Three weeks of scaffolding, and here’s where the model act starts to touch the ground. Week 1 named the object; Week 2 set the default to no; last week’s § 3 built the authorization to get past that default and insisted it scale with intrusiveness. Once the government is authorized to collect, though, a second question opens up, and it’s the one where digital law quietly falls apart: how must the collection be carried out? That’s the procedure section, and its governing idea is a word straight out of the Fourth Amendment — particularity.
Go back to the text I quoted in the very first post. A warrant must be supported by probable cause and must “particularly describe the place to be searched, and the persons or things to be seized.” For a physical search, that requirement does muscular work. It names the house, sometimes the room, and the things the officers may take, and it confines them: a warrant for a stolen car doesn’t authorize opening desk drawers. Particularity in a physical search is a limit on execution. It follows the officer around the house.
But particularity is only half of what the Fourth Amendment demands, and for digital collection the other half carries more weight. Particularity lives in the Warrant Clause; it disciplines the description — the precision with which the warrant names what may be seized. The Amendment’s opening command is broader and blunter: the people are to be secure against unreasonable searches and seizures. Reasonableness doesn’t govern the description; it governs the conduct— the manner, the scope, and the duration of what the government actually does. And the two are not the same requirement. A warrant can be flawlessly particular and still be executed unreasonably; the Supreme Court has long held that even a valid warrant must be carried out in a reasonable manner. Particularity asks whether the target was described with enough precision. Reasonableness asks whether the government’s search — how much it took, how it took it, how long it held on — was itself reasonable. Keep both in view, because a device search often fails on each.
Now watch what happens to particularity in a device search. The warrant particularly describes its target — say, evidence of drug trafficking. And then the government images the entire phone. Every photograph, every message, every location point, every app’s stored data, the responsive 5 percent and the intimate 95 percent alike, copied in a single act. The description was particular. The seizure was total. Particularity survived in the warrant’s words and died in the execution, because the way we collect digital evidence is to take all of it and sort later. A warrant that particularly describes its target and then authorizes copying a person’s entire digital life is a general search wearing a particular warrant. And the deeper failure is the one particularity can’t even see: copying everything, retaining the untouched remainder, and returning to it later are questions about the manner and duration of the search — questions of reasonableness — and a particular warrant answers none of them. Particularity has nothing to say about what happens after the seizure; reasonableness has everything to say, and today almost nothing is required to satisfy it.
Title III saw this problem coming, in its own idiom, and § 2518 answered it by building limits into the execution, not just the application. Four of them matter here. There is minimization: § 2518(5) requires agents to conduct the intercept so as to minimize the capture of communications not subject to the order — a real-time limit on what actually enters the government’s hands. There is duration: an order runs thirty days, extendable only on a fresh showing, so the collection has an end. There is sealing: § 2518(8)(a) requires the recordings to be sealed under the judge’s direction, and the seal is a precondition to using the evidence — an integrity mechanism welded to admissibility. And there is inventory: § 2518(8)(d) requires that the target eventually be told it happened. Minimization, duration, sealing, notice — every one of them constrains how the surveillance is carried out and what survives it. Digital collection, as currently practiced, has none of them by default.
So § 4 of the model act imports the idea Title III got right — procedure that binds execution — and updates each piece for a forensic image instead of a phone tap. Four mechanisms.
First, particularity that limits collection, not just description. The warrant should define and bound the scope of what may be collected, not merely the offense under investigation, and should prefer targeted extraction to whole-device imaging wherever the technology allows it. When comprehensive imaging is genuinely necessary — the § 3 necessity showing from last week — that necessity is stated and approved, not assumed.
Second, minimization, relocated. Here I’ll concede something to the forensics up front: you cannot minimize a forensic image in real time the way an agent minimizes a live call, because you have to copy the drive in order to search it at all. Fine. Then the minimization duty moves downstream and becomes a duty to segregate: separate the responsive material from the non-responsive remainder, through filter teams or search protocols, and strictly limit what may be retained from the rest. Minimization stops being a real-time act and becomes a mandatory post-imaging one — but it does not disappear, which is often what happens today.
Third, sealing and integrity, adapted. The forensic image should be hashed and sealed under judicial direction at the moment of collection, and every later operation on it logged in a handling record — a chain of custody for data. This is the one piece the government should want, because it’s the same thing that authenticates the evidence in court: the Federal Rules already let electronic data be self-authenticated by hash verification. An integrity requirement isn’t a burden bolted onto collection; it’s the service that makes the evidence admissible, and it happens to double as the record that lets the person later prove what was done to their data.
Fourth, a hook to retention. Collection is not a license to keep. Section 4 connects to the retention clock that § 6 will build, so that the procedure for taking data carries within it the beginning of the rule for eventually letting it go.
Now the objection, which is real and technical. You cannot know what’s responsive, the government will say, until you’ve imaged and examined the whole drive; targeted extraction is often impossible; the complete image is required to authenticate the evidence and to give the defense what it’s entitled to; and “real-time minimization” of a static data set is close to incoherent. Image everything and sort later, on this view, isn’t a Fourth Amendment dodge — it’s the only way the work can be done.
I accept the premise and reject the conclusion. Yes, making the full image is often a technical necessity. No, that necessity does not justify retaining the full image indefinitely, or re-searching it later for new purposes, or treating the 95 percent that fell outside the warrant as a permanent government asset. This is the same move I made in Week 1: separate the technically unavoidable act from the legal permissions attached to it. Permit the full image for the purpose of the search. Then require segregation, put a clock on the non-responsive remainder, seal and hash the image — which serves the very authentication interest the government just invoked — and log the handling. The forensic reality argument proves that imaging must be allowed. It proves nothing about what may be kept, and collapsing those two questions is exactly how the general search smuggles itself back in. Courts have already begun groping toward execution-stage rules — the Ninth Circuit’s search-protocol experiment is the famous example, and it’ll get its own post — but a protocol improvised case by case is a poor substitute for a procedure written down in advance.
We saw the cost of the missing procedure two weeks ago. The government lost Richman in large part because, when it imaged and collected, it never segregated the responsive material from the rest — the precise failure § 4 is built to prevent. And notice which clause did the work: the court didn’t hold that the warrants lacked particularity. It held that the government’s continued possession had become unreasonable. Particularity described the target; reasonableness is what finally reached the conduct — years later, one case at a time. Particularity that binds only the warrant’s language, and not the government’s hands, is a formality we perform on the way to taking everything. Section 4 writes reasonableness into the procedure up front — bounded collection, segregation, sealing, a clock — so the reckoning doesn’t have to wait for a judge to call the retention unreasonable long after the data has been gone through.
Next Tuesday, § 5: once the data is lawfully collected, what governs the analysis of it — the stage Title III never had to imagine, and the one where the modern search actually happens. And Thursday, we follow the data past the moment of collection into a stranger place, where two federal laws give the government opposite orders about what to do with it.
(Model statutory text for § 4 comes in Pass 2 this fall. This post is the why; the drafting is next.)
DigitalCrimPro 
Leave a comment