Nobody Screens the Envelope: Who Owns No-Contact Enforcement Inside the St. Clair County Jail?

A sheriff’s office answered a complaint honestly, and in doing so described an enforcement architecture in which every contact channel the facility digitized is controlled, and the one it did not is nobody’s assigned job.

The Answer Was More Useful Than the Outcome

I filed a complaint. I got an answer back. The answer resolved nothing, and it was the most useful document I have read on this subject all year.

On July 21, 2026, a lieutenant in the Corrections Division of the St. Clair County Sheriff’s Office replied internally to a web contact submission reporting alleged contact from a person held at the county jail. The reply was not hostile. It was not evasive. It was candid, procedurally accurate, and almost certainly correct on every point of policy it described. That is precisely why it is worth publishing. In a handful of sentences it lays out the entire enforcement architecture of a no-contact order inside a county jail, including the portion of that architecture that does not exist.

Read that as an operations statement rather than as a rejection, because that is what it is. The facility built controls around the channels it can measure. Email routes through a vendor platform with an approval list. Visitation routes through a scheduling system with the same list. Messaging is a permission that can be switched off, and in this instance it was. Each of those systems produces a log. Each log can be audited. Each control can be enforced at the moment of the attempt, before contact occurs.

Paper produces none of that. Paper leaves the housing unit, enters the mail stream, and arrives. There is no approval gate, no attempt log, and no audit trail on the facility side. The lieutenant did not describe a failure of will. He described the absence of a system, and then he correctly identified whose problem that becomes.

The Four Custodians of a No-Contact Order

A no-contact provision, whether it arrives as a personal protection order under MCL 600.2950 or 600.2950a, a bond condition, or a probation term, is not self-executing. It is a command that depends on four separate actors to function. Each holds real authority. The question that determines whether the order works in practice is narrower than authority: which of them is assigned to notice when the order is breached.

What Custody Actually Contains

There is a widely held assumption, shared by protected parties and by a fair number of people who work in courts, that incarceration functions as containment. The person is inside. The order is in place. The problem is therefore managed.

Custody contains movement. It does not contain communication. It contains the communication channels a facility elected to route through auditable systems, and it releases the rest into the general mail stream with no facility-side record that anything left.

The distinction matters because the two categories are drifting apart. Jails have digitized aggressively over the past decade. Electronic messaging, video visitation, and tablet programs are now standard, and each one arrived with an approval list, a permission toggle, and a log. That build-out produced genuine control. The lieutenant’s account of the account restrictions is evidence of it working: two approved contacts, texting restricted, everything checked against the list.

Physical mail did not get that treatment because there was no vendor incentive to build it and no operational reason to want it. Outgoing mail generates no revenue, creates constitutional exposure when handled badly, and consumes staff hours at a facility already short on them. So the one channel that predates the digitization is the one channel that remains outside it, and it is also the one channel a person subject to an electronic restriction would most predictably use.

Why the Gap Is Structural, Not Local

Nothing in this analysis is a criticism of St. Clair County. The response describes a condition common to county jails across Michigan and well beyond it, and the legal framework explains why.

Federal law permits facilities to regulate outgoing correspondence. In Procunier v. Martinez, 416 U.S. 396 (1974), the Supreme Court held that censorship of outgoing mail must further a substantial governmental interest unrelated to suppressing expression and must be no broader than necessary to serve that interest. Later decisions confined that standard to outgoing correspondence and applied the more deferential reasonableness framework of Turner v. Safley, 482 U.S. 78 (1987), to incoming material. The operative point for enforcement purposes survives all of it: a facility that screens outgoing mail to prevent contact with a protected party is on solid legal ground. Preventing harassment of a person protected by court order is about as substantial as governmental interests get.

So the constitutional question is not the obstacle. Facilities may screen. The obstacle is that nothing requires them to. County jail mail policy in Michigan is set facility by facility, and a sheriff’s office weighing finite corrections staffing against a discretionary screening function will reach the answer the lieutenant reached, every time, and will be correct to reach it.

On the judicial side the machinery is equally intact and equally reactive. A personal protection order issued under MCL 600.2950 or 600.2950a is enforceable through contempt proceedings governed by MCR 3.708, and Michigan law authorizes warrantless arrest where an officer has reasonable cause to believe such an order has been violated under MCL 764.15b. Both mechanisms work. Both require someone to arrive with the violation already in hand.

The Burden Lands Where It Was Supposed to Be Lifted

A protective order is, in theory, a transfer of burden. Before the order, a person managing unwanted contact carries the whole weight of avoiding it. After the order, the state assumes part of that weight and backs it with contempt power and arrest authority.

In this channel the transfer does not complete. The order issues. The state’s enforcement apparatus stands ready. And the protected party is still the only party positioned to notice that anything happened, still the party who must preserve the envelope intact, still the party who must produce a police report, and still the party who must draft and file the motion that gives the court something to act on.

That is not a small logistical inconvenience. It is a structural inversion. The person with the least institutional power in the sequence holds the only operative duty in it, and holds it because everyone else’s duty was defined narrowly enough to exclude it.

There is a second cost that shows up later. When the protected party is the sole detection mechanism, the record of violations is necessarily incomplete and necessarily self-generated, which is exactly the record that draws skepticism when it eventually reaches a courtroom. A person reporting repeated contact with no institutional corroboration begins to look like a person with a pattern of reporting rather than a person subject to a pattern of contact. The absence of a screening system does not merely permit violations. It degrades the credibility of the person reporting them.

What Enforcement Costs the Person Holding the Order

I am going to shift registers for a moment, because what follows is not analysis. It is experience, and I am labeling it as experience so you can weigh it accordingly.

I have held personal protection orders. In my experience they have worked considerably better as documents than as protections. The order issues. The contact does not reliably stop. And the response I have most often received when reporting it, in various phrasings from various people, amounts to the same instruction: take it back to the judge.

I understand why that instruction gets given. An officer arriving on a report of unwanted contact is being asked to make a rapid judgment about the scope of an order that officer has not read, about conduct that is often ambiguous on its face, in a category of case that competes for attention with everything else on the shift. Declining to act is the low-risk option for the responding agency and the high-cost option for the person reporting. Nothing about that requires anyone to be indifferent. It only requires the incentives to point the direction they point.

What that instruction actually means, though, is worth spelling out for anyone who has not lived it. It means a motion. It means a filing. It means service. It means a hearing date, and then frequently an adjournment, and then another date. It means time off work, and childcare, and mileage, and parking, and sitting in a building with the person the order was supposed to keep away from you, waiting for a docket call. It means doing all of that competently enough to survive an opposing argument, usually without counsel, because protective order enforcement is rarely where a person’s limited legal budget goes. And then it means doing it again the next time, because the ruling addresses the incident, not the pattern.

This is where a purely structural analysis usually stops, and it should not, because the cost has a consequence the record never captures. People stop filing. Not because the contact ended. Because the price of responding to it exceeded what they had left to spend. The order stays in place, the file stays quiet, and the quiet reads to every institution downstream as resolution.

That is the part that should concern anyone doing policy work in this space. Enforcement data on protective orders measures the violations that someone had the resources to pursue. It cannot measure the ones abandoned for cost, and abandonment is invisible by design. A system that requires the protected party to fund its own enforcement will systematically undercount its own failures and will read attrition as success.

None of this means the order is worthless. It establishes the line, it creates the predicate that makes later conduct chargeable, and it puts a fact in writing that would otherwise be one person’s account. What it does not do is watch the line. That was never assigned to anyone, and the previous sections explain why. What people want when they seek one of these orders is not litigation. It is quiet. It is the ability to go to work and raise their children and answer their own phone without doing threat assessment first. The instrument they are handed to obtain that quiet requires them to litigate for it, repeatedly, at their own expense, on their own time, for as long as the other person is willing to keep going.

The Counterargument, Taken Seriously

The facility’s position is defensible and should be stated at full strength before it is answered.

Universal screening of outgoing mail is genuinely expensive. A jail processes a substantial daily volume, and reading it all would require dedicated staff a county corrections budget does not contain. Legal mail cannot be read at all, so any screening program requires a sorting protocol and staff trained to apply it, which introduces error and liability. Over-screening carries real constitutional exposure under the Martinez line of cases. And delay in outgoing mail imposes a cost on the large majority of incarcerated people writing to family, counsel, employers, and children, none of whom are subject to any order.

All of that is correct, and none of it answers the actual question, because the actual question is not whether a facility should read everything. It is whether a facility that already maintains an approved contact list can also maintain a prohibited addressee list.

What Fixing It Would Actually Require

The remedy is a records integration problem, not a staffing problem, and it is small.

None of this requires reading mail. None of it requires new corrections officers. It requires a court to send a document it already generates to a facility that already maintains a contact database, and it requires someone to look at the address on the envelope before it goes in the bag.

Why This Matters Beyond One County

Every county jail in Michigan holds people subject to no-contact conditions. Every one of those facilities has digitized its electronic contact channels over the past decade and left its outgoing mail policy roughly where it stood before. Every issuing court operates on the assumption that a violation will be brought to it. And in every one of those counties, the person who has to bring it is the person the order was written to protect.

The gap is not the product of anyone’s bad faith. It is the product of four institutions each defining their duty accurately and none of them defining it broadly enough to close the seam. That is how most institutional failures actually happen. Not through a decision to permit harm, but through a series of defensible narrow readings that add up to nobody being responsible for the outcome.

I want to be direct about my own position here, because it would be dishonest to pretend I came to this document as a neutral observer. I came to it as a person who filed a complaint and got a candid answer. The candor is what made it publishable. A defensive response would have told me nothing. This one told me exactly where the machine stops, which is more than most institutions will put in writing.

And I would rather have the answer than the outcome, because the outcome only fixes my problem. The answer identifies a seam that runs through every county in this state, and a seam that has been named is a seam that can be closed.