cir. 1970)). In other words, a warrant must be “sufficiently specific to permit the rational exercise of judgment [by the executing officers] in selecting what items to seize.” United States v. LaChance, 788 F.2d 856, 874 (2d Cir. 1986) (quoting United States v. Vargas, 621 F.2d 54, 56 (2d Cir. 1980)). Importantly, the degree of particularity “turns on what was realistic or possible in this investigation.” See Archer v. Chisholm, 870 F.3d 603, 616 (7th Cir. 2017). Thus, “a broader sweep” can be lawful “when a reasonable investigation cannot produce a more particular description” of the things to be seized prior to obtaining and executing the warrant. United States v. Griffith, 867 F.3d 1265, 1276 (D.C. Cir. 2017) (citing Andresen v. Maryland, 427 U.S. 463, 480 n.10, 96 S.Ct. 2737, 49 L.Ed.2d 627 (1976)).
Ultimately, “[t]he touchstone of the Fourth Amendment is reasonableness, and the reasonableness of a search is determined ‘by assessing, on the one hand, the degree to which it intrudes upon an individual's privacy and, on the other, the degree to which it is needed to promote legitimate governmental interests.’ ” United States v. Knights, 534 U.S. 112, 118–19, 122 S.Ct. 587, 151 L.Ed.2d 497 (2001) (quoting Wyoming v. Houghton, 526 U.S. 295, 300, 119 S.Ct. 1297, 143 L.Ed.2d 408 (1999)); see also Maryland v. King, 569 U.S. 435, 448, 133 S.Ct. 1958, 186 L.Ed.2d 1 (2013) (“This application of ‘traditional standards of reasonableness’ requires a court to weigh ‘the promotion of legitimate governmental interests’ against ‘the degree to which [the search] intrudes upon an individual's privacy.’ ” (alteration in original) (quoting Houghton, 526 U.S. at 300, 119 S.Ct. 1297)); United States v. Montoya de Hernandez, 473 U.S. 531, 537, 105 S.Ct. 3304, 87 L.Ed.2d 381 (1985) (stating, “The Fourth Amendment commands that searches and seizures be reasonable,” which “depends upon all of the circumstances surrounding the search or seizure and the nature of the search or seizure itself,” and thus “[t]he permissibility of a particular law enforcement practice is judged by ‘balancing its intrusion on the individual's Fourth Amendment interests *77 against its promotion of legitimate governmental interests’ ” (quoting United States v. Villamonte-Marquez, 462 U.S. 579, 588, 103 S.Ct. 2573, 77 L.Ed.2d 22 (1983))). Reasonableness is assessed before the warrant is issued, not only in response to a motion to suppress. See Zurcher v. Stanford Daily, 436 U.S. 547, 565, 98 S.Ct. 1970, 56 L.Ed.2d 525 (1978) (recognizing that the “preconditions for a warrant” are “probable cause, specificity with respect to the place to be searched and the things to be seized, and overall reasonableness”); see also id. at 570, 98 S.Ct. 1970 (Powell, J. concurring) (stating that “the magistrate must judge the reasonableness of every warrant in light of the circumstances of the particular case, carefully considering the description of the evidence sought, the situation of the premises, and the position and interests of the owner or occupant”); Order at 4, In the Matter of the Search of 26 Digital Devices, No. 21-sw-233 (GMH) (Aug. 18, 2021) (filed under seal) (“[T]he Court finds that it has the authority—and, indeed, a constitutional duty—to evaluate the ‘overall reasonableness’ of the warrant the government seeks prior to its issuance.” (quoting Zurcher, 436 U.S. at 565, 98 S.Ct. 1970)).
III. DISCUSSION
The Court now applies these principles to the geofence warrant at hand.
A. Probable Cause
The probable cause analysis proceeds in two steps: (i) probable cause that a crime was committed; and (ii) probable cause that the search will uncover evidence of that crime. See United States v. Conley, 4 F.3d 1200, 1204–05 (3d Cir. 1993) (citing Gates, 462 U.S. at 236, 238–39, 103 S.Ct. 2317). Again, to safeguard the government's ongoing investigation in this case, all the Court will say on the first issue is that, based on the materials reviewed, the government has made the requisite showing that a federal crime occurred. See Google IV, 542 F.Supp.3d at 1155 (in assessing government's request for geofence warrant, declining to detail how the government established probable cause that a crime was committed when the criminal investigation remained ongoing).
Likewise, there is probable cause—a “fair probability”—that the search of Google's servers will uncover useful evidence—i.e., the identities of the suspects inside the [Redacted] center.[15] First, there is more than a “fair probability” that the suspects were within the geofence during the time windows the government established. The requested geofence encompasses the [Redacted] center and its parking lot. See Warrant Aff., Attach. A, at *78 4. The CCTV footage obtained by the government shows the suspects inside the [Redacted] center. See Warrant Aff., ¶¶ 35–36. Further, it is likely that the suspects accessed the [Redacted] center's customer service area from the adjoining parking lot. Although the warrant affidavit does not say whether the suspects were in the parking lot, magistrate judges may “draw such reasonable inferences as [they] will from the material supplied to [them] by applicants for a warrant.” Gates, 462 U.S. at 240, 103 S.Ct. 2317.
Second, the government has evidence that the suspects were actually using cell phones during the time windows set in the warrant. The CCTV footage apparently shows the suspects utilizing their devices while inside the [Redacted] center. See Warrant Aff., ¶ 35. This stands in stark contrast to other cases where courts have denied geofence warrant requests. In Google IV, for instance, the Court denied the government's geofence warrant application, noting that “the affidavit does not suggest that any relevant perpetrator or witness even had a smartphone.” 542 F.Supp.3d at 1157. Although in that case, like this one, the suspect “was caught on surveillance footage,” there was no “suggestion that the surveillance footage shows that the individual had a cell phone.” Id. Not so here, where the suspects are seen using their phones on the CCTV footage. In the Court's view, however, it is not necessary that the government actually know that suspects are using their phones within the geofenceSee Google III, 497 F. Supp. 3d at 355 (granting geofence warrant despite there being “no evidence in the affidavit that any of the suspects possessed cell phones or used cell phones in the commission of the offense”). The core inquiry here is probability, not certainty, and it is eminently reasonable to assume that criminals, like the rest of society, possess and use cell phones to go about their daily business. See id. at 356 (“Unlike virtually any other item, it is rare to search an individual in the modern age during the commission of a crime and not find a cell phone on the person.”); see also United States v. James, 3 F.4th 1102, 1105 (8th Cir. 2021) (“Even if nobody knew for sure whether the [suspect] actually possessed a cell phone, the judges were not required to check their common sense at the door and ignore the fact that most people ‘compulsively carry cell phones with them all the time.’ ” (quoting Carpenter, 138 S. Ct. at 2218)).
Third, the affidavit's failure to specifically allege that the suspects, while on their phones, were using applications or other features that would communicate location data to Google, is also not fatal to the warrant application. The probability that the phones were communicating location information to Google is, at the very least, “fair,” and that is all that is required. See Google III, 497 F. Supp. 3d at 355 (granting geofence warrant even though government did not present “any additional evidence that perpetrators or witnesses of the crime used Google applications or operating systems that would store location data”). Indeed, as the government has suggested here and in other cases, it would be the “relatively rare” case when a cell phone does not transmit location information to Google. Google II, 481 F. Supp. 3d at 734; see also Warrant Aff., ¶ 19. Roughly three-quarters of all phones worldwide contain Google's OS, and even those phones without Google's OS nonetheless have access to popular Google applications, the use of which can cause location information to be transmitted to Google.[16] See *79 Warrant Aff., ¶ 19. Further, Google has designed certain applications in such a way that makes them of limited functionality if users do not “sign in” with their Google account, which contains the user's identifying information. See id., ¶ 11. And even if only a third of Google OS users opt-in to the “Location History” service, that figure—which numbers in the “numerous tens of millions” of users—likely underestimates the volume of location information Google possesses, since (a) the government avers that Google collects location data even for users who have requested that such data not be gathered and (b) Google can collect location information from non-Google devices (e.g., iPhones) if those device users utilize Google accounts on those devices. See id., ¶ 19. Accordingly, “there is reasonable probability that a cellular phone, regardless of its make,” transmits location information to Google. Id. Thus, not only has the government demonstrated that the suspects were using cell phones within the geofence in the time windows described by the warrant, but there is a “fair probability” those phones communicated location information to Google.
Fourth, there is also a “fair probability” that Google is in possession of identifying information for the users of phones found within the geofence. Users of Google OS devices—which comprise 74 percent of the world's smartphone market—are prompted to sign-in to a Google account when they begin using the device, and that account contains identifying information on the user, including their name and possibly a physical address or other email accounts. See id., ¶¶ 10, 19, 20. The government avers that “[n]early every device using [Google's OS] has an associated Google account.” Id., ¶ 10. Additionally, even iPhone users (23 percent of the worldwide market) may sign-in to their Google accounts on their devices in order to access their Gmail accounts and get optimum functionality out of popular applications like Google Maps and YouTube. See id., ¶¶ 11, 19. Finally, because Google can “associate” location history data “with the relevant Google account,” the location data transmitted to Google can ultimately be traced back to a particular user. Id., ¶ 17. Although the ability to actually perform this tracing depends on a number of user decisions, including whether to sign-in to a Google account and activate the Location History service, the information presented here suggests there is, at minimum, a fair chance that tracing location information to a user can be achieved.
Thus, because there is a “fair probability” that (i) the suspects were inside the geofence, (ii) were using their cell phones inside the geofence, (iii) those phones communicated location information to Google, and (iv) Google can trace that information back to a particular device, accountholder, and/or subscriber, there is probable cause that the search will produce evidence useful to the government's investigation of the criminal activity in question.
B. Particularity
“The Fourth Amendment's particularity requirement has three components: A warrant ‘must identify the specific offense’ for which law enforcement has established probable cause; it must ‘describe the place to be searched’; and it must ‘specify the “items to be seized by their relation to designated crimes.” ’ ” Matter of Search of [Redacted] Washington, D.C., 317 F. Supp. 3d 523, 527 n.3 (D.D.C. 2018) (quoting *80 United States v. Williams, 592 F.3d 511, 519 (4th Cir. 2010)). Here, the government has satisfied each of those requirements.
The government identified the offenses for which probable cause has been established in its warrant application materials. See Warrant Aff., Attach. B, at 6. The Court will not list those offenses to avoid compromising the government's ongoing investigation.
The government has also appropriately described the things to be seized in the search of Google's location data. Here, the warrant sets forth eight distinct categories of information “that constitutes evidence of violations of [the crimes under investigation].” Warrant Aff., Attach. B, at 6. This construction is constitutionally permissible. See, e.g., United States v. Bundy, 195 F. Supp. 3d 1170, 1175–76 (D. Or. 2016) (denying a motion to suppress warrant using the “information that constitutes evidence” construction, finding the warrant “not overbroad”); United States v. Ciancia, No. CR13-902, 2015 WL 13798666, at *7 (C.D. Cal. Sept. 24, 2015) (denying a motion to suppress and finding the warrant that permitted the government to seize “all information ... that constitutes evidence, contraband, fruits, or instrumentalities of violations of [the crime at issue]” left the executing officers with no discretion (first alteration in original)). The categories, which limit the seizure of data and information to evidence of the designated crimes, “enable the executing officer to ascertain and identify with reasonable certainty those items authorized to be searched and seized.” United States v. Ray, 541 F.Supp.3d 355, 377 (S.D.N.Y. 2021) (citing Groh, 540 U.S. at 557–59, 124 S.Ct. 1284); see also United States v. Burke, 633 F.3d 984, 992 (10th Cir. 2011) (finding a warrant sufficiently particularized where it authorized law enforcement to search and seize “contraband, evidence, fruits, or instrumentalities of [certain] crime(s),” and then listed three categories of information); United States v. Jacobson, 4 F. Supp. 3d 515, 524 (E.D.N.Y. 2014) (finding warrants were sufficiently particularized “[b]ecause the [warrants] referenced particular crimes and used illustrative lists as a means of limiting the items to be seized”); United States v. Reed, No. 2:13-CR-29-1, 2013 WL 5503691, *4 (D. Vt. 2013) (“The warrant in this case incorporated a list of evidence to be seized that included a description of the crime ... [;] the warrant in this case [thus] provided adequate guidance limiting the scope of the search for evidence of a specific crime.”). Suffice to say, the government has provided far more specificity here than its agents did in Google I. There, the warrant merely sought “ ‘evidence or instrumentalities of’ the listed offenses,” but the government provided “nothing more” beyond that description. Google I, 2020 WL 5491763, at *3. This language was “completely devoid of any meaningful limitation,” and the court denied the warrant application for that reason, among others.[17] Id.
Finally, the government has satisfied the particularity requirement as to the place to be searched because, as discussed below, it has appropriately contoured the temporal and geographic windows in which it is seeking location data. That is, the government has limited the place to be searched in time and location, and its warrant application *81 is not otherwise overly-broad, but is “confined to the breadth of the probable cause that supports it.” United States v. Thorne, 548 F.Supp.3d 70, 94 (D.D.C. 2021); see also Google III, 497 F. Supp. 3d at 353.
1. Time
Here, the government seeks a total of 185 minutes of geofence data on 8 days across a five-and-a-half month period. See Warrant Aff., ¶ 36. The individual time periods range from 2 to 27 minutes, and most are for time windows falling within the afternoon between 1 p.m. and 4 p.m. See id. Although a total of 185 minutes of geofence data is more than the Google III court sanctioned, the government's request in this case is reasonable. See 497 F. Supp. 3d at 352–53 (authorizing geofence warrant that sought 139 minutes of location data). To begin, while 3-plus hours of geofence data may sound significant, it is 3 hours spread out over almost 6 months. That hardly “provides [the] all-encompassing record of the [phone] holder's whereabouts” that drew the Supreme Court's concern in Carpenter, 138 S. Ct. at 2217. Rather, the geofence only provides cell phone users’ whereabouts in a single area for a handful of minutes on the days in question, not the sum-total of their daily movements. Thus, viewed in proper context, the government's request is limited and reasonable. Cf. Jones, 565 U.S. at 430, 132 S.Ct. 945 (Alito, J., concurring) (“[R]elatively short-term monitoring of a person's movements on public streets accords with expectations of privacy that our society has recognized as reasonable.”).
Further, the time windows requested by the government are closely keyed to the periods during which the suspects were inside the [Redacted] center. The government has determined which time segments to request by reviewing the CCTV footage and determining, based on the CCTV timestamps, the approximate times at which the suspects were in the [Redacted] center.[18] See Warrant Aff., ¶ 36. “Thus, the warrant does not seek location data for days or even hours to track the whereabouts of the perpetrators, but rather location data that is tailored and specific to the time of the [alleged crimes] only.” Google III, 497 F. Supp. 3d at 357.
The government's geofence warrant application here is altogether different from the application presented in Google IV. There, the government requested only an hour of geofence data, but failed to narrow the timeframe to only the points at which evidence of a crime could reasonably be found. See Google IV, 542 F.Supp.3d at 1158. Similar to this case, the government in Google IV had surveillance footage showing the suspect at three distinct times. See id. Yet, instead of tailoring the warrant to request geofence data for only the approximate times at which the suspect appeared on the footage, the government demanded geofence data for “just before the second sighting to approximately 10 minutes after the [third sighting].” Id. The Google IV court questioned why “data for the entire period between the *82 second and third sighting” was sought when the suspect was totally out of sight, and found the warrant not sufficiently particularized. Id. In this case, the government avoided this infirmity by only seeking geofence data for the approximate times during which the government's investigation showed the suspects were in the [Redacted] center.
2. Location
The geographic scope of the government's requested geofence warrant also passes constitutional muster. The inquiry here is whether the “target locations [are] drawn to capture location data from locations at or closely associated with the [crime].” Google III, 497 F. Supp. 3d at 358. The government has carried that burden.
The requested geofence covers a portion of the front half of the [Redacted] center and its parking lot. See Warrant Application, Attach. A, at 4. So, it encompasses only the location of the suspects (i.e., the [Redacted] center) and an area “closely associated with the” location of the suspects (i.e., the parking lot outside the entrance to the [Redacted] center). It is reasonable to assume that, in order to access the [Redacted] center—which is located in an industrial area—the suspects drove and parked their vehicles in the parking lot in front of the [Redacted] center, or at least walked through it. Thus, for each of areas covered by the geofence, “there is a fair probability that the location data of perpetrators, co-conspirators and witnesses to the incidents will be uncovered.” Google III, 497 F. Supp. 3d at 358.
3. Overbreadth
Overall, the geofence warrant sought by the government is appropriately limited in scope—that is, it is not constitutionally overbroad. “[T]he proper scope of a warrant is confined to the breadth of the probable cause that supports it.” Thorne, 548 F.Supp.3d at 94. Here, the duration and location of the requested geofence closely track the probable cause presented in the government's warrant application.
As to duration, the government established that the suspects were inside the [Redacted] center for limited periods of time, and therefore it has cabined its geofence request to those approximate time periods. See Warrant Aff., ¶¶ 36–37. As to location, the government has established probable cause that the suspects were inside the [Redacted] center. See id., ¶ 35. Accordingly, the boundaries of the geofence include only that location and the parking lot directly adjoining the [Redacted] center. True, the geofence may also capture the location information for persons who are not suspects, namely the other customers inside the [Redacted] center. According to the government, the surveillance video shows that, on those occasions when the suspects are not alone in the [Redacted] center, there are, on average, two to three other customers in the [Redacted] center with the suspects. See id., ¶ 38. For several reasons, the warrant's potential to collect location information from these other individuals does not render it deficient.
As an initial matter, constitutionally permissible searches may infringe on the privacy interests of third persons—that is, persons who are not suspected of engaging in criminal activity. The Supreme Court has long recognized and accepted that third party privacy interests could be impacted by lawful searches. In Zurcher, the Court was asked to invalidate a search warrant on the basis that the “possessor of the premises to be searched” was not suspected of criminal activity, and therefore any search would violate the possessor's reasonable expectations of privacy. 436 U.S. at 550, 98 S.Ct. 1970. The *83 Court, unconvinced that a “new rule denying search warrants against third parties ... would substantially further privacy interests without seriously undermining law enforcement efforts,” rejected this rationale. Id. at 560, 98 S.Ct. 1970. As the majority observed, the “Fourth Amendment has itself struck the balance between privacy and public need, and there is no occasion or justification for a court to revise the Amendment and strike a new balance.” Id. at 559, 98 S.Ct. 1970. Instead, the Court found that “where the State does not seek to seize ‘persons’ but only those ‘things’ which there is probable cause to believe are located on the place to be searched, there is no apparent basis in the language of the [Fourth] Amendment” to require “probable cause to believe that the third party is implicated in the crime.”[19] Id. at 554, 98 S.Ct. 1970. Zurcher thus elevated “the fundamental public interest in implementing the criminal law” above the privacy interests which could be *84 indirectly impacted by a legal search backed by probable cause. Id. at 560–61, 98 S.Ct. 1970.
Against this backdrop—and given the often inherently intrusive task that is evidence gathering, even when performed lawfully by the police—it is neither novel nor surprising that reasonable searches intrude on the privacy interests of individuals who are not the target of criminal investigation. See Google III, 497 F. Supp. 3d at 361 (“[T]he fact that one uninvolved individual's privacy rights are indirectly impacted by a search is present in numerous other situations and is not unusual.”). Indeed, for nearly every suspect's text or email account lawfully seized, or house searched, there are frequently other “innocent” and “uninvolved” persons whose privacy is compromised. See, e.g., United States v. Bowen, No. CR-18-01013-001, 2019 WL 2462793, at *2 (D. Ariz. June 13, 2019) (denying in part a motion to suppress text messages in which the defendant communicated information concerning alleged wrongdoing “with other individuals”); United States v. Ali, 870 F. Supp. 2d 10, 39–40 (D.D.C. 2012) (denying a motion to suppress emails even though the government may have “seized emails that were ... not pertinent to the subject matters specified in [the] warrants”); United States v. Fuentez, No. CR 19-37, 2019 WL 6311885, at *1 (D.D.C. Nov. 25, 2019) (denying a motion to suppress items recovered during the search of defendant's “two-story townhome ... where [he] lived with his family, including minor children”); see also Google III, 397 F. Supp. 3d at 361. Consider also a search of a business’ filing cabinet or of a home office, where “it is certain that some innocuous documents” which may belong to persons other than the target of the investigation “will be examined, at least cursorily, in order to determine whether they are, in fact, among those papers authorized to be seized.” Andresen, 427 U.S. at 482 n.11, 96 S.Ct. 2737. And yet, such searches are routinely permitted. See, e.g., United States v. Oloyede, 982 F.2d 133, 139 (4th Cir. 1992) (finding that the government had probable cause to search and seize “all of [the defendant's] business and client records”); United States v. Brown, No. 19-CR-110, 2019 WL 7838276, at *7 (D. Minn. Sept. 20, 2019) (recommending denial of a motion to suppress evidence recovered pursuant to a search warrant executed on the defendant's business, during which “client files, computers, and various business records” were seized), report and recommendation adopted, 2019 WL 6607240 (D. Minn. Dec. 5, 2019); United States v. Barber, No. 4:12Cr17, 2018 WL 11187587, at *3 (N.D. Fla. Aug. 7, 2018) (noting that “agents executed a search warrant at one of [the defendant's] offices and seized 55 boxes of client files and business records”), report and recommendation adopted, 2018 WL 4291733 (N.D. Fla. Sept. 7, 2018). The Fourth Amendment was not enacted to squelch reasonable investigative techniques because of the likelihood—or even certainty—that the privacy interests of third parties uninvolved in criminal activity would be implicated. See Google III, 497 F. Supp. 3d at 362 (“The proper line of inquiry is not whether a search of location data could impact even one uninvolved person's privacy interest, but rather the reasonableness of the search, the probability of finding evidence at the location, and the particularity of the search request.”). Rather, the Fourth Amendment seeks to ensure that privacy interests are not infringed by law enforcement activities without a showing of probable cause and a particularized description of the place to be searched and the things to be seized. See Zurcher, 436 U.S. at 565, 98 S.Ct. 1970.
*85 The warrant before the Court meets that standard. It may be the case that the requested geofence, when considering its margin of error, will capture the location information for other customers inside the [Redacted] center or motorists merely driving by the [Redacted] center on the abutting road or an employee in the adjoining business during the requested time segments. The possibility of capturing those persons’ location information is not fatal to the warrant's constitutionality. “The particularity inquiry turns on what [is] realistic or possible in this investigation,” Archer, 870 F.3d at 616, and in this case it appears physically impossible for the government to have constructed its geofence to exclude everyone but the suspects. See Google III, 497 F. Supp. 3d at 362 (“[I]n the geofence context, there is no way to exclude the possibility that at any given time, a delivery truck may drop off a parcel within the geofence location.” (emphasis added)). As the D.C. Circuit has counseled, in situations “when a reasonable investigation cannot produce a more particular description” of the place to be searched or the things to be seized, courts “may allow a broader sweep.” Griffith, 867 F.3d at 1276. This is one those cases.
More, the potential infringement of third-party privacy interests is modest in this case. The government represents that, in the time periods for which it is seeking information from Google, the suspects are either in the [Redacted] center alone, or accompanied by (on average) two or three other customers. See Warrant Aff., ¶ 38. Further, as explained, the geofence, as drawn by the government, falls within an industrial area and does not encompass residences or other particularly sensitive locations.
That said, the [Redacted] center shares a building with another business, and therefore a device in that business may fall within the margin of error. So, too, may a driver traveling on the road abutting the [Redacted] center's parking lot. But in contrast to the other published cases on geofence warrants, the request for location information here does not have the potential of sweeping up the location data of a substantial number of uninvolved persons. In Google I, the requested geofence covered nearly eight acres of land “in a congested urban area” and included “residences, businesses, and healthcare providers.” 2020 WL 5491763, at *5 (footnote omitted); see also Google II, 481 F. Supp. 3d at 743 (describing the same area as “a busy commercial and residential area on a major arterial street in a major U.S. city”). The court concluded that “the vast majority of cellular telephones likely to be identified in this geofence will have nothing whatsoever to do with the offenses under investigation,” and therefore found the warrant insufficiently particularized. Google I, 2020 WL 5491763, at *5. Similarly, in Google IV, within the requested geofence was another business in the same building as the target business and “two public streets.” 542 F.Supp.3d at 1158. Further, residences and other businesses were located within the margin of error. Id. In these cases, the warrant applications were denied.
These overbreadth concerns are not present here. The government's proposed geofence appears less than an acre in size and covers only the [Redacted] center and its parking lot. Unlike in Google I, the geofence drawn here is located in an industrial area, not a “congested urban area,” and no residences can be seen within the geofenceSee Warrant Aff., ¶ 38 & Attach. A at 4. In addition, the geofence has been constructed in such a way as to exclude both the business sharing the same building as the [Redacted] center and the abutting road, although both could *86 fall within the margin of error. See Warrant Aff., ¶ 38 & Attach. A at 4. While that road, like those in Google IV, is a “public street[ ],” the government emphasizes that the road is a “secondary road.” Warrant Aff., ¶ 38. The road is therefore not the “major arterial street” that concerned the courts in Google I and IIGoogle II, 481 F. Supp. 3d at 743. And though the government's requested time windows do not cover periods in the dead of night, which the Google III court noted as another limit on the warrant's breadth, the limited size of the geofence here is more than sufficient to narrow its scope. See Google III, 497 F. Supp. 3d at 358. In this case, most of the time periods sought are in the early to mid-afternoon—during normal business hours—but the target area is small and lightly trafficked enough to render the search reasonable.[20],[21] See Warrant Aff., ¶ 36.
*87 Moreover, any overbreadth concerns raised by the requested geofence are further addressed by the warrant's two-step search procedure, which ensures identifying information associated with devices found within the geofence will be produced only pursuant to a further directive from the Court. Similar protocols—albeit less restrictive than the one imposed here—have been met with some skepticism by courts to have considered them in the context of geofence warrants. In the only published decision granting an application for such a warrant, the Google III court upheld a two-step protocol whereby Google would first furnish the government with an anonymized list of devices found within the geofenceSee Google III, 497 F. Supp. 3d at 353. Then, at step two, the government could, “at its discretion,” request from Google “account identifier and subscriber information” of any device found within the geofence which Google would then provide. Id. In considering that two-step procedure, the court found that the exercise of discretion by law enforcement to identify users of devices found within the geofence was immaterial because the government had established probable cause to seize all the location data found within the geofence. Id. at 362 (“Whether [the government] chooses to obtain all that information, or partial information, is of no matter to the Court's consideration of the constitutionality of the warrant under the Fourth Amendment.”). Yet this same two-step process was criticized in earlier cases for giving “law enforcement agents unbridled discretion to obtain identifying information about each device detected in the geofences.”[22] Google II, 481 F. Supp. 3d at 746; see also *88 Google I, 2020 WL 5491763, at *6 (“There is no objective measure that limits the agents’ discretion in obtaining information as to each cellular telephone in the geofence.”). In Google II, the government even removed the second step of the process—i.e., the step at which it would request identifying information from Google—but admitted it could simply replace that second step with a subpoena to Google ordering it to identify the accounts revealed in step one. Unsurprisingly, the court rejected that approach, too. 481 F. Supp. 3d at 747–48.
In this case, the government initially proposed a search protocol virtually identical to the one sanctioned in Google III and criticized in Google I and II: At step one, Google would produce an anonymized list of devices found within the geofence during the requested time segments, which the government would review. At step two, law enforcement would request from Google identifying information about devices on the list “at its discretion.” Because this warrant proposed no rule or condition that would adequately cabin that exercise of discretion by law enforcement, the Court rejected it. See, e.g., United States v. Zemlyansky, 945 F. Supp. 2d 438, 453 (S.D.N.Y. 2013) (“Courts implement the particularity requirement by insisting that warrants not ‘leave to the unguided discretion of the officers executing the warrant the decision as to what items may be seized.’ ” (quoting Riley, 906 F.2d at 844)); see also United States v. Triumph Cap. Grp., Inc., 211 F.R.D. 31, 57 (D. Conn. 2002) (“A warrant only needs to be specific enough to permit the executing officer to exercise reasonable, rational and informed discretion and judgement in selecting what should be seized.”). The warrant application which the Court granted, on the other hand, eliminated law enforcement's discretion at step two by requiring it to return to the Court and justify any device deanonymization based on its review of the anonymized information provided by Google and other evidence in the case.[23] See Warrant Aff., ¶ 37(d). This second step serves two functions. First, it delimits the government's discretion in the search and seizure process. The government is free to choose the devices for which it seeks identifying information, but the ultimate decision as to which subscribers, if any, Google will be compelled to identify lies with the Court.[24] See *89 id. at ¶ 37(e). The second step also ensures that the government's search is particularized; that is, before any identifying information is disclosed to the government, it must justify the specific devices for which it seeks that information, consistent with its showing of probable cause.[25]
The two-step process also ameliorates possible overbreadth concerns. Even if third-party devices appear within the government's requested geofence, those persons’ location information will be anonymized at step one.[26] See Warrant Aff., *90 ¶ 37(b). And assuming the government is able to filter out devices at step two “not likely to be relevant to the investigation”—as it says it can do here—these persons’ location data will likely remain anonymized. Warrant Aff., ¶ 37(c). Alternatively, the two-step process can also address a geofence that returns a larger number of anonymized devices than expected, and which cannot be further narrowed by law enforcement to make the result useful. In such a case, assuming law enforcement seeks to deanonymize any devices at all, additional justification for such request can be required. In other words, the two-step process can serve as a court-supervised filter, winnowing geofence results that are unlikely to provide useful evidence to the government, thereby further minimizing third-party privacy concerns.
In sum, the government has drawn “narrowly tailored geofence zones for a sufficiently limited amount of time ... and minimize[d] through that zone design and subsequent investigation [through the two-step protocol] the possibility of sweeping in large amounts of location data for uninvolved individuals.” Google III, 497 F. Supp. 3d at 363; see also Order at 22, In re Search of Information Associated with Cell Towers (filed under seal) (granting tower dump warrant where “[t]he government has carefully tailored the warrants to the greatest degree possible to obtain cell phone data ... to assist in identifying the” suspect). Thus, the scope of the warrant is constitutionally permissible.
CONCLUSION
Given that most of us have our cell phones by our side most of the time, their power to constantly monitor their own location has essentially turned them into highly-accurate, personal tracking devices, identifying our location wherever we are in the world to within 20 meters. For better or worse, this feature that makes mobile devices useful to us—that gives us directions when we are lost and helps us find our phones when they go missing—leaves a digital trail behind us, the disclosure of which many would consider an invasion of privacy. Unsurprisingly, our phones’ ability to trace our steps has also proven of great value to law enforcement when trying to determine who was at a particular location when criminal activity occurred. It is that tension between privacy concerns and legitimate law enforcement interests the Fourth Amendment seeks to address, and which warrants careful consideration of constitutional requirements when assessing the lawfulness of geofence applications. These competing interests are assessed in recognition of the Fourth Amendment's prohibition of only “unreasonable” searches and seizures.
Applying those Fourth Amendment principles here, the government has established probable cause that criminal activity, which it has identified, occurred within the proposed geofences, and that evidence related to that activity, which it has specified, will be found within them. Further, it has carefully limited the scope of the geofences to the approximate location where *91 the criminal activity occurred and timeframes during which it took place. The fact that the geofences may reveal the location information of non-suspects does not render the government's warrant unreasonable given that the geofences and the two-step protocol have been crafted to minimize privacy concerns to the greatest degree possible while also preserving “the fundamental public interest in implementing the criminal law.” Zurcher, 436 U.S. at 560, 98 S.Ct. 1970.
Accordingly, for all these reasons, the government's warrant application is GRANTED.

Footnotes

The Court granted the government's request to seal the warrant application because the criminal investigation is not public and revealing the existence of the warrant could adversely impact the government's investigation, including by causing the subjects of the investigation to flee or destroy evidence. See Motion to Seal Warrant and Related Documents and To Require Non-Disclosure Under 18 U.S.C. § 2705(b), at 5. Accordingly, the public version of this memorandum opinion will not disclose facts that may identify the government's investigation or the targets of it. Further, the Court has provided the government with an opportunity to review this opinion and redact from the public version material sensitive to its investigation.
See Matter of Search of Info. Stored at Premises Controlled by Google, as further described in Attachment A, No. 20 M 297, 2020 WL 5491763 (N.D. Ill. July 8, 2020) (“Google I”) (denying warrant application); Matter of Search of Info. Stored at Premises Controlled by Google, 481 F. Supp. 3d 730 (N.D. Ill. 2020) (“Google II”) (denying warrant application); Matter of Search Warrant Application for Geofence Location Data Stored at Google Concerning an Arson Investigation, 497 F. Supp. 3d 345 (N.D. Ill. 2020) (“Google III”) (granting warrant application); Matter of Search of Info. that is Stored at Premises Controlled by Google, LLC, 542 F.Supp.3d 1153 (D. Kan. 2021) (“Google IV”) (denying warrant application).
Geofence issues have also been litigated at length in an ongoing criminal matter in the Eastern District of Virginia. See United States v. Chatrie, No. 3:19-cr-00130 (E.D. Va.). In Chatrie, authorities arrested the defendant after determining, through information gathered via a geofence warrant, that his cell phone was in the vicinity of a bank when it was robbed. See Government's Response in Opposition to Defendant's Motion for Suppression of Evidence Obtained Pursuant to the Google Geofence Warrant at 4–5, Chatrie, No. 19-cr-130 (E.D. Va. Nov. 19, 2019), ECF No. 41. The defendant filed a motion to suppress the evidence gathered from the geofence in October 2019. See Defendant's Motion to Suppress Evidence from a Geofence “General” Warrant, Chatrie, No. 19-cr-130 (E.D. Va. Oct. 29, 2019), ECF No. 29. Since then, the district court has held three hearings on the motion—including a two-day evidentiary hearing in March 2021—and the parties have engaged in successive rounds of supplemental briefing. See Chatrie, No. 19-cr-130 (E.D. Va.), ECF Nos. 104, 109, 135, 139, 205, 214. Further, Google submitted an amicus brief and several declarations in the case that articulate its views on geofence warrants and describe how Google responds to them. See Chatrie, No. 19-cr-130 (E.D. Va.), ECF Nos. 59-1, 96, 96-1, 96-2, 110-1. Though the court in Chatrie has not yet ruled on the defendant's motion to suppress, this Court has reviewed the documents filed on the public docket, and will refer to them to the extent they assist in further illuminating generally the technology on which geofence warrants are based.
See Zach Whittaker, “Google says geofence warrants makeup nearly one-quarter of all US demands,” TechCrunch (Aug. 19, 2021) (noting that “Google received 982 geofence warrants in 2018, 8,396 in 2019 and 11,554 in 2020”), available at https://techcrunch.com/2021/08/19/google-geofence-warrants/; “Supplemental Information on Geofence Warrants in the United States”, GOOGLE (stating that geofence warrants “constitute[e] more than 25% of all warrants [Google] receive[s] in the United States”), available at https://s3.documentcloud.org/documents/21046081/google-geofence-warrants.pdf (last visited Dec. 2, 2021). In an amicus brief submitted in the Chatrie case, Google said it had received 15-times more geofence warrant requests in 2018 than it did in 2017, and five-times more in 2019 than it did in 2018. See Brief for Google as Amicus Curiae at 8, Chatrie, No. 3:19-cr-00130 (E.D. Va. May 22, 2020), ECF No. 59-1 (“Google Amicus Brief”).
Geolocation coordinates are synonymous with latitude and longitude coordinates. Latitude and longitude coordinates provide a system for pinpointing exact locations on Earth's surface. For instance, the latitude and longitude of Capitol One Arena (located at 601 F Street NW in Washington, D.C.) is 38° 53’ 50.784” N, 77° 1’ 15.24” W.
For instance, by enabling Bluetooth, Google OS phone users allow the device to “scan[ ] its environment to identify Bluetooth devices.” Warrant Aff., ¶ 8. Those Bluetooth devices then communicate their “identifier code” to the phone, which then sends it to Google. See David Yanofsky, “Google can still use Bluetooth to track your Android phone when Bluetooth is turned off”, Quartz (Jan. 27, 2018) (explaining that Android devices (which are made by Google and contain Google OS) “use Bluetooth to collect location-related data and transmit that data to Google. [An Android device] does this by sending Google, among other things, the unique identifier codes of Bluetooth broadcasting devices it encounters. Such devices, known as beacons, are often used in stores, museums, and other public places to help phones ascertain their locations within buildings.”), available at https://qz.com/1169760/phone-data/. Those “identifier codes” correspond to a particular Bluetooth beacon in a particular location. The phone sends the identifier codes to Google, which can then determine where the Bluetooth beacon—and, by extension, the phone—is located. See id. (“When either Bluetooth or Bluetooth scanning is enabled, a report containing a list of nearby Bluetooth beacons is sent to Google any time an app refreshes Android location services.”); id. (explaining that “when an Android phone sends Google a unique Bluetooth identifier, the company can not only track you around town, but also into a mall, through a store, and up to a specific rack of clothes”).
In a suppression hearing in the Chatrie case, a government witness testified that “Google location records in a normal interval occur approximately every two minutes.” Suppression Hearing Transcript at 236, Chatrie, No. 19-cr-130 (E.D. Va. Mar. 5, 2021), ECF No. 202.
Google estimates that even one-third of its account users numbers in the “numerous tens of millions.” McGriff Decl. ¶ 13, Chatrie, No. 19-cr-130 (E.D. Va. Mar. 11, 2020), ECF No. 96-1.
The Associated Press reported in 2018 “many Google services on Android devices and iPhones store ... location data even if [the user activated] a privacy setting that says it will prevent Google from doing so.” Ryan Nakashima, “AP Exclusive: Google tracks your movements, like it or not”, The Associated Press (Aug. 13, 2018), available at https://apnews.com/article/north-america-science-technology-business-ap-top-news-828aefab64d4411bac2... (finding that “[e]ven with Location History paused, some Google apps automatically store time-stamped location data without asking”). For instance, according to the Associated Press’ reporting, Google Maps, web searches, and weather updates can all send information on a device's location to Google despite the device user having “paused” Location History. See id. Notably, several days after the Associated Press’ reporting, Google confirmed that even with Location History “paused,” “some location data may be saved as part of your activity on other services, like Search and Maps.” Ryan Nakashima, “AP NewsBreak: Google clarifies location-tracking policy,” The Associated Press (Aug. 16, 2018), available at https://apnews.com/ef95c6a91eeb4d8e9dda9cad887bf211/APNewsBreak:-Google-clarifies-location-tracking-....
Similar to Google device users, Apple device users who utilize Google applications can also “opt out” of location tracking. See, e.g., Dave Johnson, “How to stop Google from tracking your iPhone's location”, Insider (Nov. 24, 2020), available at https://www.businessinsider.com/how-to-stop-google-tracking-iphone.
In Google I, “the government assert[ed] that approximately 97% of smartphones in the world use Google applications or Google's operating system.” 2020 WL 5491763, at *3.
For reference, an acre of land is 4,046.9 square meters.
In a prior version of its warrant application, the government's targeted area was in the form of a circle that centered on the [Redacted] center. The circle had a radius of approximately 35 meters and appeared to capture part of the road abutting the [Redacted] center. Further, the circle also included part of a building behind the [Redacted] center, as well as the business that shares the [Redacted] center's building. Following discussions with the Court, the government further limited the scope of the geofence to exclude these areas in which it had no evidence that criminal activity occurred. While the initial circle was small—less than an acre in size—the more restricted triangular geofence area the government included in its revised warrant application appears smaller still.
The government has not alleged whether the suspects were using phones with Google OS, or non-Google OS phones like iPhones. However, the government indicates the suspects are from a foreign country, and Google's phone market share outside the United States is 74 percent. See Warrant Aff., ¶¶ 19, 30.
It is possible that the government's collection of location information, particularly location data covering, as here, only short periods of time, would not be deemed a “search” requiring a warrant. See Carpenter v. United States, ––– U.S. ––––, 138 S. Ct. 2206, 2234, 201 L.Ed.2d 507 (2018) (Kennedy, J., dissenting) (noting the majority's “suggest[ion] that less than seven days of [CSLI] may not require a warrant”); United States v. Jones, 565 U.S. 400, 430, 132 S.Ct. 945, 181 L.Ed.2d 911 (2012) (Alito, J., concurring) (explaining that only when the government “secretly monitor[s] and catalogue[s] every single movement of an” individual's car “for a very long period” does the surveillance become a search); see also United States v. Hammond, 996 F.3d 374, 387–92 (7th Cir. 2021) (holding that real-time tracking of a specified cell phone over a period of approximately six hours was not a search).
In the Chatrie case, Google has maintained that, irrespective of the constitutional analysis, the Stored Communications Act requires the government to seek a warrant in order to obtain its location information. See Google Amicus Brief at 19–23. Here, as the government has applied for a warrant, thus conceding, if only for the purposes of this application, that a warrant is required, the Court need not take a position on either issue.
Here, it is possible the government could have sought identifying information for the other customers inside the [Redacted] center as potential witnesses to alleged criminal activity. See Google III, 497 F. Supp. 3d at 362 (“The government is entitled to search for evidence of the crime pursuant to a valid warrant and that evidence includes the identity of witnesses to the offense.”). For instance, it is possible that these persons could corroborate the CCTV footage and place the suspects inside the [Redacted] center. Had the government properly articulated a basis for these persons’ identifying information, the Court may not have required additional legal process to deanonymize the location data. See infra note 24. In other words, in the proper case, such as when “there is a fair probability that almost all location data retrieved will be for individuals who are either the perpetrators, co-conspirators, or witnesses to the crime,” additional legal process—whether in the form of a warrant, order, or other court directive—may not be required. Id. at 353 (emphasis added). Of course, the government is not precluded from seeking the identities of witnesses if it returns to the Court with a request to deanonymize the location data it received from Google at step one of the search protocol.
In this case, the government has indicated that the suspects are from a foreign country, increasingly the likelihood that the suspects’ phones have a Google OS. See Warrant Aff., ¶ 19 (explaining that Google's share of the worldwide cell phone market is 74 percent, while its share of the U.S. market is 46 percent).
It is notable that, in response to the Google I court's concern about the general and vague description of the “things to be seized,” the government “offer[ed] a modification to the search warrant that spell[ed] out in greater detail the type of information that agents could seek from Google.” 2020 WL 5491763, at *7. That is what the government did in the first instance here.
The Court is not troubled by the government's request for “a short time range for the location data,” rather than data for “an exact time at the location.” Warrant Aff., ¶ 37. Because the timestamps on the CCTV footage were “manually set,” the time reflected in the CCTV footage may have been off slightly, necessitating a broader time request. Id. Such approximation does not undermine the validity of the warrant. See, e.g., Google III, 497 F. Supp. 3d at 357 (authorizing geofence warrant where government's requested timeframes were “approximate ... based on the government's investigation”); see also Archer, 870 F.3d at 616 (“The particularity inquiry turns on what [is] realistic or possible in this investigation.”).
Some of the public criticism directed toward geofence warrants is that they fall into the category of so-called “reverse warrants,” which do not seek information on a particular suspect, but rather seek information that would lead to the identity of a suspect, compromising third parties’ privacy interests in the process. But the Fourth Amendment does not and has never required that law enforcement know a suspect's identity for certain or even have a suspect in mind to obtain a search warrant. Although law enforcement will often have identified a suspect or group of potential suspects before seeking warrants, many cases remain “whodunnits” at the time the government begins to seek warrants. Indeed, that was the case in Zurcher: There, the Supreme Court approved a search warrant that authorized seizure from a newspaper of photographs of unidentified individuals who had assaulted police officers. See 436 U.S. at 555, 98 S.Ct. 1970.
Said another way, a suspect's identity is not a prerequisite to a search warrant, which itself can be lawfully used to determine who a suspect is or develop a group of potential suspects. In fact, that is the often entire purpose of warrants seeking cell-site location information (“CSLI”) via tower dumps. Such warrants are routinely upheld. See, e.g., Order at 22, In re Search of Information Associated with Cell Towers, Nos. 21-sc-59, et al. (D.D.C. Jan. 17, 2021) (Howell, C.J.) (filed under seal) (granting tower dump warrant whose purpose was to uncover the identity of a suspect). A number of other courts have also acknowledged, directly and implicitly, that Fourth Amendment principles do not require that law enforcement have identified a suspect before a search warrant is issued. See, e.g., Google III, 497 F. Supp. 3d at 351 (granting a geofence warrant that sought “evidence pertaining to the identity of the [crime] suspects and their co-conspirators”); United States v. Clark, No. 17-CR-34, 2018 WL 11240486, at *3 (W.D. Wis. Apr. 20, 2018) (denying the defendant's motion to suppress evidence gathered pursuant to a search warrant whose affidavit did not identify the suspect and finding that “the suspect's identity ... was not material to the probable cause analysis”), aff'd in part, rev'd in part on other grounds and remanded, 935 F.3d 558 (7th Cir. 2019); United States v. Darby, 190 F. Supp. 3d 520, 526 (E.D. Va. 2016) (denying a motion to suppress evidence gathered pursuant to a search warrant whose aim was to identify suspects), aff'd, 721 F. App'x 304 (4th Cir. 2018); see also United States v. Williams, 616 F.3d 760, 765 (8th Cir. 2010) (noting that, for purposes of an arrest warrant, “[p]robable cause does not require certainty regarding [a suspect's] identity”); Avila v. Dailey, 246 F. Supp. 3d 347, 353 (D.D.C. 2017) (noting that a search whose purpose was to gather “evidence that would assist [the authorities] in identifying and locating the suspect” was lawful), on reconsideration in part, No. 15-CV-2135, 2017 WL 9496067 (D.D.C. Aug. 1, 2017); United States v. Olaya, No. 15-CR-20200, 2017 WL 1967500, at *4 (E.D. Mich. Apr. 19, 2017) (declining to suppress information gathered from a warrant that would “aid in identifying individuals involved in” criminal activity). Rather, “the preconditions for a warrant” are probable cause that a crime occurred and that evidence will be found in the place to be searched, “specificity with respect to the place to be searched and the things to be seized, and overall reasonableness.” Zurcher, 436 U.S. at 565, 98 S.Ct. 1970. Probable cause that a specific person committed the crime is not necessary at the warrant stage.
It is worth noting that the government did not seek location information for another [Redacted] center allegedly used in the underlying crimes because it is “a regional ship center with a large customer service area and multiple customer service counters.” Warrant Aff., ¶ 38. Although this does not impact the particularity analysis for the geofence under consideration, it does reflect the government's good-faith effort to narrow the scope of the information it seeks, thereby reducing the infringement on privacy interests the warrant imposes.
Notably, courts have also authorized the collection of CSLI through tower dumps, which arguably have far greater potential to indiscriminately sweep-up individuals’ location information than does the geofence at issue here. CSLI are “time-stamped records that are automatically generated by and for the wireless carrier whenever a mobile device connects to a cell site (i.e., the physical radio antennas that make up the cellular network).” Google Amicus Brief at 13–14. CSLI data reveals that a certain phone number connected to a certain cell site in a certain area at a certain time. It does not provide a pinpoint location for the cell phone user, but rather indicates the cell phone was in a particular “sector” serviced by the cell site when the cell phone “pinged” the site. This data can place the user of the cell phone “within a wedge-shaped sector ranging from one-eighth to four square miles.” See Carpenter, ––– U.S. ––––, 138 S. Ct. at 2211–12, 2218; see also United States v. Graham, 824 F.3d 421, 426 n.3 (4th Cir. 2016) (noting that the CSLI data in that case “could only determine the four-square-mile area within which a person used his cell phone”); Google Amicus Brief at 15 (explaining that CSLI “typically reflects location on the order of dozens to hundreds of city blocks in urban areas rather than a matter of meters, and up to forty times more imprecise in rural areas”). Whereas some warrants seeking historical CSLI merely direct a mobile phone carrier to search through their records for a single, identified cell phone's CSLI, a “tower dump” seeks “every phone that connected to a particular cell site ... in a particular period.” Google Amicus Brief at 14 (emphasis added); see also United States v. Pembrook, 876 F.3d 812, 816 (6th Cir. 2017) (“A ‘tower dump’ is a chronological list of every phone number that used the tower for any purpose (voice call, text, internet connection, etc.) regardless of provider (e.g., Verizon, AT&T).”), judgment vacated on other grounds, 138 S. Ct. 2676 (2018). A number of courts have recognized that tower dumps collect a broad swath of information while upholding the warrants authorizing them. See, e.g., United States v. Rhodes, No. 1:19-CR-73, 2021 WL 1541050, at *1 (N.D. Ga. Apr. 20, 2021) (acknowledging that the tower dump at issue resulted in the collection of “[d]ata on calls made by hundreds of individuals”); United States v. Pembrook, 119 F. Supp. 3d 577, 589 (E.D. Mich. 2015) (similar); see also Order at 20, In re Search of Information Associated with Cell Towers (filed under seal) (granting application to issue tower dump warrant despite concern that perhaps “thousands[ ] of cellphone identifiers, many of which would not belong to a suspect or witness” would be gathered by the warrant). For example, the Eighth Circuit recently affirmed the denial of a motion to suppress evidence gathered from a series of wide-ranging tower dumps that collected 90 minutes of CSLI at each of 4 different locations. See James, 3 F.4th at 1104. Although James does not say how many individuals’ CSLI the warrants captured, it is well-understood that “[a]ny order authorizing a cell tower dump is likely to affect at least hundreds of individuals’ privacy interests.” In re Search of Cellular Telephone Towers, 945 F. Supp. 2d 769, 770 (S.D. Tex. 2013). Evidently, these overbreadth issues did not concern the James panel, which found the tower dump warrants adequately particularized without ever mentioning the potential privacy interests of the persons other than the suspect whose location information was collected. See 3 F.4th at 1106; see also United States v. James, No. 18-CR-216, 2018 WL 6566000, at *5 (D. Minn. Nov. 26, 2018) (dismissing the defendant's argument that “[t]he [CSLI] warrants allowed law enforcement to identify the location of hundreds if not thousands of cell phone users on specific days during specific time frames” because, as in this case, “the warrant applications seek information that is constrained—both geographically and temporally—to the [crimes] under investigation”), report and recommendation adopted, 2019 WL 325231 (D. Minn. Jan. 25, 2019). Whatever the magnitude of the third-party privacy interests implicated in James and these other cases, they appear to far outweigh those at issue in this case.
A key issue driving the disparate views in Google I and II and Google III with respect to the two-step protocol appears to have been the scope of probable cause. In Google III, the court found probable cause to seize all the location data inside the requested geofences, because it was likely that “almost all” the location data within the geofence would likely belong either to “perpetrators, co-conspirators, or witnesses.” 497 F. Supp. 3d at 353. The courts in Google I and II maintained that while the government had furnished probable cause to seize the location information of the “unknown suspect,” it had not done the same for the “location information of persons not involved in the crime.” Google II, 481 F. Supp. 3d at 751. A further, more fundamental disconnect between Google I and II and Google III is their divergent views regarding the constitutional relevancy of the data collection process. Whereas Google I and II assessed the search protocol as part of the particularity inquiry, Google III did not, viewing the protocol as “merely a process established for practical concerns rather than constitutional necessity.” 497 F. Supp. 3d at 362 (noting that “Google has established [the two-step] procedure”); see also Google Amicus Brief at 17–19 (noting that “Google developed a multi-step anonymization and narrowing protocol to ensure privacy protections for its users”). As explained below, this Court is of the view that, with respect to geofence warrants, a search protocol can play a role in satisfying the Fourth Amendments requirements.
The Court is mindful that its constitutional role is generally not to manage the government in the execution of a search warrant. See Dalia, 441 U.S. at 257, 99 S.Ct. 1682 (“[I]t is generally left to the discretion of the executing officers to determine the details of how best to proceed with the performance of a search authorized by warrant.”); see also Lo-Ji Sales, Inc. v. New York, 442 U.S. 319, 327, 99 S.Ct. 2319, 60 L.Ed.2d 920 (1979) (holding that a magistrate judge abandoned his proper judicial role when he participated in the execution of a search warrant for obscene materials). What was done here is not Lo-Ji 2.0. When presented with the initial warrant application in this case, the Court noted its concerns, which the government then cured in a subsequent submission. To be sure, those concerns revolved around the manner in which the government planned to execute its warrant. But the two-step process the government first proposed implicated Fourth Amendment issues that a magistrate judge must address. By doing so, the Court has not become a “participat[ant] with the police and prosecutors in [the] execution” of the geofence warrant—i.e., it will not be present (either physically or virtually) when the government serves the warrant on Google, when Google searches its location records, or when the government reviews and analyzes any anonymized location data Google produces and determines the devices for which it will seek identifying information. Lo-Ji, 442 U.S. at 328 n.6, 99 S.Ct. 2319.
To be clear, in the Court's view, the two-step process the government will follow here is not the only way in which a geofence warrant can be tuned to harmonize it with the Fourth Amendment. For instance, in a future case, the government could propose firm disclosure rules and/or conditions in the warrant, consistent with its showing of probable cause, whereby Google would be required to disclose identifying information only on devices which met those rules. The Court having approved those rules when it issued the warrant, adherence to them during the warrant's execution would take the discretion of law enforcement out of the equation. If, say, the government believes that a single suspect or a group of suspects committed crimes at multiple locations, it could write a condition into its search protocol stating that Google is required to disclose identifying information only for the devices that were found in more than one of those locations at or around the time the crimes occurred. See Google I, 2020 WL 5491763, at *7 (noting that “if the government had ... limited the cellular telephone numbers for which agents could seek additional information to those numbers that appear in all three defined geofences, the government would have solved the issues of overbreadth and lack of particularity”). Or, if consistent with the government's theory of the case, the warrant could condition the release of identifying information on the duration a particular device is found within the geofence. If a criminal act took at least 15 minutes to complete, then the government might request identifying information for only those devices within the geofence for that specified amount of time. And, as stated previously, there may also be situations, as in Google III, where the duration and location of the government's proposed geofence and nature of the crime under investigation effectively means that “almost all location data retrieved will be for individuals who are either the perpetrators, co-conspirators, or witnesses to the crime.” 497 F. Supp. 3d at 353. In such cases, additional legal process to obtain identifying information on all the devices falling within the geofence may not be required.
The government has indicated to the Court that it will submit a second warrant seeking to deanonymize the location information it receives at step one of the process, consistent with its showing of probable cause.
Indeed, it is likely that Google's disclosure of anonymized location information to the government at step one does not implicate significant privacy concerns. In a recent case involving anonymous location data, a federal court in California observed that it was “[o]bvious[ ]” that “a person does not have a reasonable expectation of privacy over information that cannot even be connected to her.” Sanchez v. L.A. Dep't of Transp., No. CV 20-5044, 2021 WL 1220690, at *3 (C.D. Cal. Feb. 23, 2021) (concluding that riders of electric scooters did not have a reasonable expectation of privacy in the scooter's anonymized location information that was provided to local government); cf. Moreno v. S.F. Bay Area Rapid Transit Dist., No. 17-CV-02911-JSC, 2017 WL 6387764, at *8 (N.D. Cal. Dec. 14, 2017) (concluding that transmission of anonymized location information to public transit agency via smartphone application did not state claim under state privacy law because transmission of such data was not “an egregious breach of social norms” (quoting Hill v. NCAA, 7 Cal.4th 1, 26 Cal.Rptr.2d 834, 865 P.2d 633, 655 (1994)). Courts have reached similar conclusions with respect to anonymized medical records. See, e.g., In re Nat'l Hockey League Players’ Concussion Inj. Litig., No. 14-2551 (SRN/JSM), 2016 WL 3815132, at *3 (D. Minn. July 13, 2016); United States ex rel. Roberts v. QHG of Ind., Inc., 1:97–CV–174, 1998 WL 1756728, at *8, 1998 U.S. Dist. LEXIS 23512, at *29–*30 (N.D. Ind. Oct. 8, 1998) (“[T]he privacy interests of a patient in his or her medical records is tied to identity information contained in the records. Once the identifying information is removed from the record, the patient's privacy interest is essentially eliminated.”) (citations omitted)); cf. U.S. Dep't of State v. Ray, 502 U.S. 164, 176, 112 S.Ct. 541, 116 L.Ed.2d 526 (1991) (noting, in the FOIA context, that “disclosure of such personal information constitutes only a de minimis invasion of privacy when the identities of the [persons] are unknown”). But what the government's warrant ultimately seeks at step two is identifying subscriber information concerning the cell phones falling within the geofence. And that is the point at which Fourth Amendment protection may well be triggered—an issue that need not be decided here because the government has effectively conceded the point by applying for a warrant. See supra note 14. Nevertheless, it remains true that there is little to no infringement of personal privacy implicated by the anonymized location records disclosed at step one, or, for that matter, where the government concludes after step one that no useful information has been collected and does not seek the deanonymization of any phones at all.