How to Check Whether a Public Tax Case Archive Is Complete

An online case page can look complete because it has a title, a status, and a tidy list of documents. Those elements do not establish that every filing, attachment, or later event is present. A source archive may publish only selected records; a copy may have been captured before the next filing; an attachment may fail to load even while its docket row remains visible. Before drawing a conclusion from an apparent absence, examine what the archive actually covers.

This guide offers a research method for describing the scope of a public tax case archive. It does not tell a party whether a filing is required, whether a deadline has passed, or whether an official record is legally complete. For those questions, consult current official sources and qualified advice. The examples here are invented to illustrate how to avoid overstating what an online copy shows.

Define the completeness question first

“Is this archive complete?” has no single answer until you say complete for what purpose. You may want to know whether all attachments shown on a particular docket page are available, whether the archive covers events through a certain date, whether a listed decision can be opened, or whether every document mentioned by an order appears in the file. Each is a different test. A case page with every file from one captured day can still be outdated the next day.

Write the question in an observable form. “Do the four attachment links visible on the source docket for entry 8 each resolve to a readable file?” is testable. “Does this record contain every fact about the dispute?” is not. Many facts may exist outside the published docket, and the public may not have access to everything. A well-defined test also tells you what evidence would change your conclusion.

Start by separating source completeness from archive completeness. The official docket may itself be a public view with limits. An independent archive may reproduce that public view at a point in time. If a file is absent from the archive, you have not yet established whether the official site had it, whether the archive failed to collect it, whether it was removed, or whether it never existed. Keep those possibilities open until you can compare sources.

Avoid a yes-or-no badge in your research notes. A statement such as “all five attachment links listed in the archived docket snapshot opened on October 13” is much more informative than “complete.” It defines both the population and the check date. If you later inspect additional entries, you can add another result without contradicting the first.

Record the archive’s time boundary

Most case pages have several dates: a filing date, docket-entry dates, dates printed on PDFs, and an archive or access date. Only the last group describes when you observed the web copy. A PDF dated in July does not prove that the archive was updated in July; an archive date in October does not prove the case had no later activity. Label each date in your notes.

If the archive displays a captured or last-updated date, record exactly what that label says. Do not assume it applies to every individual PDF. Check the latest docket row actually shown. A useful statement is “the newest visible docket row is dated September 9 in the copy checked October 13.” This says what the page contains without asserting what happened after September 9.

The Michigan Tax Tribunal’s e-filing FAQ explains that users can view filed documents through the Tax Docket Lookup. That is a reason to compare an independent copy with the official lookup when current completeness matters. The public interface and its available files can change. If the official source cannot be reached during your review, write “not compared with the official docket,” not “official docket unchanged.”

Save the URL and access date for each source you compare. A screenshot or private note of the row count and latest entry can help explain a later difference. Do not rely on memory, especially if you work through many cases. The apparent absence of a later row can mean only that it was absent from the view you inspected at that time.

Inventory docket rows and attachment slots separately

A docket row and an attachment are different units. One row may have no file, one file, or several. A continuation row can carry a document associated with an earlier event. Counting rows as files will make an archive seem complete or incomplete for the wrong reason. Create an inventory with one line per visible attachment and preserve its parent docket entry.

For each row, capture the row number or label, event description, displayed date, number of offered attachments, and any note about a continuation. For each attachment, record the title or filename, file type if shown, and whether the link opens. If a ZIP file expands into several documents, list the members while retaining the relationship to the original ZIP and docket row. The goal is to distinguish “no file offered” from “file offered but unavailable.”

If a row offers no attachment, do not treat it as a broken download. The docket may be recording an event, notice, or administrative action without a public file. Conversely, a visible filename is evidence that a file was offered at that point, even if today’s link returns an error. Those two states deserve different labels in your inventory.

When several files have the same visible name, compare their parent rows and content before merging them. Duplicate-looking scans can be genuinely different versions. Two entries may refer to the same document for a legitimate reason. Preserve both associations in your inventory, even if the bytes are identical, so a reader can reconstruct the docket history.

Test whether files open and contain what their labels promise

A successful click is only the first check. The link might open an HTML error page, a zero-page PDF, a placeholder image, or an unrelated file. Look at the response in a browser, confirm that the viewer loads pages, and compare the visible heading with the attachment label. A downloadable file is not necessarily the correct file.

For a long PDF, spot-check the first page, a middle page, and the last page. Check whether page count is plausible and whether pages are blank, clipped, sideways, or duplicated. If the source offered a Word document or image that an archive converted to PDF, inspect the reading copy for lost tables, cropped margins, or missing frames. Conversion can preserve the filename while changing how content appears.

Searchable text is another layer, not the entire completeness test. A scanned PDF can be visually complete yet have no text layer. OCR can produce text while missing a stamp or table. A negative text search does not prove the words are absent from the visible document. For an important claim, inspect the rendered page, and mark any page that is unreadable rather than calling it empty.

If a PDF refuses to open once, retry under ordinary conditions and note the result. A temporary network failure is different from a persistent missing attachment. Do not repeatedly hammer a source. A bounded second check or a comparison with the official source is usually more useful than dozens of identical requests.

Compare the archive against an independent reference

The most direct comparison is the official docket visible at the same time. Match case number, parties, and jurisdiction before comparing rows. Similar names can lead you to the wrong case. Then compare the latest visible row, entry labels, dates, and offered attachments. Record differences without immediately assigning a cause.

For example, imagine an archived case has entries 1 through 12, while the official lookup shows entries 1 through 14. The narrow conclusion is that the archive copy you inspected does not show two entries visible at the official source. It does not establish when the archive last ran, whether the missing entries are public files, or whether the official site changed after a prior archive capture. Those questions need further evidence.

A second independent reference may be a document that explicitly mentions another filing. If an order says “the motion filed August 2,” but you cannot locate that motion in the archive, record a cross-reference gap. The motion could be outside the public view, filed under an unexpected title, or absent from the copy. The order’s language supports that it refers to a motion; it does not prove why the file is missing.

Use official terminology carefully. The Michigan Tax Tribunal glossary distinguishes motions, orders, notices, and decisions. A row title suggesting a motion is not a substitute for opening the motion. A later order may respond to a motion, but not always in the way a casual reader expects. Completeness checks should map the references, not invent the missing content.

Recognize deliberate limits without guessing at the reason

Some material may be withheld, redacted, or excluded from a public view. Personal identifying information may be protected. The Michigan Tax Tribunal glossary identifies categories of information that should not appear in public documents, including certain identification and financial account numbers. An apparent gap can therefore have a privacy-related explanation, but the gap alone does not establish that it was intentionally redacted.

Do not reverse-engineer hidden text or infer private details from surrounding fragments. If the public PDF contains a black box, describe it as a visible redaction. If a page is absent entirely, describe it as absent from the copy you checked. Those are observations; they are not conclusions about what the missing material says. A respectful archive review can identify availability limits without trying to defeat them.

Some docket events may have no public attachment even though parties received communications elsewhere. A case summary can omit context from correspondence or proceedings. Never treat a public archive as a complete portrait of the people named in it. The research question is what is publicly visible and verifiable, not what might exist privately.

If you need an official certification or a legally complete record, an independent website cannot supply that merely by displaying many PDFs. Use the agency’s current process. The public archive remains useful for orientation, comparison, and retrieval, but its completeness has to be described in terms of the materials you actually tested.

Use a gap log that preserves uncertainty

A compact gap log has fields for case number, docket row, expected item, observed behavior, check date, comparison source, and next action. Choose neutral status labels: offered-and-readable, offered-but-unavailable, not-offered, content-mismatch, unreadable-page, or comparison-not-done. These labels describe the evidence better than a red “missing” flag for everything.

Suppose row 6 says “Exhibits” and offers two files. One opens; the other returns a server error. Record one readable attachment and one unavailable attachment, with the date and URL you checked. Do not infer that the unavailable file was never filed. If it later opens, update the row and retain the old observation as part of your review history.

A separate notes column can record ambiguity. “The second attachment may be a duplicate; compare page count and content” is a better working note than deleting it from the inventory. If a filename has changed but the document appears identical, document what you compared. Checksums can help identify byte-for-byte identity, while visual comparison helps when a file has been recompressed or OCR-processed.

Avoid making the public page itself sound more certain than the evidence allows. A site that labels an attachment “preview unavailable” should not simultaneously imply that all documents were successfully collected. If you publish a research finding, include the relevant limits: which source was checked, the date, and which items were unreadable or unverified.

Understand what an empty search result means

An archive search returning no result is not proof that no case exists. The name may differ, the case may be outside the site’s jurisdictions or date range, the source may not expose it publicly, or the index may not yet contain it. Search one or two reliable alternate identifiers, such as an exact docket number, before making a negative claim. Then state the search boundary: “No result appeared in this archive under these terms on this date.”

The same reasoning applies within a PDF. No match for “decision” in extracted text does not prove the visible pages contain no decision. Scans may have weak OCR, and the document may use another term. Open the pages or inspect the docket row that should contain the material. The conclusion should reflect what was checked, not the confidence suggested by a search box.

If the source site has a date range or division filter, include it in the note. A query limited to Michigan Small Claims cases cannot establish what exists in another division or state. Search interfaces are access paths into a collection; they are not declarations that the collection is comprehensive.

Report a scoped finding, then revisit it when needed

A good completeness finding might say: “On October 13, the archived page showed 12 docket entries and 17 offered attachment links. Sixteen opened as readable PDFs; one returned an error. The official lookup showed two later entries that were not visible in the archived copy.” That statement can be checked and updated. It does not claim that every private or public case document has been found.

For a short blog citation, you may not need the full inventory. You do need to avoid phrases such as “the complete case file shows” when you checked only a record summary. When the scope matters, give a brief qualifier and a link to the exact document you used. The reader should know whether you relied on an archive snapshot, an official current docket, or both.

Set a revisit interval based on the question. A historical case with a final decision may need less frequent checking than an active case, but a visible final label alone is not a guarantee that nothing will change. Corrected files, appeals, or site maintenance can alter what is publicly available. Add a new dated observation rather than overwriting the old one.

The practical standard is modest: identify what the site offered, confirm what you could actually read, compare with the relevant official source when currentness matters, and describe remaining uncertainty. That gives readers a clear account of the archive’s coverage without promising a completeness that the web page cannot prove.

For collaborative research, save the inventory and gap log with the date of the check. Another reviewer should be able to repeat the same link tests and see whether the results changed. If a previously broken file opens later, retain both dated observations. That history distinguishes a resolved access problem from an item that was never offered. It also keeps a later article correction honest: the original statement may have been accurate for the copy available at the time, even if a newer copy is more complete.

One useful quality-control step is to have a second person inspect a small sample without seeing your classifications. Ask them to find the same docket row, count its offered files, and open the same attachment. If their result differs, investigate whether you used different filters, a cached page, or a changed source. Agreement on a small sample does not prove the entire archive is complete, but disagreement can reveal a flawed counting method before it is repeated across hundreds of cases.

When reporting a batch result, give both a numerator and a denominator. “Ninety-eight readable PDFs out of 100 offered attachment links” is meaningful; “98 percent complete” is less so if the source might have unlisted material or if some PDFs contain unreadable pages. Say what the denominator covers. This helps a reader understand the finding without mistaking a link-availability score for a legal completeness judgment.