As regular readers know, I’m very interested in the new practice some magistrate judges have adopted of conditioning computer warrants on ex ante restrictions that they themselves create and attempt to impose on law enforcement. A district judge recently decided a new case on the role of the restrictions that in my view demonstrates why such restrictions are troubling. The case is United States v. Salceda, decided on February 27 by Judge Snyder of the Central District of California.
First, a brief background on those who are unfamiliar with the practice. (If you want more details, read my recent article Ex Ante Regulation of Computer Search and Seizure.) Imagine the government wants to conduct a search, and they apply for a warrant. Traditionally, the role of magistrate judges has been to review the warrants applications for probable cause and particularity, as the text of the Fourth Amendment requires. If the warrant application establishes probable cause and satisfies the particularity requirement, the judge must issue the warrant. In the last few years, however, some magistrate judges have taken on a new role. These judges worry that computer searches are more invasive than traditional searches. They also don’t trust the process of litigating the constitutionality of computer warrant searches after-the-fact. As a result, they have begun conditioning the issuance of computer warrants on special conditions that they devise.
For example, a judge might say that he won’t allow the warrant unless the government agrees to not use any evidence discovered in plain view. Or the judge won’t allow the government to have the warrant unless the government agrees to search a computer a particular way. The basic idea is to use the magistrate judge’s power to issue or deny the warrant ex ante to try to regulate what happens later, with each magistrate judge free to impose whatever restrictions he or she wants (beyond probable cause and particularity) to try to ensure that the resulting search won’t be too broad. Most magistrate judges have not imposed these new restrictions, but a number have. The lawfulness of imposing such restrictions currently remains uncertain. The Vermont Supreme Court heard argument nine months ago on what likely will be the first appellate case to directly consider the lawfulness of the practice, but no decision has been handed down yet.
That brings us to the Salceda case. The police had probable cause to believe that Salceda had child pornography on his home computer. They asked for a warrant to enter Salceda’s home, seize his computer, and look for evidence. Magistrate Judge Parada agreed that the government had established probable cause and particularity, but then decided to impose conditions on the warrant. Specifically, Magistrate Judge Parada decided to impose a condition designed to deal with the fact that the police can’t know what is inside a computer when they execute the initial search at the suspect’s home, and therefore for practical reasons have to seize each computer they find and search it back at the lab to see if it has the evidence sought. Appellate precedent has allowed this initial seizure on grounds that such a practice is reasonable. But the magistrate judge imposed a condition that law enforcmeent had to conduct an initial search back at the lab within 60 days of seizing each device to determine if the device had evidence described in the warrant. The warrant then included the following instruction:
If, after conducting such an initial search [within 60 days from the date of the execution of the search warrant], the case agents determine that a digital device is an item to be seized or contains any data falling within the list of items to be seized pursuant to this warrant, the government will retain the digital device for further analysis; otherwise, the government will return the digital device.
In Salceda, the government seized the suspect’s computer, searched it in the initial 60 day period, and found lots of child pornography on it. The agents then put the computer aside. Almost two years later, as the case picked up again, the agents decided that they wanted to search the computer a second to find the rest of the child pornography on it. But at this point the agenta ran into a problem: Judge Parada’s warrant restriction wasn’t clear about whether they could do this. The warrant said that they could “retain the digital evidence for further analysis,” but it’s not clear if that meant that they could search the computer again or merely hold on to it or analyze files already searched.
The police applied for a second warrant to search the rest of the computer based on the evidence of the child pornography already discovered during the first search. As far as I can tell, the request for the second warrant was submitted to the duty judge, who happened to be a different judge. The second judge, Judge Zarefsky, told the investigators to go ask Judge Parada to clarify what the search restriction was intended to do instead of obtaining a second warrant, and denied the second warrant application.. The agents decided instead to search the computer again based on the first warrant, taking the view that the restriction allowed the agents to conduct “further analysis” such as searching the computer. The second search of the computer discovered more child pornography. The defendant then moved to suppress the extra child pornography discovered during the second search.
Held: The extra child pornography must be suppressed. Here’s the analysis:
The Court finds that any evidence discovered during the September and October 2011 forensic searches is inadmissible. The original warrant’s “further analysis” language is ambiguous as to whether it permits the government to analyze defendant’s digital devices for additional evidence of contraband, or whether it merely permits additional analysis of previously discovered contraband. Given this ambiguity, suppression of the evidence is appropriate. See Transfiguracion, 442 F.3d at 1228 (construing ambiguities in plea agreements in favor of a defendant because the government, as drafter of the agreement, bears the “responsibility for any lack of clarity”) (internal quotation marks and citation omitted); Wilhelm, 425 F.3d at 463 (finding fourth amendment violation because officer “recognized the warrant as ambiguous before the execution of the warrant, but failed to immediately stop execution and seek the necessary clarification of a warrant in order to make certain the warrant particularly described the place to be searched. . . .”). The Court’s decision is informed by the fact that the government sought–and was denied–a new warrant from Judge Zarefsky, and thereafter sought ex parte relief from this Court before conducting the subsequent searches. At the very least, the government’s conduct demonstrates its acknowledgment that the “further analysis” language is ambiguous. In denying the government’s ex parte application, the Court directed the government to seek clarification from the issuing judge to cure any ambiguities. Because the government chose to proceed with the searches without clarification or without securing a new search warrant, that evidence must be suppressed. See Wilhelm, 425 F.3d at 463.
Salceda is premised on the wrong analysis, I think. The Fourth Amendment requires reasonableness, and the Supreme Court and circuit courts have created a doctrinal structure for what reasonableness means and when unreasonable searches justify an exclusionary remedy. In my view, it is improper for magistrate judges and district court judges to simply ignore that binding appellate caselaw and replace it with an inquiry into whether agents properly followed a particular magistrate judge’s ex ante restrictions. In Salceda, the court seems to envision the warrant as some sort of contract between the government and the magistrate judge. She then asks if the agents lived up to their end of the deal, without ever considering any actual Fourth Amendment principles. But that’s just wrong, in my view. Warrants are not contracts, in which judges get to strike deals with investigators about what investigators can do. The Fourth Amendment requires reasonableness, and that allows investigators to do some things and not others. Reasonableness is up to the appellate caselaw, not individual magistrate judges with particular preferences.
The important Supreme Court precedent here is Richards v. Wisconsin, 520 U.S. 385 (1997). In Richards, a magistrate judge refused to allow agents to execute a warrant without first knocking and announcing their presence. The agents did so anyway. The defendant moved to suppress the evidence on the ground that the search was unlawful because it violated the magistrate’s express condition of granting it. The Supreme Court disagreed in a unanimous decision by Justice Stevens:
In arguing that the officers’ entry was unreasonable, Richards places great emphasis on the fact that the magistrate who signed the search warrant for his hotel room deleted the portions of the proposed warrant that would have given the officers permission to execute a no knock entry. But this fact does not alter the reasonableness of the officers’ decision, which must be evaluated as of the time they entered the hotel room. At the time the officers obtained the warrant, they did not have evidence sufficient, in the judgment of the magistrate, to justify a no knock warrant. Of course, the magistrate could not have anticipated in every particular the circumstances that would confront the officers when they arrived at Richards’ hotel room. [n.7] These actual circumstances–petitioner’s apparent recognition of the officers combined with the easily disposable nature of the drugs–justified the officers’ ultimate decision to enter without first announcing their presence and authority.
In my view, that’s the right way to analyze Salceda, too. The issue should be whether the search was reasonable, and if not, whether the exclusionary rule should apply. The details of whether the agents did or did not follow the search restriction — and how that ambiguous search restriction should be interpreted — should be irrelevant because it is not relevant to the reasonableness of the search itself.