Week 1: final_FINAL_revised2

Why humans version so badly even when they are smart, and why the filename, not the file, has been doing the work all along

22 February 2027

Open the slides

final_FINAL_revised2.docx was not written by someone who doesn’t understand version control. It was very likely written by someone who understands the problem perfectly well and has no tool at hand that solves it.

The filename is standing in for something else#

Here is the folder it lives in:

report.docx
report_v2.docx
final.docx
final_FINAL.docx
final_FINAL_revised2.docx

A filename is one string, and by the last entry that string is carrying four separate assertions at once:

• Supersession: this replaces final.docx, which replaced something before it.
• Sequence: revised2 says a second round happened, and implies a first one you are expected to already know about.
• Currency: as of now, this is the one to use.
• Finality: final, twice, asserting that no further version is coming.

None of the four is metadata the filesystem tracks. There is no supersedes field, no current flag, no round counter. So all four go into the name, because the name is the only field in the entire arrangement that everyone involved can see, write to, and agree to read.

That is the whole mechanism, and it explains the shape of the result. The name has no schema, so a claim cannot be corrected, only appended to, which is why these strings grow rightward and never shrink. It has no authority behind it, so final is a claim and not a guarantee. And it is not exclusive, so anyone can file a competing claim in the same field. final_FINAL.docx is that happening once. final_FINAL_revised2.docx is it happening twice.

This is not a failure of discipline. Given a folder, an email thread, and no shared tool for tracking revisions, it is roughly the best available solution. It is also a bad one, and the badness is structural rather than personal.

Versioning is older than version control#

Treat “versioning” and “version control” as different things. Versioning is the practice of keeping track of which state of a work is current, and what changed to get there. It is as old as revision itself: monks correcting a manuscript, an editor marking up a typescript in red pen, a filing cabinet with folders labelled “drafts” and “final.” Version control, software that does this automatically with a shared and enforced record, is barely sixty years old, and most people writing documents today have never opened it. Week 5 covers what came before it. This week is about what people do in its absence.

Where this goes#

Next week asks a question this one keeps dodging: what a file actually is, underneath the name someone gave it. Once you can say precisely what makes two files the same or different, you can ask why anyone would ever need to write revised2 to say so.