Build a disclosure schedule DOCKET

The job this replaces takes two days and misses things

Building a disclosure schedule by hand means opening every file in a folder, one at a time, writing its details into a spreadsheet, and remembering to look at the properties as well as the contents. It is done under time pressure, usually by whoever is available, and the failure mode is not that somebody does it badly; it is that nobody can read four hundred documents carefully in the time there is.

What actually goes wrong is not the visible content. It is the things a file carries that nobody opened it to see: text deleted under tracked changes and still in the file, comments left attached, an author name that identifies who really wrote it, GPS coordinates in a photograph, a hidden sheet in a spreadsheet, a document that calls out to a server the moment the other side opens it.

So for every file this records the path, type, size, dates and SHA-256, and then opens it far enough to say what it holds beyond what is on the page. The digest matters as much as the rest: it is what lets you show later that the copy you scheduled is the copy you produced.

What it does not decide

It does not read the substance of a document and it does not decide what is disclosable. That judgement belongs to the person making the disclosure, it is a legal judgement, and a tool that offered one would be worse than useless because it would be relied on.

What it produces is a list of what each file contains and which ones warrant a look before they go out. That is a different and more honest claim: it narrows four hundred documents to the eleven with tracked changes and the three with comments, and a person reads those eleven properly instead of skimming all four hundred.

Everything runs in this tab. For disclosure material that is not a nicety: the folder you are scheduling is privileged or confidential by definition, and uploading it to a service to be indexed is the kind of thing that ends up being a separate problem from the case.

Doing it in the right order

Schedule first, then clean, then produce, and keep the three apart. Run DOCKET over the folder as received, so the schedule describes what you actually hold. Then deal with what it found: SCRUB and DOCSCRUB remove the metadata and comments, CLOAK strips a photograph's GPS, FLATTEN burns a filled-in form so it cannot be changed, and DISARM removes anything that reaches out to a server. Then schedule the cleaned set again, because the digests have all changed and the schedule has to match what goes out.

Keep both schedules. The difference between them is the record of what was removed and why, which is the question you will be asked if anybody later notices that a document's properties are not what they were.

Around it: CHAIN logs who held the material, ATTEST certifies the bundle with a hash of every file, EXHIBIT binds exhibits into one PDF with Bates numbers, and STRATA digs the earlier versions out of a PDF that was saved by appending.