Where the index stands
Assessment status: Definitive. Score 34 / 100 (Low Transparency). Overall evidence coverage 98.8% across 6 of 6 categories and 53 distinct requests.
Of 37 indicators, 36 carry a score, 1 are partial and carry no score, and 0 are unassessed or not applicable. A partial indicator is recorded as null, never as zero, and it lowers the coverage figure rather than the score.
| Assessment period | May 23, 2025 – November 2, 2026 |
|---|---|
| Evidence cutoff date | November 2, 2026 |
| Last updated | November 2, 2026 |
| Methodology version | v1.2.0 (released October 25, 2026) |
| Assessment note | Expanded assessment under methodology v1.2.0. All 16 archived FOIA requests, all 6 published case watches, the seven Aker-Horvath requests documented inside that case watch, the released documents and the county's public FOIA pages were reviewed against 37 indicators. FOIA 25-244 (Brady/Giglio service records, October 2025) was added on October 25, 2026 and lowered the scores on aggregate-data availability, the cost of obtaining existing data, nonexistence certifications, search-scope disclosure and fee communication. The Aker-Horvath records (26-136, 26-137, 26-139 and 26-140 through 26-143) were added on October 25, 2026. They corroborate the existing judgments rather than overturn them: four responses on the fourth business day with no fee support the timing and fee scores, while a bare certificate of nonexistence for sixteen categories of fingerprint records, two same-dated plea sheets that disagree on bond, and a request whose subject line reads 'Finger Prints' while enumerating plea documents support the existing low scores on nonexistence certifications and internal consistency. No indicator score was changed on this review. Case watches are admitted as evidence only where the page reproduces primary records verbatim and cites them. Indicators without enough traceable evidence carry no score; null is not zero. FOIA 25-277 (annual statistical reports, submitted November 25, 2025 and answered December 18, 2025) was added on October 30, 2026. It is the first itemization in the archive whose arithmetic reconciles, which raised the fee-itemization and fee-timeline scores; the letter and the itemization also agree on a 50% deposit and a 180-day clock for the first time. Against that, the response produced no records at all, priced data the office concedes exists at $34,720.00, omitted the fee appeal form, certified a redaction it had not yet performed, left the public-interest waiver section blank, and rested the whole estimate on a volume figure - roughly 1,000 files per year - that appears in five requests in this archive and has never been derived from a count. Twelve indicator scores moved: A1, A3, A4, A5, B1, C1, C2, C3, C4, E2, E3, E4, F1 and F5, with eight rising and six falling. All outstanding reassessment flags were resolved on this review. FOIA 25-278 (plea-bargaining practices, January 1, 2022 to present), submitted November 25, 2025 and answered December 18, 2025, was added on October 30, 2026. It is the literal twin of 25-277 - same requester, same filing day, same response day, same coordinator, same formula - which makes consistency across similar requests directly testable for the first time; the office is consistent, and the consistency is mechanical. The response is also the first of its class to produce actual records: a trial count by court and year totaling 62 trials over four years, plus SCAO Form MC 414 and a plea-offer revocation template. It volunteers a fuller migration account than the archive had, stating that the transition completed in June 2024 and explaining the field mismatch, which raised the historical and migration-transparency score. Against that, the office issued a second itemization the same day for the same review and conceded in its own footnotes that the second is 'moot' if the first is paid - $57,866.66 combined, $28,933.33 in deposits - and again priced a 4,000-file manual review to answer a question the internal calendar answered in 62 lines. Twenty indicators were reviewed and their notes rewritten; fourteen moved and six were confirmed unchanged. All 25-278 reassessment flags were resolved on this review. FOIA 25-279 (personnel and resource data, FY 2020 through the current year), submitted November 25, 2025 and answered December 18, 2025, was added on October 30, 2026. It is the third request filed by the same requester on the same evening and answered by the same coordinator on the same day, and it is the most revealing of the three. The letter grants dozens of sub-items in prose - position classification lists back to 2009, headcount, turnover, training records by employee, thirteen travel authorizations, annual budgets, grant awards and fourteen grant-funded job descriptions - and encloses none of them. The 74-page packet is a letter, a denial form, two statute excerpts, a blank appeal form, a policy notice and three invoices. The invoices quote $176.96, $34,432.00 and $64.83, totaling $34,673.79 with $17,336.89 in deposits and clocks of 30, 180 and 30 days, and all three contain unreconciled arithmetic errors. The volume premise behind the largest invoice moved from '4 YEARS requested' in the twin requests to '6 YEARS requested' here without explanation. Against that, the response discloses real search scope, a fuller six-year position history than the archive had, and the clearest statement yet of the office's size - 8 employees, 3 attorneys, 12-15 vacancies - and of the limits that size imposes. Twenty indicators were reviewed and their notes rewritten; nineteen moved and one was confirmed unchanged. All 25-279 reassessment flags were resolved on this review. FOIA 26-21 (total count of criminal convictions, 2023 through 2025), submitted January 30, 2026 and answered February 5, 2026, with an amended response dated February 12, 2026, was added on October 30, 2026. It is one of three statistical requests faxed the same day by the same requester - 26-20, 26-21 and 26-22 - all answered within days, all certified as producing no existing document, and all met with a fee quote built on the same estimate of case volume. The count was denied as a compilation under MCL 15.233(5) with a certificate of nonexistence under MCL 15.233(1), and the underlying files were priced at $9,548.00 with a $4,774.00 deposit and a 270-day estimate. The distinguishing fact is the amendment. The February 5 letter stated that the office processes 'approximately 1,500 files per year'; at the itemization's own 3-minute and 4-minute rates and $6.20 quarter-hour charge, that volume implies $13,020.00 and a $6,510.00 deposit, $3,472.00 more than the office quoted. The February 12 amendment changed the figure to 1,100 files per year - matching the invoice - and changed nothing else, including the fee, the deposit, both footnotes, the appeal paragraph and two typographical errors. A fee the office justified by volume did not move when volume fell 27 percent. Footnote 2 names requests 25-21 and 25-22 that do not exist in the archive, and the fee waiver was requested and never decided. Twenty-nine indicators were reviewed and their notes rewritten; thirteen moved and sixteen were confirmed unchanged. All 26-21 reassessment flags were resolved on this review. FOIA 26-29 (total number of inherited criminal cases resulting in conviction) and FOIA 26-30 (total number of criminal cases initiated from the election date through December 31, 2025) were added on October 30, 2026. Both were submitted in the same week as 26-20, 26-21 and 26-22, making five statistical requests from one requester in a single week. 26-29 was received February 6 and answered February 12, 2026; the office reported it could not locate a record containing the total, defined 'inherited' by reference to a request outside the archive (26-28), and granted the underlying data subject to segregation at $1,033.33 with a $516.66 deposit, on a baseline of 500 files against an estimate of 800 to 1,000 open at any one time. Its itemization states that partial time increments must be rounded down, and at that rule line 3a is $409.20 rather than the $413.33 printed - a $4.13 overcharge - while the letter gives 180 days and the itemization 270 days, both dated the same day. 26-30 was received February 9 and answered February 12, 2026; the office named eight responsive documents and produced none, citing Densmore v. Department of Corrections as '203 Mich 363 (1994)' - correctly 203 Mich.App. 363 - while also certifying under MCL 15.233(5) that no document containing the total exists and stating that the total 'can be gleaned' from the documents it declined to produce. Six of the eight documents match what 25-195 produced; the two 21st Century Reports covering 2024 and 2025 do not appear in 25-94 or 25-195, and the Board of Commissioners' June 27, 2025 decision on the 25-94 appeal records those periods as denied, not produced. Sixteen indicators were reviewed and their notes rewritten; fifteen moved and one was confirmed unchanged. All 26-29 and 26-30 reassessment flags were resolved on this review, along with thirteen stale 25-279 flags left open by the prior scan. FOIA 26-28 (total number of criminal cases pending or in progress at the time the current prosecutor assumed office) was added on October 30, 2026. It is the first of the six statistical requests and the missing piece behind 26-29: that letter defines 'inherited' by citing FOIA 26-28 by number, and the citation is now verifiable in this archive and checks out. The office denied the existence of any record containing the total, certified that under MCL 15.233(1), granted the underlying data subject to segregation, and priced it at $1,653.33 with an $826.66 deposit - on a baseline of 800 files, drawn from the same sentence estimating 'approximately 800 to 1000 files open at any one time' that 26-29 used to reach 500 files on the same day in the same template. The itemization repeats 26-29's $4.13 overcharge against its own rounding rule ($657.20 rather than $661.33), its blank waiver section for the sixth consecutive file, and its 270-day estimate contradicting the letter's 180 days. The letter is the only one in the cluster to supply a worked example - a 2001 authorization arraigned in 2025 - and it is the first packet in the archive to supply both the denial appeal form and the excess-fee appeal form together. Sixteen indicators were reviewed and their notes rewritten; fifteen moved and one was confirmed unchanged. All 26-28 reassessment flags were resolved on this review. FOIA 26-32 (impact of the system migration on the office's ability to respond to FOIA requests) was added on October 30, 2026. It is the largest production in the archive - five named records, 95 pages - and it answers the office's own earlier denial by releasing the migration memos it had previously withheld. The June 16, 2025 memo on Appeal of FOIA 25-94 states that the office 'went live with Karpel' on June 10, 2024; that 'the data did not transition completely, or accurately'; that the failure 'was a possibility that we were warned of prior to the implementation'; that 'access to the previous case tracking system or its exact data is no longer available'; and that files are 'organized by name (only by name),' so year-based search is 'not viable.' The January 2026 email thread adds that the office has not reentered the data because 'we don't have time,' and that a commissioner proposed a cheaper technical route through ProsecutorByKarpel's court-form ingestion. The denial in 26-32 is the narrowest in the archive: contact information only, MCL 15.243(1)(a), every substantive record released. Against that, the file exposes three different per-file time estimates for the same work - 20 minutes, 14 minutes and 5 minutes - and repeats the $4.13 fractional-increment overcharge on two lines of the 25-278 itemization. Twenty-three indicators were reviewed and their notes rewritten; twenty-one moved and two were confirmed unchanged. All 26-32 reassessment flags were resolved on this review. FOIA 26-32 (impact of the system migration on the office's ability to respond to FOIA requests) was added on October 30, 2026. It is the largest production in the archive - five named records, 95 pages - and it answers the office's own earlier denial by releasing the migration memos it had previously withheld. The June 16, 2025 memo on Appeal of FOIA 25-94 states that the office 'went live with Karpel' on June 10, 2024; that 'the data did not transition completely, or accurately'; that the failure 'was a possibility that we were warned of prior to the implementation'; that 'access to the previous case tracking system or its exact data is no longer available'; and that files are 'organized by name (only by name),' so year-based search is 'not viable.' The January 2026 email thread adds that the office has not reentered the data because 'we don't have time,' and that a commissioner proposed a cheaper technical route through ProsecutorByKarpel's court-form ingestion. The denial in 26-32 is the narrowest in the archive: contact information only, MCL 15.243(1)(a), every substantive record released. Against that, the file exposes three different per-file time estimates for the same work - 20 minutes, 14 minutes and 5 minutes - and repeats the $4.13 fractional-increment overcharge on two lines of the 25-278 itemization. Twenty-three indicators were reviewed and their notes rewritten; twenty-one moved and two were confirmed unchanged. All 26-32 reassessment flags were resolved on this review. FOIA 26-35 (the January 2026 plea-agreement count) was added on October 30, 2026. It is the first file in the archive in which the office answered a request for an aggregate figure with the figure, at no charge, inside the statutory period. The letter reports 52 pleas drawn from District and Circuit Court dockets, concedes that the 52 includes misdemeanor traffic tickets and is therefore a ceiling rather than an exact count, states that no document or record of an aggregate numerical total exists, and reports that the compilation took approximately 85 minutes and was done 'out of a courtesy and effort in transparency.' Two things keep the file from raising the assessment much. First, the office's account of its own dates cannot be reconciled with the packet: the letter dates the request February 10 when it is dated February 9, states that the request was received February 17 when the enclosed fax notification records 2:05:51 p.m. on February 9, and says the plea was observed on February 9 in one sentence and January 9 five sentences later. Counted from February 9 the response is timely on the fifth business day only because February 16 is a state holiday. Second, the file sharpens the archive's central inconsistency: the same claim that no aggregate record exists produced a certification and a $1,653.33 invoice on February 12 and a free number on February 17, and the 85-minute figure is the only measured labor time in the archive -- about 1.63 minutes per plea, against the 20-, 14- and 5-minute per-file estimates quoted for comparable work elsewhere. Sixteen indicators were reviewed and their notes rewritten; fourteen moved and two were confirmed unchanged. No 26-35 reassessment flags were outstanding. FOIA 26-35 (the January 2026 plea-agreement count) was added on October 30, 2026. It is the first file in the archive in which the office answered a request for an aggregate figure with the figure, at no charge, inside the statutory period. The letter reports 52 pleas drawn from District and Circuit Court dockets, concedes that the 52 includes misdemeanor traffic tickets and is therefore a ceiling rather than an exact count, states that no document or record of an aggregate numerical total exists, and reports that the compilation took approximately 85 minutes and was done 'out of a courtesy and effort in transparency.' Two things keep the file from raising the assessment much. First, the office's account of its own dates cannot be reconciled with the packet: the letter dates the request February 10 when it is dated February 9, states that the request was received February 17 when the enclosed fax notification records 2:05:51 p.m. on February 9, and says the plea was observed on February 9 in one sentence and January 9 five sentences later. Counted from February 9 the response is timely on the fifth business day only because February 16 is a state holiday. Second, the file sharpens the archive's central inconsistency: the same claim that no aggregate record exists produced a certification and a $1,653.33 invoice on February 12 and a free number on February 17, and the 85-minute figure is the only measured labor time in the archive -- about 1.63 minutes per plea, against the 20-, 14- and 5-minute per-file estimates quoted for comparable work elsewhere. Thirty-one indicators were reviewed and their notes rewritten; sixteen moved and fifteen were confirmed unchanged. No 26-35 reassessment flags were outstanding. FOIA 26-290, 26-291 and 26-292 -- the three follow-ups drafted from this site's own potential-request list and sent October 25, 2026 -- were added on November 2, 2026. All three were received Monday October 26 and answered Thursday October 29, the third business day of the five the statute allows, making this the fastest batch in the archive. Speed is the only thing the batch demonstrates. 26-290 asked six questions about victim-services staffing and grant funding; five are labelled 'granted' and withheld under Densmore v Department of Corrections on the ground that the same records were produced in FOIA 25-279, a production that this site documents as containing no responsive documents at all -- a denial, a blank appeal form, a policy notice and three cost itemizations totalling $34,673.79 -- and the sixth question is not answered. 26-291 asked for a privilege log for what 25-195 withheld and received a nine-row table of documents that were produced, all authored by the Prosecutor, under a citation to 'MCL 15.243(9),' a subsection that does not exist; four paragraphs later the same letter says the records 'were not deemed privileged.' 26-292 asked for twelve monthly felony-authorization counts and was refused on the strength of the requester's own cost-saving scope limitation, which the office read as an exclusion of the number it had just said 'could be located and provided.' That is the third written concession in this archive that the underlying data exists while the aggregate is withheld, after 25-278 and 26-28. Across the batch: no fee quoted, no cost estimate, no Notice of Denial form, no count of records withheld in full or in part, no waiver determination, and the sixth item of every request left unanswered. Thirty-six indicators were reviewed and their notes rewritten; twenty-eight moved and eight were confirmed unchanged. No 26-29x reassessment flags were outstanding. FOIA 26-34 (the aggregate number of criminal cases resolved through diversion programmes, 2022-2025), faxed February 9, 2026 and answered February 17, 2026, was added on November 2, 2026. It is the sibling of 26-35: the two requests arrived in the same afternoon's fax traffic at 1:18 and 2:05 p.m., were clarified by the same 1:22 p.m. e-mail, and were answered on the same day by the same coordinator. 26-35 was granted in full, with a count of 52 pleas produced in 85 minutes at no charge. 26-34 was denied outright, and the denial is definitional rather than exemption-based: the office narrowed 'diversion programs' to juvenile diversion administered by Probate Courts under the Juvenile Diversion Act, Act 13 of 1988, MCL 722.821 et seq., asserted that 'There are no other diversion programs within Delta County', and checked box 2, Record Does Not Exist, on the Denial Form. The certificate states that a search occurred but names no system searched, no custodian consulted and no date range applied, and the letter cites no subsection of MCL 15.243. The packet also carries three different receipt dates for one fax -- February 9 on the request form and the IPFone log, February 10 on the Denial Form, and February 17 in the response letter -- and the office's own forwarded e-mail of February 9 at 1:22 p.m. states that the two requests 'just came in' that day. The response is timely on all three readings because February 16, 2026 was a state holiday, so the response-timing indicator is unaffected and its score is unchanged. Nine indicator scores were reviewed: A3, B5, C1, C3, E2, F1 and F4 fell; A6 rose, because the office expressly invited clarifying information and offered to 'review the matter anew'; E5 was already at its floor and is unchanged in score but not in substance. All outstanding reassessment flags were resolved on this review. |
The live widget on the home page reads the same data files. Previously published scores are kept on the historical score archive.
Purpose and limitations
The WWTI condenses the public record of requests to the Delta County Prosecutor's Office into one number that a reader can trace back, indicator by indicator, to a document. It is built to be checked, not trusted.
It has real limits. It measures what the documents in this archive show, and the archive contains only the requests this site has received responses to; it is not a sample chosen to be representative. It does not decide whether any withholding was lawful: that is a question for the Board of Commissioners on appeal or a court. It is an opinion, labelled as one, and its rating bands are defined by WickmanWatch and carry no official standing. Where the evidence is thin, the index says so instead of producing a number.
Six categories and fixed weights
The weights below are read from data/transparency-methodology.json, the single source of truth. They do not change within a methodology version.
| Category | Weight | What it asks |
|---|---|---|
| A. FOIA Responsiveness and Procedural Compliance | 10% | Whether the office answers requests on time and in the form the Michigan Freedom of Information Act requires. |
| B. Record Accessibility and Completeness | 20% | Whether the records released match what was asked for, are usable, and are redacted no more than the law requires. |
| C. Existing Data Availability and Accessibility | 30% | Whether data the office already holds, or its case-management system can already produce, is available to the public at reasonable cost, and whether the office explains clearly when it says it is not. |
| D. Proactive Disclosure | 10% | What the office publishes without being asked. |
| E. Denial Justification and Appeals Accountability | 20% | Whether denials are justified, documented consistently, and open to meaningful appeal. A denial is assessed for how well it is justified, never for the fact that it was a denial. |
| F. Public Communication | 10% | How clearly and accurately the office communicates with requesters and the public about records. |
| Total | 100% |
How cost, complexity and accessibility are measured
Cost is measured as burden, proportionality, itemization, alternatives and outcome, never as a finding of illegality.
- absolute fee burden;
- deposit barrier;
- relative fee;
- implied hourly cost;
- cost per record or page;
- estimate-to-final variance;
- fee-waiver accessibility;
- cost-related abandonment;
- partial fulfillment and affordability;
- electronic delivery and avoidable costs;
Complexity is classified only when the evidence permits. Unknown complexity stays unknown.
- record types;
- date ranges;
- custodians;
- systems searched;
- search hours;
- review hours;
- records expected;
- records produced;
- already public;
- existing data vs new compilation;
- clarification;
- scope narrowed;
- scope-to-cost explanation;
Cost is never scored from a single large number. A fee is compared with the median for comparable requests, with the share of the estimate demanded up front as a deposit, with any stated hourly rate and hours, and with whether a less expensive option, a narrowing option or a waiver was offered. A large fee is treated as a measure of burden, not as proof that a law was broken.
Request-level analytics and the limited-sample rule
- Samples smaller than five are flagged limited. This is a WickmanWatch reporting rule, not a statistical guarantee.
- Requests are never ranked solely by dollar amount.
- Every percentage must carry the count it was calculated from (required).
Every percentage on this site is shown with its denominator. A sample smaller than 5 is labelled limited sample and is never used to generalise about the office. Where the record does not support a figure, the figure is null and is displayed as a dash; it is never displayed as zero. Non-payment of a deposit is not by itself treated as abandonment, and abandonment is never attributed to cost without a document that says so.
How case watches are used as evidence
A published case watch is an evidence source of type `case_watch`. It is admitted only where the page reproduces primary records verbatim and cites each one by date and document number; a case watch that argues from inference is recorded as `analysis` and can never establish sufficiency on its own.
A case watch carrying a FOIA number counts as one distinct FOIA request for the distinct-request check. A case watch without a FOIA number supplies evidence but does not count toward that check and is not counted as a FOIA request.
A case watch never substitutes for the underlying document where the document is published. It is a second, citable route to the same record.
Admitting case-watch evidence flags the indicators it touches for review. It does not by itself raise a score or mark an indicator sufficient; that requires the written re-review recorded on the indicator.
What changed in v1.2.0
v1.2.0 (October 25, 2026). Admits the five published case watches into the evidence base as `case_watch` sources and adds the rule governing when a case watch may be cited. No indicator, weight, band or threshold changes, so scores remain comparable with v1.1.0. Eight indicators were re-reviewed against the new evidence: six moved, two held.
v1.1.0 (October 25, 2026) · breaking change. Expanded from 26 to 37 indicators inside the same six categories and fixed category weights. Adds cost, complexity, timeline, production, migration, proactive-disclosure, denial and communication measures, and rebalances each category's internal weights to 100. Because the scoring structure changed, scores are not comparable with v1.0.0. No v1.0.0 snapshot had been published; the baseline was 'Insufficient Evidence to Score'.
v1.0.0 (October 25, 2026). Initial release: six categories with fixed weights, twenty-six indicators, evidence-coverage rules and rating bands.
Scores are compared only across versions marked as comparable. v1.1.0 changed the scoring structure and is not comparable with v1.0.0; v1.2.0 changed only the evidence base, not the structure, so its scores remain comparable with v1.1.0. No v1.0.0 snapshot was ever published: the baseline was “Insufficient Evidence to Score”, which is not a score of zero.
Overall formula
Each category score runs from 0 to 100 and is the internal-weight average of that category's indicator scores. Scores are held at full precision and shown to the nearest whole number, with halves rounding up; the rating band is read from the whole number shown. An indicator's share of the overall index is its category weight multiplied by its internal weight. For example, an indicator weighted 25% inside the 30% category C represents 7.5% of the overall index.
Rating bands
| Score | Rating |
|---|---|
| 0–19 | Critically Opaque |
| 20–39 | Low Transparency |
| 40–59 | Limited Transparency |
| 60–79 | Generally Transparent |
| 80–100 | Highly Transparent |
These bands are independent WickmanWatch ratings. They are not official government ratings, and no government body has reviewed, approved or adopted them.
What counts as evidence
Every assessment separates five kinds of statement. Only the first two can make an indicator sufficient, and an indicator needs at least one of them.
| Kind | Meaning |
|---|---|
| documented fact | Taken directly from a released record, dated and traceable to a document. |
| official statement | What the public body asserted, in a letter, form or statement to the press. Recorded as an assertion, not as a verified fact. |
| analysis | WickmanWatch's reasoned reading of the record, including arithmetic and comparison across requests. |
| inference | A conclusion drawn from the record that the record does not itself state. Never sufficient on its own. |
| unresolved question | A question the record raises and does not answer. Never sufficient on its own and never scored as a failure. |
A FOIA denial is not treated as unlawful and is not treated as a failure. A fast response does not earn a high score. The categories that look at denials (B, C and E) score how a refusal was explained, supported and applied, not that a refusal occurred; each indicator that could be misread this way carries an explicit note in the table below saying what it does not credit or penalize.
Evidence coverage
Each indicator has exactly one status.
| Status | Meaning |
|---|---|
| sufficient | Adequate, reviewed, traceable evidence supports a score. The only status that counts toward evidence coverage and the only status that carries a score. |
| partial | Some evidence exists but not enough to score. No score is recorded. |
| insufficient | Evidence was reviewed and found inadequate to score. No score is recorded. |
| unassessed | Not yet reviewed. No score is recorded. |
| not_applicable | Does not apply to the period or the facts. Excluded from the coverage denominator only when a written rationale is recorded. |
Category coverage uses the same calculation over that category's internal weights. Only sufficient indicators count toward coverage. An indicator marked not applicable leaves the denominator only when a written rationale is recorded; without one it stays in. Coverage measures how much of the methodology has adequate evidence behind it. It says nothing about whether transparency is good or bad, and it is reported separately from the score.
A category is treated as sufficiently assessed, and receives a score, only when at least 70% of its applicable internal weight is sufficient. A category that falls short carries no score at all. It is stored as null, never as zero.
Definitive, provisional and insufficient
Definitive
All of the following must hold:
- overall evidence coverage of at least 80%;
- at least 4 of the six categories sufficiently assessed;
- evidence coverage of at least 70% in category C (Existing Data Availability and Accessibility) and at least 70% in category E (Denial Justification and Appeals Accountability);
- at least 20 distinct FOIA requests cited by scored indicators, once 20 or more relevant requests are available;
- traceable evidence for every scored indicator; and
- a documented assessment period, evidence cutoff date and methodology version.
Provisional
Overall evidence coverage of at least 50%, at least 3 categories sufficiently assessed, and traceable evidence for every scored indicator. A provisional score normalizes the weights across the categories that were assessed. It is always labelled provisional, it names every category that was left out, and it states the original weights of those categories.
Insufficient Evidence to Score
If neither standard is met, the widget shows “Insufficient Evidence to Score” and no number.
If the data files cannot be loaded, or fail their own checks in the visitor's browser, the widget shows the same state. It does not guess.
Missing evidence and inapplicable indicators
Missing evidence is not proof of poor transparency and is never assigned a score of zero. An indicator with no sufficient evidence carries no score. Where an indicator depends on public sources (for example whether statistics are published), the absence of something is scored only after a dated, documented search; an unsearched source is unassessed. An indicator that does not apply, such as use of the extension where no extension was taken, is excluded from the denominator only with a written rationale that appears in the published assessment.
How denials, exemptions, partial releases and appeals are evaluated
A denial or partial denial is read against what the statute asks of the office, which is to say what the response must contain: the basis for the exemption, a certificate where a record does not exist, a description of any separated material, and notice of the right to appeal (MCL 15.235(5)). The index asks whether the stated basis is specific, whether it fits what was actually released, and whether the Notice of Denial agrees with the letter. It does not ask whether the office was permitted to withhold. The office may withhold exempt material; the privacy exemption in MCL 15.243(1)(a) is discretionary and is often properly applied, so redaction of private identifiers is not penalized.
Where a requester asks for a summary or count that does not exist, MCL 15.233(4) and (5) do not require the office to compile or create one. Refusing to compile is not scored as a failure. What is scored is whether existing data was released, whether the limit relied on was named, and whether a certificate of nonexistence held up against later events.
Appeals are scored on how the right was explained and, where an appeal was decided, on the reasons given and the timing MCL 15.240 contemplates, including the rule that a board or commission is not deemed to receive an appeal until its next regularly scheduled meeting. Where an appeal is known to have been filed but its record has not been located, its status is N/A. Its absence is neither evidence that no appeal exists nor a failure by the office.
Statutory references are provided for public understanding and are not legal advice.
Historical records and the June 2024 ProsecutorByKarpel transition
The office is reported to have begun using ProsecutorByKarpel in June 2024. No exact day is established in the record, and none is assumed. Requests are split at June 2024 rather than faulted for system-generated reports that could not exist earlier.
This is an indicator (C4, Historical Records and System Migration Transparency) inside category C. It is not a seventh category. It looks at documented evidence about:
- Historical records from before implementation;
- Records created during the transition;
- Records created after implementation;
- Whether historical records were migrated;
- Existing reports, exports, archives or other sources of historical information;
- Explanations given when records are described as nonexistent or unavailable;
- Documented search methods and their limitations;
The exact implementation date is not established in the record and is not assumed. Records are not assumed destroyed, migrated or accessible, and software capabilities are not assumed; each is credited only on verified evidence. Requests for records before June 2024 are not faulted for lacking system-generated reports. Where a request spans the transition, it is split at that date.
Indicators and scoring criteria
There are 37 indicators. Internal weights within each category total 100%. Each indicator is scored 0–100 against the anchors shown; reviewers interpolate between anchors and write the reasoning into the evidence record.
A. FOIA Responsiveness and Procedural Compliance 10% of overall
A1 Response timing — 20% of category · 2% of overall
Question. Was each response issued inside the period MCL 15.235(2) allows, measured from the documented date of receipt? Timing is measured in business days and reported as a median and as the share meeting the applicable deadline. Timing is measured in business days and reported as a median and as the share meeting the applicable deadline. Timing is measured in business days and reported as a median and as the share meeting the applicable deadline. Timing is measured in business days and reported as a median and as the share meeting the applicable deadline.
- Score 80–100
- Nearly every request answered inside the statutory period, with receipt and response dates both documented in the packet.
- Score 40–59
- Mixed record: some responses late or dates that cannot be reconciled from the packet.
- Score 0–19
- Most responses late, or no response to a request that was received.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(1); MCL 15.235(2)
- Not penalized or credited
- Speed alone. A fast response that omits required content is scored under A3 and Categories B and E; it earns nothing extra here.
A2 Use of the extension — 12% of category · 1.2% of overall
Question. When the office took the single extension MCL 15.235(2) permits, was it used once, by written notice, with a stated reason and a response date?
- Score 80–100
- Extension used once, notice issued inside the original period, reason stated and a date given.
- Score 40–59
- Extension used but notice incomplete (reason or date missing or illegible).
- Score 0–19
- Extension used without notice, used more than once, or the stated date was missed.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 2 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(2)
- Not penalized or credited
- Requests answered without an extension are not evidence for this indicator; mark it not_applicable only with a documented rationale.
A3 Required content of the response — 18% of category · 1.8% of overall
Question. Does each response contain what MCL 15.235(5) requires: the basis for any exemption, a certificate of nonexistence where applicable, a description of separated material, and notice of the right to appeal?
- Score 80–100
- Every element that applies is present and specific.
- Score 40–59
- Some elements present, others generic or blank.
- Score 0–19
- Required elements routinely absent.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(5)(a); MCL 15.235(5)(b); MCL 15.235(5)(c); MCL 15.235(5)(d)
A4 Fee itemization and deposits — 18% of category · 1.8% of overall
Question. Where fees were quoted or a deposit demanded, was the itemization in the form MCL 15.234 requires, and were deposit rules observed? Itemization is also read for stated labor hours and rates, a per-task split, and the deposit percentage. Itemization is also read for stated labor hours and rates, a per-task split, and the deposit percentage. Itemization is also read for stated labor hours and rates, a per-task split, and the deposit percentage. Itemization is also read for stated labor hours and rates, a per-task split, and the deposit percentage.
- Score 80–100
- Itemization states hourly wage and hours, fees and any deposit are consistent with MCL 15.234, and the requester could check the arithmetic.
- Score 40–59
- Itemization present but incomplete, or the basis for part of the fee is unclear.
- Score 0–19
- No usable itemization, or a deposit or fee demanded without the statutory basis.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.234(2); MCL 15.234(4); MCL 15.234(8); MCL 15.234(13); MCL 15.234(14)
- Not penalized or credited
- A quoted fee is not itself a transparency failure. A fee is scored on how it was itemized and justified, and a $0 fee is not evidence of good practice if the response is otherwise deficient.
A5 Fee estimate and deposit timeline — 14% of category · 1.4% of overall
Question. How quickly after receipt was a fee estimate issued, and how long did the deposit-to-production stage take where it occurred?
- Score 80–100
- Estimates were issued promptly and within the statutory response window, and any deposit-to-production interval is documented and reasonable.
- Score 40–59
- Estimates were issued, but timing is inconsistent or the deposit-to-production stage is undocumented.
- Score 0–19
- No estimate was issued where one was required, or the timeline cannot be reconstructed from the record.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.234(1); MCL 15.234(8); MCL 15.234(14)
- Not penalized or credited
- A fast estimate is not by itself a transparency success; it is scored together with accuracy and itemization. An unknown deposit-to-production date is not scored as a delay.
A6 Disposition, clarification and outstanding actions — 18% of category · 1.8% of overall
Question. Did the response identify the request's disposition, state any outstanding action, request clarification where needed, and explain any revised estimate?
- Score 80–100
- Every response states the disposition, any outstanding action and the next step, and any clarification or revised estimate is explained.
- Score 40–59
- Dispositions are usually stated, but outstanding actions or revised estimates are sometimes left unexplained.
- Score 0–19
- Responses do not state a disposition or next step, or the requester is left without notice of what remains outstanding.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.235(1); MCL 15.235(2); MCL 15.235(5)
- Not penalized or credited
- An unanswered request is not classified as a denial without evidence. A request that needed no clarification is not faulted for lacking one.
B. Record Accessibility and Completeness 20% of overall
B1 Responsive records produced — 28% of category · 5.6% of overall
Question. Compared line by line with the request, how much of what was requested was actually produced? Production is classified as full, partial or none, and record types supplied and not supplied are counted. Production is classified as full, partial or none, and record types supplied and not supplied are counted. Production is classified as full, partial or none, and record types supplied and not supplied are counted. Production is classified as full, partial or none, and record types supplied and not supplied are counted.
- Score 80–100
- Everything requested that exists was produced, or each omission is identified and explained.
- Score 40–59
- A substantial part produced; remaining parts neither produced nor explained.
- Score 0–19
- Little or none of what was requested produced, without explanation.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- A denial is not automatically a failure here: where a record is exempt or does not exist, the question is whether the response said so.
B2 Usability of the records produced — 18% of category · 3.6% of overall
Question. Are the produced records legible, complete pages, and in a form a member of the public can use? Usability includes machine-readable and searchable formats and original electronic files where appropriate. Usability includes machine-readable and searchable formats and original electronic files where appropriate. Usability includes machine-readable and searchable formats and original electronic files where appropriate. Usability includes machine-readable and searchable formats and original electronic files where appropriate.
- Score 80–100
- Legible, complete, and in the format requested or a usable equivalent.
- Score 40–59
- Usable with effort: poor scans, missing pages, or an unrequested format.
- Score 0–19
- Illegible, truncated, or unusable.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
B3 Segregation and redaction — 24% of category · 4.8% of overall
Question. Did the office separate exempt from nonexempt material and release the rest (MCL 15.244), and do the redactions match the exemption cited?
- Score 80–100
- Redactions are limited to material the exemption reaches, and the response says what was removed and why.
- Score 40–59
- Redactions broadly consistent with the exemption but some are unexplained or appear wider than the exemption.
- Score 0–19
- Whole records withheld where only parts were exempt, or redactions with no stated basis.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.243(1)(a); MCL 15.244
- Not penalized or credited
- Redaction of private identifiers under MCL 15.243(1)(a) is not penalized: that exemption is discretionary and often properly applied.
B4 Unexplained gaps and internal consistency — 16% of category · 3.2% of overall
Question. Does the production hang together? Are there omissions, contradictions or unexplained differences between what was produced and what the office itself has said elsewhere?
- Score 80–100
- No unexplained gaps; the production is consistent with the response letter and with the office's own statements.
- Score 40–59
- One or more gaps or inconsistencies that the record neither explains nor rules out.
- Score 0–19
- Material omissions or contradictions that the record shows and the office never addressed.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- A gap is scored only where the record documents it. Suspicion that something is missing is an unresolved question, not a finding.
B5 Search scope and withheld-material disclosure — 14% of category · 2.8% of overall
Question. Did the office explain the scope of its search and identify what, if anything, was withheld?
- Score 80–100
- The search scope, systems or categories searched, and any withheld material are identified clearly enough to assess the response.
- Score 40–59
- Some of the search scope or withheld material is identified; parts are asserted without detail.
- Score 0–19
- Neither the search nor the withheld material is described, so the response cannot be assessed.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.233(4); MCL 15.235(5)
- Not penalized or credited
- A lawful withholding is not penalized; only the explanation and identification are scored.
C. Existing Data Availability and Accessibility 30% of overall
C1 Availability of existing aggregate data — 20% of category · 6% of overall
Question. When asked for statistics or counts, did the office provide data it already maintains or can already generate?
- Score 80–100
- Existing reports, exports or counts provided, or their absence clearly and specifically explained.
- Score 40–59
- Some existing data provided; the rest unexplained.
- Score 0–19
- Existing data refused or described as nonexistent with no verifiable basis.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.233(4); MCL 15.233(5)
- Not penalized or credited
- MCL 15.233(4) and (5) do not require the office to compile a summary or create a new record. A refusal to compile is not scored as a failure; the question is whether existing data was released and whether the office explained the limit it relied on.
C2 Cost of obtaining existing data — 16% of category · 4.8% of overall
Question. Where the office quoted a fee for data, was the cost proportionate to the work described and checkable by the requester? Cost is also read as a relative burden against comparable requests and, where counts exist, per record or page. Cost is also read as a relative burden against comparable requests and, where counts exist, per record or page. Cost is also read as a relative burden against comparable requests and, where counts exist, per record or page. Cost is also read as a relative burden against comparable requests and, where counts exist, per record or page.
- Score 80–100
- Existing data supplied at no or nominal cost, or any fee tied to documented work that is itemized.
- Score 40–59
- A fee that is itemized but whose scope is hard to reconcile with the data requested.
- Score 0–19
- A fee that operates as a barrier and is not tied to documented work.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.234(2)
C3 Handling of 'record does not exist' certifications — 16% of category · 4.8% of overall
Question. When the office certified that a record does not exist, was the basis explained, the search described, and the position consistent across similar requests?
- Score 80–100
- Certificate states what was searched and why nothing responsive exists; consistent with other responses.
- Score 40–59
- Certificate given with no description of the search.
- Score 0–19
- Certificate contradicted by records the office itself later produced, with no explanation.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(5)(b)
- Not penalized or credited
- A certificate of nonexistence is lawful where true. This indicator scores how it was supported and whether it held up, not the fact of the certificate.
C4 Historical records and system migration transparency — 20% of category · 6% of overall
Question. How clearly does the record explain what exists from before, during and after the June 2024 move to ProsecutorByKarpel, and how to find it?
- Score 80–100
- The office documents what historical records exist and where, whether and how records were migrated, what reports or exports can be run, and the search it performed, and it applies one consistent explanation across requests.
- Score 40–59
- Some of these are documented; others are asserted without support or explained inconsistently across requests.
- Score 0–19
- Records are described as unavailable or nonexistent with no explanation of the transition, the search or any alternative source.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- The exact implementation date is not established in the record and is never assumed. Requests whose subject period is before the June 2024 go-live are not faulted for lacking system-generated reports; they are split at that date. Records are not assumed destroyed, migrated or accessible, and software capabilities are not assumed. Each is credited only on verified evidence.
C5 Documentation of system capability — 8% of category · 2.4% of overall
Question. Is verifiable documentation (contracts, manuals, vendor material) available showing what the case-management system can report?
- Score 80–100
- Capability documented in records the office released or in vendor documentation that was verified.
- Score 40–59
- Partial documentation; capability claimed on one side but not verified.
- Score 0–19
- No documentation; capability asserted by either side without support.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 1 reviewed record(s), drawn from at least 1 distinct FOIA request(s). Only one system contract has been obtained, so this indicator is scored on that single reviewed document; the limit is recorded in the assessment note.
- Not penalized or credited
- Software capability is never assumed. Only documented or verified capability counts.
C6 Structured data and export capability — 10% of category · 3% of overall
Question. Does the record show that existing reports, exports or structured data can be produced from the office's systems?
- Score 80–100
- The record documents existing reports or exports, structured data, and the search and retrieval capability to produce them.
- Score 40–59
- Capability is documented in a contract or system description, but not shown in a produced report or export.
- Score 0–19
- No documentation of reporting, export or structured-data capability exists in the record.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 1 reviewed record(s), drawn from at least 1 distinct FOIA request(s). A contract proves capability was purchased, not that it was used. The distinction is recorded in the evidence note.
- Not penalized or credited
- A purchased capability is not treated as proof that a report was actually produced. Capability is never assumed from the vendor's marketing.
C7 Retrieval limitations and alternative sources — 10% of category · 3% of overall
Question. Where records were unavailable, did the office explain whether they were not retained, not retrievable, or available elsewhere?
- Score 80–100
- The office distinguishes unavailable from not-retained records, explains technical limits, and identifies alternative sources.
- Score 40–59
- Some limitations or alternatives are explained; others are asserted without support.
- Score 0–19
- Records are described as unavailable with no explanation and no alternative source.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.233(4); MCL 15.233(5)
- Not penalized or credited
- A nonexistence certification is not itself a failure; the explanation and alternatives are what is scored.
D. Proactive Disclosure 10% of overall
D1 Published statistics and reports — 22% of category · 2.2% of overall
Question. Does the office publish case statistics or annual reports where the public can find them?
- Score 80–100
- Regular, dated, findable statistics or reports.
- Score 40–59
- Occasional or hard-to-find publications.
- Score 0–19
- No published statistics located after a documented search.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 1 reviewed record(s). Based on a dated review of the prosecutor's public page, which is the office's own publication channel.
- Not penalized or credited
- Absence is scored only after a dated, documented search of the office's and the county's public sites. An unsearched site is unassessed, not zero.
D2 Published FOIA procedures, forms and fee policy — 20% of category · 2% of overall
Question. Are the FOIA procedures, request form and fee policy posted and current? Published procedures are also read for fee-waiver and indigency instructions. Published procedures are also read for fee-waiver and indigency instructions. Published procedures are also read for fee-waiver and indigency instructions. Published procedures are also read for fee-waiver and indigency instructions.
- Score 80–100
- Procedures, forms and fee policy posted, current and easy to find.
- Score 40–59
- Posted but incomplete or out of date.
- Score 0–19
- Not posted after a documented search.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), from a dated search of public sources. At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
D3 Proactive release of frequently requested records — 18% of category · 1.8% of overall
Question. Are categories of records that are requested repeatedly released without a request?
- Score 80–100
- Frequently requested categories are published.
- Score 40–59
- Some are published.
- Score 0–19
- None are published although the same records are repeatedly requested and released.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
D4 Disclosure about data systems — 14% of category · 1.4% of overall
Question. Does the office publish what systems it uses, how long records are kept, and what data fields exist?
- Score 80–100
- System, retention schedule and data fields are publicly described.
- Score 40–59
- Partial description.
- Score 0–19
- Nothing published.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), from a dated search of public sources. At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
D5 Machine-readable datasets and downloadable archive — 14% of category · 1.4% of overall
Question. Does the office publish machine-readable datasets, historical reports or an organized, dated public document archive?
- Score 80–100
- Machine-readable datasets, historical reports and a dated, described, downloadable archive are published and reachable.
- Score 40–59
- Some reports or documents are published, but they are not machine-readable or not organized as an archive.
- Score 0–19
- No machine-readable dataset, historical report or organized archive is published.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 1 reviewed record(s). Based on a dated review of the public site.
- Statutory reference
- MCL 15.234(5)
- Not penalized or credited
- A document is not counted as published merely because it exists; it must be reachable and described.
D6 Requests for already-published information — 12% of category · 1.2% of overall
Question. Where a request sought information already public, did the published material actually answer it?
- Score 80–100
- Requests for already-public information were answered by pointing to published material that fully answered them.
- Score 40–59
- Some requests overlapped published material, but the published material only partly answered them.
- Score 0–19
- Requests for already-public information were treated as new work without checking the published material.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). Requires verifying that the published material answered the request, not merely that it existed.
- Not penalized or credited
- A request is not assumed unnecessary because related information appeared on a website.
E. Denial Justification and Appeals Accountability 20% of overall
E1 Statutory basis cited — 20% of category · 4% of overall
Question. Does each denial or partial denial name the specific statutory exemption relied on?
- Score 80–100
- Each denial cites a specific provision.
- Score 40–59
- Some cite a provision; others give a general reason only.
- Score 0–19
- Denials give no statutory basis.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(5)(a); MCL 15.243(1)(a)
E2 Explanation of how the exemption applies — 20% of category · 4% of overall
Question. Does the office explain why the exemption fits the record withheld, and does the explanation fit what was released?
- Score 80–100
- Explanation is specific to the material and consistent with the release.
- Score 40–59
- Explanation is generic or fits only part of what was withheld.
- Score 0–19
- No explanation, or an explanation contradicted by the office's own release.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- Whether the withholding was lawful is a legal question the index does not decide. The score reflects whether the justification is stated and coherent.
E3 Completeness of the Notice of Denial — 12% of category · 2.4% of overall
Question. Is the Notice of Denial complete: reason box, statutory subsection, description of separated material, and boxes that agree with the letter?
- Score 80–100
- Every field that applies is completed and agrees with the letter.
- Score 40–59
- Some fields blank or inconsistent with the letter.
- Score 0–19
- Form blank, contradictory, or absent where a denial was issued.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.235(5)
E4 Appeal notice, handling and timeliness — 18% of category · 3.6% of overall
Question. Was the right of appeal explained, and when an appeal was decided, was it handled as MCL 15.240 contemplates?
- Score 80–100
- Appeal rights explained accurately; any appeal decided with reasons and within the periods MCL 15.240 sets, allowing for the rule that a board is not deemed to receive an appeal until its next regular meeting.
- Score 40–59
- Appeal rights explained but an appeal decision gave limited reasons or its timing is unclear.
- Score 0–19
- Appeal rights omitted, or an appeal decided without reasons or outside the statutory periods.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 1 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Statutory reference
- MCL 15.240(1)(a); MCL 15.240(2); MCL 15.240(3); MCL 15.240a
- Not penalized or credited
- Where an appeal was filed but its record has not been located, the status is N/A. Absence of an appeal record is not evidence that no appeal exists and is not scored as a failure.
E5 Consistency across similar requests — 12% of category · 2.4% of overall
Question. Are similar requests treated alike, or is the same kind of record certified nonexistent in one response and produced in another? Consistency includes whether nonexistence claims agree with other evidence in the record. Consistency includes whether nonexistence claims agree with other evidence in the record. Consistency includes whether nonexistence claims agree with other evidence in the record. Consistency includes whether nonexistence claims agree with other evidence in the record.
- Score 80–100
- Similar requests receive consistent answers, or differences are explained.
- Score 40–59
- Occasional unexplained differences.
- Score 0–19
- Repeated unexplained reversals on the same kind of record.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- Different outcomes can be lawful where facts differ. The score reflects whether the difference is explained.
E6 Denial rates and nonexempt-portion consideration — 8% of category · 1.6% of overall
Question. What share of requests were denied or partly denied, and did the office show that nonexempt portions were considered for release?
- Score 80–100
- Denial rates are documented and each denial shows that nonexempt portions were considered and released where possible.
- Score 40–59
- Denial rates can be computed, but nonexempt-portion consideration is not documented.
- Score 0–19
- Denials are recorded without any indication that nonexempt portions were considered.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.244; MCL 15.243(1)
- Not penalized or credited
- A high denial rate is not itself misconduct; it is a descriptive measure.
E7 Fee challenges and appeal outcomes — 10% of category · 2% of overall
Question. Were fee challenges documented separately from access appeals, and did any appeal change the fee or the disclosure outcome?
- Score 80–100
- Fee challenges and access appeals are documented separately, with outcomes and any resulting change explained.
- Score 40–59
- Appeals are documented, but fee challenges are not separated from access appeals or outcomes are incomplete.
- Score 0–19
- No appeal or fee-challenge outcome is documented where one was filed.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 1 reviewed record(s), drawn from at least 1 distinct FOIA request(s). An upheld denial is not proof of misconduct and a reversal is not proof of wrongdoing.
- Statutory reference
- MCL 15.240; MCL 15.240a
- Not penalized or credited
- The absence of a fee challenge is not scored as a failure; only documented outcomes are scored.
F. Public Communication 10% of overall
F1 Clarity and accuracy of written communications — 24% of category · 2.4% of overall
Question. Are response letters clear, internally consistent and accurate about what they enclose?
- Score 80–100
- Letters are clear and accurate; enclosures match what the letter says.
- Score 40–59
- Occasional errors or ambiguities.
- Score 0–19
- Letters are contradictory or misdescribe what was enclosed.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- Typographical errors alone are not scored.
F2 Identified contacts and clarification — 18% of category · 1.8% of overall
Question. Is a named contact given and does the office engage when a request needs clarification?
- Score 80–100
- Named contact, working channel, and documented clarification where needed.
- Score 40–59
- Contact given but no documented engagement.
- Score 0–19
- No contact and no engagement.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 2 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
F3 Consistency between public statements and records — 20% of category · 2% of overall
Question. Do the office's statements to the press and public square with what its own records and responses show?
- Score 80–100
- Public statements are consistent with the records.
- Score 40–59
- Some statements cannot be reconciled with the records.
- Score 0–19
- Public statements are contradicted by the office's own records.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 2 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- A press statement is an official statement, not a documented fact. It is weighed against the record, not assumed true or false.
F4 Correction and explanation of errors — 14% of category · 1.4% of overall
Question. When an error or inconsistency is identified, does the office correct or explain it?
- Score 80–100
- Errors are corrected or explained in writing.
- Score 40–59
- Errors acknowledged informally or partially.
- Score 0–19
- Documented errors are neither corrected nor explained.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s), drawn from at least 1 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement; analysis, inference and unresolved questions cannot establish sufficiency alone.
- Not penalized or credited
- Unassessed unless an identified error and the office's response to it are both documented.
F5 Cost and options communication — 14% of category · 1.4% of overall
Question. Did the office explain its costs, waiver and appeal options, less-costly alternatives, and ways to narrow the request?
- Score 80–100
- Costs are explained clearly, waiver and appeal options are stated, and less-costly alternatives or narrowing options are offered.
- Score 40–59
- Some cost explanation is given, but waiver, appeal, alternatives or narrowing options are missing.
- Score 0–19
- Costs are stated without explanation and no waiver, appeal, alternative or narrowing option is communicated.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 3 reviewed record(s), drawn from at least 3 distinct FOIA request(s). At least one cited record must be a documented fact or an official statement.
- Statutory reference
- MCL 15.234(2); MCL 15.234(4); MCL 15.240
- Not penalized or credited
- A waiver that was not requested is not scored as a failure to offer one.
F6 Public FOIA instructions and accessibility — 10% of category · 1% of overall
Question. Are clear public FOIA instructions, contacts and fee procedures published and easy to find?
- Score 80–100
- Clear instructions, contacts, forms and fee procedures are published together and are easy to find.
- Score 40–59
- Instructions and contacts exist, but fee procedures or forms are hard to find or incomplete.
- Score 0–19
- No usable public FOIA instructions are published.
- Between anchors
- Scores between the anchors (20-39 and 60-79) are interpolated by the reviewer and the reasoning is written into the evidence record.
- Evidence needed
- At least 2 reviewed record(s). Based on a dated review of the public site.
- Statutory reference
- MCL 15.234(4); MCL 15.234(5)
- Not penalized or credited
- A published procedure is credited only if it is reachable and current enough to use.
Adding documents, corrections and history
This site is static. There is no database and no live backend, so nothing re-scores itself. When a FOIA document is added, a maintenance script run by the publisher scans the document directory and the request records, lists the new or changed files, and flags the indicators they may bear on for human review. It does not change a status or a score. A document that cannot be classified is flagged for manual review. The prior assessment stays in place until the new evidence has been reviewed and the assessment file has been updated by hand. Only then are coverage and scores recalculated and a new snapshot published.
Published snapshots are never overwritten. Each is appended to the history archive with its score, status, category scores and original weights, coverage, assessment period, cutoff date, methodology version and a reference to the evidence it rested on. Each snapshot carries a hash that includes the one before it, so a later edit to an earlier snapshot is detectable. A correction is published as a new snapshot with a note saying what changed; the earlier one remains visible.
When the methodology changes, its version number is incremented and the change is recorded below. Snapshots made under different versions are marked as not directly comparable whenever the change affects how a score is produced.
| Version | Date | Comparable with earlier? | Change |
|---|---|---|---|
| v1.2.0 | October 25, 2026 | Yes | Admits the five published case watches into the evidence base as `case_watch` sources and adds the rule governing when a case watch may be cited. No indicator, weight, band or threshold changes, so scores remain comparable with v1.1.0. Eight indicators were re-reviewed against the new evidence: six moved, two held. |
| v1.1.0 | October 25, 2026 | No | Expanded from 26 to 37 indicators inside the same six categories and fixed category weights. Adds cost, complexity, timeline, production, migration, proactive-disclosure, denial and communication measures, and rebalances each category's internal weights to 100. Because the scoring structure changed, scores are not comparable with v1.0.0. No v1.0.0 snapshot had been published; the baseline was 'Insufficient Evidence to Score'. |
| v1.0.0 | October 25, 2026 | First release | Initial release: six categories with fixed weights, twenty-six indicators, evidence-coverage rules and rating bands. |
Data files
Everything on this page and in the widget comes from static files that anyone can download and recalculate: methodology, current assessment, score history and FOIA evidence index.