FOIA 26-35: The office counted 52 pleas in 85 minutes — eight days after saying the same kind of aggregate did not exist

Watchdog Alerts

Why isn’t this basic information available to the public?

1. The office granted this request and gave the number: 52 pleas in January 2026, read off the District and Circuit Court dockets, at no charge. It is the only request in the February 2026 statistical cluster that produced a substantive answer.
2. In the same letter the office wrote that ‘there does not exist a document or record of an “aggregate numerical total” of plea agreements entered into and accepted’ — the identical proposition it used seven days earlier, in 26-28, to issue a certificate of nonexistence and quote $1,653.33 for the underlying data.
3. The office says the review ‘took approximately 85 minutes of time.’ That is 1.63 minutes per plea — the only figure in this archive where the office reports how long a records task actually took, rather than how long it estimates one will take.
4. The number is over-inclusive by the office's own admission: it ‘includes misdemeanor level traffic tickets’ because the request's exclusion was ‘too broad to allow for clear differentiation.’ No exemption was cited and no partial denial was stated. 52 is a ceiling, not the answer.
5. The letter misstates its own record. It says the request was dated February 10 and received February 17; the request form is dated February 9, and the fax notification the office itself enclosed logs receipt at 2:05:51 p.m. on February 9. The response was still timely — but only because February 16 was a state holiday.
6. Compliance is described as a favour — the count was compiled ‘out of a courtesy and effort in transparency’ — and the letter warns that ‘future requests’ may be priced. The fee waiver the request asked for was never granted or denied in writing.

Timing caveat. Nothing on this page faults the office for adopting ProsecutorByKarpel in June 2024 or for the migration being difficult. The relevant point is one the office makes itself: the 52-plea count was drawn from the courts' dockets rather than from the office's own case-management system, because a docket entry can be read and a migrated case file, on the office's own account in 26-32, cannot be searched by year. That is a statement about where the office's records live, not about bad faith.

On February 9, 2026, a requester faxed the Delta County Prosecutor's Office a one-page request for a single number: the total number of plea agreements entered into and accepted in criminal cases during January 2026. The office answered on February 17, 2026. It granted the request, and it gave a number — 52. In the same two-page letter it also wrote the sentence it had used seven days earlier to justify not producing a number at all: “There does not exist a document or record of an ‘aggregate numerical total’ of plea agreements entered into and accepted.” This time the office counted anyway, charged nothing, and said the review took 85 minutes.

That combination is the reason this file is in the archive. FOIA 26-28 asked for a headcount of open cases and was answered with a certificate of nonexistence and a price of $1,653.33 to assemble the underlying data. FOIA 26-35 asked for a headcount of pleas in one month and was answered with the number itself, free, in under an hour and a half. Both requests described the same absence: no maintained aggregate record. Both letters say so. Only one of them treats that absence as a reason the public cannot have the number.

Summary

Four numbers organize this file, and a fifth explains the rest. 52 is the number of pleas the office counted. 85 is the number of minutes it says the count took, in minutes. 1.63 is what those two numbers produce when divided — the average time the office spent per plea, in minutes. Zero is the fee it charged, in dollars, and also the number of pages of responsive documents it produced, because the answer was a figure written into a letter rather than a record released from a file.

The fifth number is the one that makes the file worth reading: eight. That is the number of days between the fax arriving at the office and the date the office says it arrived. The request form is dated February 9, 2026. The office's own fax notification records the transmission at 2:05 p.m. Eastern on February 9, two pages, received on the office's fax line. The response letter is dated February 17, 2026, and states that the request was “dated February 10, 2026, and received by the Delta County Prosecutor's Office on February 17, 2026.” A letter cannot describe eighty-five minutes of docket review and also describe a request received the same day it was answered. One of those two statements is wrong, and the fax log says which.

None of that makes the response late. Because February 16, 2026 was a Monday state holiday, the fifth business day after a February 9 receipt fell on February 17 — the day the letter is dated. The office met the deadline under MCL 15.235 by exactly one holiday. What the record will not support is the office's own account of when the clock started.

What was requested

The request is a single page, plus a second page of fee and timing boilerplate. It is captioned “Freedom of Information Act Request,” addressed to the Delta County Prosecutor's Office, dated February 9, 2026, and its operative paragraph asks for:

“The total number of plea agreements entered into and accepted in criminal cases handled by the Delta County Prosecutor's Office during the month of January 2026, excluding minor traffic infractions and small ticket matters. This also includes pleas that might have been made in December 2025 or earlier. To clarify I am asking for the amount entered and accepted in January 2026.”

Four features of the request matter to what followed. First, it is a count of events in a month, not a count of files. A plea entered on the record in January is a docket entry, and docket entries are the most routinely recorded fact in a courthouse. Second, it contains an exclusion: minor traffic infractions and small ticket matters are expressly outside the request. That clause is the source of the only real defect in the answer. Third, it anticipates the office's standard objection and closes it off in advance, stating that the request “seeks only an aggregate numerical total,” that no case files, names, identifiers or personally identifiable information are requested, that it “does not require the creation of a new public record,” and that it includes “any existing reports, summaries, database query results, or electronic exports that reflect this total.” Fourth, it asks for a fee waiver or reduction and, if any fee is anticipated, a detailed estimate before processing.

The request also volunteers a reason for believing the data exists, and the reason is the most human thing in the packet. The requester writes that he “personally observed a court proceeding on February 9, 2025, in which the Prosecutor agreed to time served and a reduced charge, which I understand to be resolved through a plea agreement.” The year is almost certainly a typo for 2026 — the request is dated February 9, 2026 and describes an observation offered as the basis for a request about January 2026. It is worth noting because the office's response then introduces its own confusion about the same event, and the office's version is harder to excuse.

One small thing about the request itself, noted for completeness rather than fault. The form asks the office to respond “within five (5) business days as required by MCL 15.235(2).” The five-business-day period is in subsection (1); subsection (2) is the extension provision. The citation is the requester's, not the office's, and it had no effect on the response.

What the office answered

The February 17 letter is two pages and makes five moves. The first is the disposition, and it is a single word: “Your request is GRANTED.” There is no “in part,” no Notice of Denial, no exemption cited, no redaction, no deposit and no fee. In a statistical cluster where every other file was denied, this one was granted outright.

The second move is definitional, and it is the office doing the requester a genuine service. “It should be noted, that a plea is only ‘accepted’ when a plea is actually entered on the record. Until that point, either side can revoke or withdraw from any agreement. As such, your request would simply require the review of court dockets for entries that include a ‘plea’.” That is correct as a matter of criminal procedure, and it is also a disclosure about method: the count was not drawn from the prosecutor's own case-management system. It was drawn from the courts' dockets.

The third move is the number. “In reviewing the dockets for District Court and Circuit Court, the Circuit Court, these documented the entry of 52 pleas in District or Circuit Court that were entered pursuant to a plea agreement.” The sentence is garbled in the middle — “the Circuit Court, the Circuit Court” — and the reader has to work to recover the meaning, but the figure is unambiguous.

The fourth move is the concession, and it is the one that keeps this file honest. “This number includes misdemeanor level traffic tickets, as a differentiation of ‘minor traffic infractions and small ticket matters’ is too broad to allow for clear differentiation in response.” The office is telling the requester, in writing, that the number it is providing is larger than the number that was asked for. That is a real disclosure, volunteered, and it belongs in the office's favor. It is also the reason 52 cannot be read as the answer to the question posed.

The fifth move is the paragraph that ties this file to the rest of the archive:

“It should be noted that while this Requester may have witnessed the taking of a plea on February 9, 2026, this does not create an aggregate statistical record as he now requests. The calculation of the number above, required the physical review and cross reference to the physical file was done out of a courtesy and effort in transparency. There does not exist a document or record of an ‘aggregate numerical total’ of plea agreements entered into and accepted. On January 9, 2026, this Requester simply saw the taking of a plea, not the creation of a record that compiles or creates statistical information responsive to his request. Due in large part to the small-time frame requested, the physical review was able to be conducted. However, the granting of this request does not mean that future requests will not be subject to a fee estimation, as even with this small sample size, the review took approximately 85 minutes of time.”

Three things sit inside that paragraph. The office says the count was compiled “out of a courtesy and effort in transparency” — compliance framed as a favor. It says no aggregate record exists, which is the same proposition it advanced in 26-28, 26-29, 26-30, 26-20 and 26-22. And it warns that granting this request “does not mean that future requests will not be subject to a fee estimation,” which is a fee waiver granted by silence and withdrawn in advance.

The paragraph also carries two different dates for the same observed proceeding. Its first sentence says the requester “may have witnessed the taking of a plea on February 9, 2026.” Five sentences later the office writes “On January 9, 2026, this Requester simply saw the taking of a plea.” The requester's own form says February 9, 2025. So three dates attach to what is plainly one afternoon in a courtroom, and the honest reading is that all three are meant to be the same day. The requester's is a year slip: the form is dated February 9, 2026 and the observation is offered as the basis for a request about January 2026, so “2025” is a keystroke. The office's first reference has it right — February 9, 2026, the day the request was faxed. The office's second reference drops a month. That is a slip too, but it is a slip in the one paragraph the office uses to argue that the requester saw an event and not a record, and it sits in a letter that also misstates the date it received the request and the date the request was signed. A typo is a typo. The pattern is that the office's errors all run in the direction of making its own handling of the file look tidier than the documents show.

The number is not the number that was asked for

The request excluded “minor traffic infractions and small ticket matters.” The answer includes “misdemeanor level traffic tickets.” The office's reason is that the exclusion “is too broad to allow for clear differentiation in response.” That is a candid explanation and it may well be accurate — distinguishing a small ticket matter from a misdemeanor traffic ticket in a docket line is not always possible from the docket alone.

What the office did not do is treat that difficulty as a partial denial. There is no Notice of Denial in the packet, no statutory basis cited for withholding the excluded category, and no statement that a portion of the request was denied. Instead the excluded category was folded into the total and the total was called granted. Under Michigan's act, a public body that cannot separate exempt from nonexempt material is expected to say so and to identify the basis; here nothing was withheld, so nothing was exempt, but the figure supplied answers a broader question than the one asked.

What was asked for, and what was counted

Element of the request	What the office counted	Responsive?
Plea agreements entered and accepted	Docket entries showing a plea entered pursuant to a plea agreement	Yes
In criminal cases handled by the office	District Court and Circuit Court dockets	Yes, by method
During January 2026	Entries dated in January 2026	Yes
Excluding minor traffic infractions and small ticket matters	Includes misdemeanor-level traffic tickets	No — over-inclusive by the office's own statement
Aggregate numerical total only	52	Yes

The practical consequence is that 52 is a ceiling, not an answer. The true figure for the request as written is 52 minus whatever misdemeanor traffic tickets were entered on a plea in January 2026. The record does not say how many that is, and the office did not offer to find out. A requester who wanted the narrower number would have to ask again, and the office has already said that future requests may be priced.

The same absence, answered two ways

This is the part of the file that matters beyond its own four corners. In the first half of February 2026 the office answered seven statistical requests from the same requester. Six were denied. One was granted. The granted one is the only one where the answer could be read off a court docket instead of assembled from the office's own migrated files.

The February 2026 statistical cluster

Request	What was asked	Answer	Fee quoted
26-20	Aggregate case totals	Denied	$0
26-21	Amendment to a prior request	Denied	$9,548.00
26-22	Conviction rate	Denied	$9,548.00
26-28	Cases pending when the prosecutor took office	Denied, with a certificate of nonexistence	$1,653.33
26-29	Inherited cases that ended in conviction	Denied	$1,033.33
26-30	Eight named documents	Denied	$0
26-35	Pleas accepted in January 2026	Granted — 52	$0

The office's position in the six denials rests on a single factual claim: the aggregate does not exist as a record, and producing it would require creating one or reconstructing it from files. In 26-28 that claim came with a certificate of nonexistence and a price of $1,653.33 to search 800 files. In 26-29 it came with a price of $1,033.33 to search 500. In 26-22 it came with a price of $9,548.00.

In 26-35 the same claim came with the number. The office did not say the aggregate existed. It said the opposite, in the plainest sentence in the packet: “There does not exist a document or record of an ‘aggregate numerical total’ of plea agreements entered into and accepted.” Then it supplied the total anyway, because the underlying facts were on a docket it could read.

That is the finding, and it is narrower than it first appears. The office is not being accused of lying about its files. A docket entry and a case file are different objects, and the work of reading 52 docket lines in two courts is not the work of reconstructing 800 case files. What the record shows is a threshold decision, made case by case, about when compiling an aggregate is worth the office's time. When the compilation required opening its own files, the answer was a certificate and an invoice. When the compilation required reading a docket, the answer was 52, free, in 85 minutes. The public has no way to know where that line is drawn, because the office has never stated one.

When the clock started

The request form is dated February 9, 2026. The fax notification in the packet is from IPFone, addressed to the office's administrative fax mailbox, and records a transmission on Monday, February 9, 2026 at 2:05:51 p.m. Eastern, two pages received. That is the office's own record of receipt, generated by its own fax service, and it is in the packet the office sent back.

The response letter says something else. It acknowledges “your Freedom of Information Act request dated February 10, 2026, and received by the Delta County Prosecutor's Office on February 17, 2026.” Both dates are wrong on the face of the documents the office itself enclosed: the request is dated February 9, and the fax log shows it arriving February 9.

The error is not merely clerical, because the receipt date is what starts the statutory clock. Michigan requires a response within five business days of receipt. Counting from the true receipt date, and excluding the weekend and the February 16 state holiday:

Five business days from the fax

Date	Day	Business day	Event
February 9, 2026	Monday	—	Fax received, 2:05 p.m. Eastern, two pages
February 10, 2026	Tuesday	1
February 11, 2026	Wednesday	2
February 12, 2026	Thursday	3
February 13, 2026	Friday	4
February 16, 2026	Monday	—	State holiday, not counted
February 17, 2026	Tuesday	5	Response letter dated and mailed

On that arithmetic the response is timely, and this page says so plainly. It is timely on the last available day, and it is timely only because of the holiday; without February 16 the fifth business day would have been the sixteenth and the letter would have been a day late. The office is entitled to the benefit of that calendar.

What the office is not entitled to is its own description of the calendar. A letter dated February 17 that claims the request was received February 17 is claiming a same-day turnaround for a task the same letter says took 85 minutes of physical docket review across two courts. Either the receipt date is wrong or the work description is. The fax log resolves it, and the resolution is that the office understated the elapsed time by eight days in a document that exists to record exactly that.

Eighty-five minutes, and the year it says it needs

The office's own measurement is the most useful number in the packet, because it is the only place in this archive where the office reports how long a records task actually took rather than how long it estimates a task will take.

Eighty-five minutes produced 52 plea entries from the dockets of two courts. That is 1.63 minutes per entry, or a little under a minute and forty seconds each, including whatever cross-reference to the physical file the letter describes. Compare that with the per-file figures the office has used to price the reconstruction of its own case files:

The office's per-item time figures, in its own documents

Source	Date	Task	Time per item	Measured or estimated
Memo re Appeal of 25-94	June 16, 2025	Reconstruct outcomes from case files	20 minutes	Estimated
25-278 Detailed Cost Itemization	December 2025	Reconstruct outcomes from case files	14 minutes	Estimated
26-28 / 26-29 / 26-30 letters	February 2026	Search and review case files	5 minutes	Estimated
26-35 response letter	February 17, 2026	Read plea entries off two courts' dockets	1.63 minutes	Measured, by the office

The comparison has to be handled carefully, and this page will not overstate it. Reading a docket line is not reconstructing a case file. A docket entry is a single dated row; a case file is a folder of documents that has to be opened, read, and summarized. The 1.63-minute figure does not show that the office's 5-, 14- or 20-minute estimates are inflated, and it would be unfair to present it that way.

What it does show is that the office is capable of measuring its own work when it chooses to, and that it did so here in a single sentence, unprompted, as a warning about future fees. In every file where the office has quoted a price, the per-file figure is an estimate offered without a count behind it — the same reconstruction priced at 20 minutes a file in June 2025, 14 minutes a file in December 2025, and 5 minutes a file in February 2026. In the one file where the office actually did the work and reported the clock, the number was 85 minutes. That contrast is the reason Pattern 22 on the Wickman Record now carries a measured figure alongside the estimates.

The waiver that was granted by silence

The request asked for “a waiver or reduction of fees” on the ground that the scope was narrow and the subject aggregate. The response letter never mentions the waiver. It does not grant it, does not deny it, and does not cite MCL 15.234 in either direction. It simply charges nothing.

The practical outcome is a full waiver: the office produced the number and asked for no money. The record of that outcome, however, is an absence. A requester cannot point to a sentence in which the office decided the request was in the public interest, because there is no such sentence. What the letter does say is the opposite of a commitment: “the granting of this request does not mean that future requests will not be subject to a fee estimation, as even with this small sample size, the review took approximately 85 minutes of time.”

Read together, those two facts produce a peculiar posture. The office waived the fee without saying so, and pre-announced that it may not waive the next one, using the time it just spent as the reason. In a cluster where the same requester has been quoted $9,548.00, $1,653.33 and $1,033.33 for comparable compilations, a free answer with a warning attached is better than an invoice. It is also not a decision anyone can rely on.

The timeline

FOIA 26-35, dated

Date	Event	Source
February 9, 2026	Request dated and faxed; two pages received at 2:05:51 p.m. Eastern	Request form; IPFone notification
February 10, 2026	Date the office says the request bears	Response letter
February 16, 2026	State holiday; not a business day	Calendar
February 17, 2026	Response letter dated; request granted; 52 pleas reported; 85 minutes of review; no fee	Response letter
February 17, 2026	Date the office says the request was received	Response letter

What the record shows

The grant is real and should be credited. This is the only request in the February statistical cluster that produced a substantive answer. The office read two courts' dockets, counted 52 entries, disclosed that the count was over-inclusive, and charged nothing. A site that documents denials has an obligation to document a grant with the same care, and this one does.

The number does not answer the question asked. The request excluded minor traffic infractions and small ticket matters. The answer includes misdemeanor traffic tickets, by the office's own admission. 52 is therefore a ceiling. The office did not cite a basis for declining to separate the excluded category, and did not describe the response as partial.

The same absence of a maintained aggregate produced opposite outcomes eight days apart. On February 12 the office certified that no aggregate existed and priced the underlying data at $1,653.33. On February 17 it said no aggregate existed and supplied the total for nothing. The difference between the two files is not the legal theory; it is where the underlying facts were stored.

The office's stated receipt date is contradicted by its own fax log. The letter says February 17; the IPFone notification says February 9 at 2:05 p.m. The response was timely on either reading, but only just, and the record of when the clock started is wrong in the office's own favor.

Compliance is described as a courtesy. The count was compiled “out of a courtesy and effort in transparency.” That phrasing treats a statutory obligation as a discretionary kindness, and it is consistent with the letter's warning that future requests may be priced.

The office can measure its own work. Eighty-five minutes for 52 entries is the first measured figure in this archive. Every priced request in the record rests on an estimate instead.

Eighty-five minutes, and the year it says it needs

The Delta County Prosecutor's Office has spent most of this archive explaining why it cannot count things. It cannot say how many cases were open when the prosecutor took office, because that is a calculation and not a document. It cannot say how many inherited cases ended in conviction, because the old system's data is gone and the files are organized by name only. It cannot produce a conviction rate without 1,077 hours of reconstruction, priced at $9,548.00, or 800 files of search, priced at $1,653.33, or 500 files, priced at $1,033.33. In file after file the answer has been the same: the number does not exist, and making it exist would cost the county more than the county is willing to spend.

Then, on February 17, 2026, the office counted 52 pleas in 85 minutes and charged nothing.

It is worth being precise about why that happened, because the easy version of this story is wrong. The office did not suddenly discover that aggregates are cheap. It discovered that this particular aggregate was already written down — not in its own case-management system, which it has described under oath-adjacent cover as unreliable since the June 2024 migration, but on the dockets of the District and Circuit Courts. A docket entry is a public row in a public register. Reading 52 of them took less time than a lunch hour.

So the office's position is not incoherent. It is a position about where information lives. If the answer is on a docket, the office will read the docket. If the answer requires opening the office's own files, the office will quote a price designed to make the question go away. Both of those are defensible in isolation. Together they describe a policy that has never been written down and that the public cannot see: a threshold, applied case by case, that determines whether a request about the office's own performance is answered or invoiced.

The 85 minutes is what makes that threshold visible. It is the office's own measurement, offered unprompted, in a letter whose purpose was to warn the requester that the next request might cost money. In the same sentence the office demonstrates that it can time its work precisely and that it has never done so when quoting a price. The reconstruction estimates — 20 minutes a file, then 14, then 5 — were not measurements. They were numbers chosen to produce an invoice. The 85 minutes was a measurement, and it was produced because the office wanted to justify a future charge.

There is a second thing the 85 minutes shows, and it is the thing the office probably did not intend. It shows how small the gap is between “we cannot tell you” and “here is the number.” Fifty-two entries, two courts, one afternoon. The office did not need a new database, a consultant, or a year and ten months. It needed somebody to read a docket. When the office says that answering a question about its own caseload would take 1.87 years, the reader now has a benchmark to hold against that claim — not a proof that the claim is false, but a demonstration that the office knows how to work quickly when the work is easy, and that it has never explained why the other work is not.

And there is a third thing, smaller and harder to forgive. The letter cannot keep its dates straight. It says the request is dated February 10 when the form says February 9. It says the request was received February 17 when the office's own fax service logged it at 2:05 p.m. on February 9. It says the requester witnessed a plea on February 9, 2026, and then, five sentences later, on January 9, 2026 — a month off, in the sentence the office uses to make its point. None of these errors changes the outcome. All of them appear in a document whose entire function is to establish, on the record, what the office received and when it answered. An office that cannot state its own receipt date in a FOIA response is asking the public to take a great deal on faith — including the 1.87 years.

The fair conclusion is not that the office lied about its files. It is that the office has discretion over how hard it looks, that it exercises that discretion without saying so, and that when it chooses to look it can produce an answer in under ninety minutes. The 52 is a real number. The 85 minutes is a real measurement. What is missing is any statement of principle that would tell a requester in advance which questions will be answered and which will be priced — and until the office writes that principle down, every denial in this archive carries a question the office has not answered: was this impossible, or was it just not convenient?

What this record does not show

This page does not show that the 52 figure is inaccurate. The office counted docket entries in two courts and disclosed the one limitation it knew about. Nothing in the packet suggests the count was careless.

It does not show that the office was legally required to compile the number. Michigan's act does not oblige a public body to create a record it does not maintain, and the office's position that no aggregate record exists is consistent with what it has said in every other statistical file.

It does not show that the fee waiver should have been granted in writing, only that it was never decided in writing. The office charged nothing, which is the outcome a waiver would have produced.

It does not show that the receipt-date error was deliberate. A mis-transcribed date is the most ordinary error in a records office. What the record shows is that the error runs in the direction that makes the office look faster, and that the office's own fax log contradicts it.

It does not establish that the office's reconstruction estimates are inflated. The 1.63-minute figure measures a different task. The comparison is offered as context, not as proof.

What to ask next

The obvious follow-up is scale. If 52 entries took 85 minutes, twelve monthly counts should take roughly the same effort each, and the answers would show whether the January figure is typical. A request for the same count across all twelve months of 2025 would test whether the office's willingness to count survives repetition, and whether the fee warning in this letter was meant.

The second follow-up is the work product. The letter describes a “physical review and cross reference to the physical file.” If that review produced notes, a tally sheet, or a marked docket, those are records, and they are the records that would show how the 52 was assembled and how the traffic tickets were handled.

The third is the one this file makes unavoidable: ask the office to state, in writing, the standard it applies when deciding whether to compile an aggregate from its own files rather than from a court docket. There may be no such standard. Asking for it is how the public finds out.

The fourth is narrower and cheaper. Ask for the same January 2026 plea count excluding misdemeanor traffic tickets, and ask the office to identify the basis if it declines to separate them. The office has already said the differentiation is difficult; a request that names the specific docket codes it considers “minor traffic infractions” would remove the difficulty or expose it.

Authorities and sources cited on this page

Show the statutes, cases, and sources cited on this page

Michigan Freedom of Information Act, MCL 15.231 et seq.; MCL 15.233(1) (certificate of nonexistence, cited in the companion files); MCL 15.234 (fee waiver and reduction, invoked by the request); MCL 15.235 (response deadlines); MCL 15.240 (appeal to the chairperson of the county board of commissioners or judicial review in circuit court, as stated in the response letter). The response letter also states that a prevailing requester may recover attorney fees, costs and disbursements, actual or compensatory damages, and punitive damages of $1,000.00.

Documents relied on: the February 17, 2026 response letter (two pages); the IPFone fax notification dated February 9, 2026 (two pages); and the request form dated February 9, 2026 (two pages). All six pages are published below with the requester's identifying information redacted.

Documents

These are redacted copies prepared for publication. Requester names, home addresses, personal phone numbers, and email addresses have been removed. Unredacted originals are not hosted on the public web server.

Related FOIA requests

• FOIA 26-292 — FOIA 26-292: The office used the requester's own cost-saving clause to withdraw the request, then offered to grant it if he removed the clause (Granted in part, denied in part — no records produced; the office conceded the data exists and refused the aggregate on the ground that the requester's own scope limitation excluded it, October 25, 2026)
• FOIA 26-291 — FOIA 26-291: The office answered a request for a privilege log with nine document titles — and cited a subsection of the FOIA that does not exist (Granted in part — a nine-row title-and-date table supplying two of the eleven requested fields, no count, and a citation to a subsection of MCL 15.243 that does not exist, October 25, 2026)
• FOIA 26-290 — FOIA 26-290: The office refused the victim-services records because it says you already have them — from a packet that contained none (Refused — five paragraphs labelled “granted” but zero pages produced, on the authority of a prior production that contained no responsive documents; the sixth paragraph received no disposition, October 25, 2026)
• FOIA 26-131 — FOIA 26-131: Twenty case files, priced at $82.66 before a single page is copied (Granted in part — release conditioned on a $41.33 deposit; no records produced yet, April 17, 2026)
• FOIA 26-34 — FOIA 26-34: The office denied a count of diversion cases by defining “diversion” down to a programme it does not run (Denied — certificate of nonexistence under MCL 15.233(1), Denial Form box 2 checked, on the ground that the only diversion programme in Delta County is juvenile diversion administered by the Probate Court; no exemption cited, no fee quoted, no waiver determination, February 9, 2026)
• FOIA 26-33 — FOIA 26-33: Asked to count deferred prosecution agreements, the office certified the total nonexistent and priced the underlying data at $9,093.33 (Granted in part and denied in part — the aggregate record denied as requiring creation of a new record, with a certificate of nonexistence and Denial Form box 2 checked; the underlying data granted subject to a $9,093.33 estimate, a $4,546.66 deposit and a 270-day production estimate; public-interest fee waiver left undetermined, February 9, 2026)
• FOIA 26-32 — FOIA 26-32: What the office's own memos say about the migration that broke its records — a 95-page production that answers its own denial (Granted in part, denied in part — five records produced across 95 pages, with only personal contact information redacted under MCL 15.243(1)(a); no fee quoted, February 6, 2026)
• FOIA 26-31 — FOIA 26-31: Asked how many crime scenes a prosecutor attended, the office listed eight documents and denied them as duplicative (Denied in full — refused as a duplicative request under Densmore v Department of Corrections, on the ground that the eight listed records were already produced in response to FOIA 25-267; Denial Form box 1 checked with the section 13 subsection left blank; no aggregate figure supplied and no fee quoted, February 6, 2026)
• FOIA 26-30 — FOIA 26-30: Eight documents, one number, and a Densmore refusal built on a premise the record does not fully support (Denied — no aggregate total produced; eight responsive documents identified but not reproduced under a Densmore prior-production rationale; no fee quoted, February 6, 2026)
• FOIA 26-29 — FOIA 26-29: A count of inherited convictions the office could not locate — priced at $1,033.33 for the work of counting (No count produced — office could not locate a record; underlying data granted subject to segregation and priced at $1,033.33 with a $516.66 deposit, February 5, 2026)
• FOIA 26-28 — FOIA 26-28: A headcount of cases open when the prosecutor took office — denied, then priced at $1,653.33 on an 800-file baseline (Denied in part, granted in part — no record states the headcount, with a certificate of nonexistence; underlying data granted subject to segregation and priced at $1,653.33 with an $826.66 deposit, February 5, 2026)
• FOIA 26-22 — FOIA 26-22: A conviction rate the office says does not exist, priced at $9,548 (Denied — record does not exist; $9,548 quote for the underlying files, January 30, 2026)
• FOIA 26-21 — FOIA 26-21: A count of convictions the office says does not exist — then a quiet amendment that changed the number behind the bill (Denied as a compilation — no count produced; fee of $9,548.00 quoted for underlying files; amended response revised the volume figure without explanation, January 30, 2026)
• FOIA 26-20 — FOIA 26-20: The aggregate case total, and the records that would let you count it (Denied — record does not exist, January 30, 2026)
• FOIA 25-279 — FOIA 25-279: The office granted the personnel and budget records on paper — then sent three invoices totaling $34,673.79 and none of the records (Granted in part on paper — dozens of records listed as enclosed and none enclosed; three cost itemizations totaling $34,673.79, November 25, 2025)
• FOIA 25-278 — FOIA 25-278: The office says it never tracked plea bargaining — then bills $23,146.66 for a 4,000-file review it says is already covered by another invoice (Granted in part — 62 trials produced from the calendar; plea-offer data certified not to exist; a $23,146.66 estimate the office says is moot if the twin invoice is paid, November 25, 2025)
• FOIA 25-277 — FOIA 25-277: No annual statistical report exists, the underlying data does, and the office prices it at $34,720 (Granted in part — no statistical report exists; the underlying data is granted at a $34,720 estimate built on a volume figure the office has never counted, November 25, 2025)
• FOIA 25-257 — FOIA 25-257: Three certificates of nonexistence, then a $19,840 bill to rebuild the data by hand (Denied — three certificates of nonexistence, with a $19,840 estimate to rebuild the same data by hand, October 31, 2025)
• FOIA 25-254 — FOIA 25-254: Two certificates of nonexistence, two documents produced, and a $74,520 itemization copied from an earlier request (Granted in part, denied in part — two documents produced, two certificates of nonexistence, and a $74,520 estimate identical to FOIA 25-244's, October 31, 2025)
• FOIA 25-244 — FOIA 25-244: Ten service slips, thirteen policy manuals, and a $74,520 quote (Granted in part, denied in part, October 10, 2025)
• FOIA 25-241 — FOIA 25-241: The special-prosecutor petition the office filed, and the 2024 memo that says it did not need another one (Granted in part, denied in part, October 10, 2025)
• FOIA 25-240 — FOIA 25-240: Officer use-of-force referrals, and the $1,863 bill for the one part the office would search (Granted in part, denied in part, October 10, 2025)
• FOIA 25-239 — FOIA 25-239: The Brady/Giglio list and how the office tracks officer credibility (Granted in part, denied in part, October 10, 2025)
• FOIA 25-234 — FOIA 25-234: Four lab reports, two certificates, and 26.4 grams where the press said four ounces (Granted in part, October 1, 2025)
• FOIA 25-202 — FOIA 25-202: No RFP, no bid, no purchase order — and the emails that show how the system was picked (Granted in part, August 18, 2025)
• FOIA 25-201 — FOIA 25-201: ProsecutorByKarpel contracts, invoices, and licensing (Granted in part, denied in part, August 18, 2025)
• FOIA 25-200 — FOIA 25-200: The software the office runs on — seven systems named, one of them retired in June 2024 (Granted in full, August 18, 2025)
• FOIA 25-197 — FOIA 25-197: A request for complaints against the office came back as the office's own civil-rights lawsuits (Granted in part, denied in part, August 13, 2025)
• FOIA 25-196 — FOIA 25-196: Audits of the office, 2019 through 2024 — five county financial statements produced, two fiscal years missing (Granted in part — five annual county financial statements produced; fiscal years 2022 and 2024 absent from the release; no exemption cited and no certificate of nonexistence, August 13, 2025)
• FOIA 25-195 — FOIA 25-195: Reports on the office's own work — four case-tracking reports produced, performance reviews certified nonexistent, county reports withheld as attorney-client (Granted in part and denied in part — ten documents produced; county reports withheld on attorney-client privilege; performance reviews certified nonexistent; $5,961.60 estimated, $2,980.80 deposit, 270 days, August 13, 2025)
• FOIA 25-194 — FOIA 25-194: Every email and text the Prosecutor sent in 2024 — a $63,028.80 estimate on a request the requester offered to narrow (Granted in part and denied in part — exemptions cited; $63,028.80 estimated, $31,514.40 deposit required, 270 days, August 13, 2025)
• FOIA 25-215 — FOIA 25-215: Three answers to one request — two certificates of nonexistence, a grant, and a $13,248 estimate (Granted in part, August 9, 2025)
• FOIA 25-180 — FOIA 25-180: Two boxes checked — the trial records both “already provided” and “do not exist” (Denied in full — certificate of nonexistence plus an exemption box checked; no fee, no pages produced, August 2, 2025)
• FOIA 25-139 — FOIA 25-139: Forfeitures in cases that ended without a conviction, certified not to exist — while the same office took an extension for the volume of records (Denied in full — certified that the record does not exist; no fee quoted, no pages produced, June 27, 2025)
• FOIA 25-137 — FOIA 25-137: The average time from charge to disposition, certified not to exist — after a ten-day extension for “voluminous” records (Denied in full — certified that the record does not exist; no fee quoted, no pages produced, June 27, 2025)
• FOIA 25-136 — FOIA 25-136: The plea-agreement count the office certified does not exist — after taking an extension for the volume of records (Denied in full — certified that the record does not exist; no fee quoted, no pages produced, June 27, 2025)
• FOIA 25-135 — FOIA 25-135: The office denied the 2024 trial summary as a record that “does not exist” — then produced eight case files on the next page (Granted in part — 8 case files produced, June 27, 2025)
• FOIA 25-134 — FOIA 25-134: The dismissed-case count the office certified does not exist — after taking an extension for the volume of records (Denied in full — certified that the record does not exist; no fee quoted, no pages produced, June 27, 2025)
• FOIA 25-133 — FOIA 25-133: The office certified the monthly count did not exist — then attached an annual one and offered to sell the underlying data for $2,815.20 (Granted in part — one annual screening table produced; monthly counts withheld; $2,815.20 quoted with a $1,407.60 deposit, June 27, 2025)
• FOIA 25-129 — FOIA 25-129: Arraignment-to-disposition times the office certified do not exist, then priced at $2,931.12 (Granted in part — two items denied as nonexistent; $2,931.12 quote on the third, June 26, 2025)
• FOIA 25-118 — FOIA 25-118: Final case outcomes for January and February 2025 (Fee estimate issued, June 18, 2025)
• FOIA 25-94 — Prosecutor Case Outcome Statistics, 2021–2025 (Granted in Part, May 23, 2025)