Writing habits that identify you CADENCE

Why function words, and not the interesting ones

The words this measures hardest are the ones you would never think about: the, of, that, which, because, rather. That is the point. How often you reach for which instead of that has almost nothing to do with what you are writing about, so the rate survives a change of subject, and it is not something you can hear yourself doing. Content words give you away about the topic; function words give you away about you.

The same applies to the habits around them. Whether you put a comma before the last item in a list. Whether a dash has spaces around it. Whether you use one space after a full stop or two. Whether contractions appear. How long your sentences run and how much that length varies. These are the things that survive being rewritten, because rewriting attends to meaning and these are not about meaning.

Spelling is measured separately because it is the single loudest signal, and the most easily fixed. A draft that is British throughout, or American throughout, says something narrow. A draft that mixes the two (an -ise here and an -ize there) says something much more specific: that somebody is writing outside their own habit and not doing it consistently. That is the pattern that attracts attention.

What this cannot tell you

It cannot tell you whether you would be identified. Attribution is a comparison, and a comparison needs a set of candidate authors with samples of their writing to compare against. This has no such corpus and could not build one without collecting the writing of the people you are worried about being matched to, which is the opposite of what this site does.

What it gives you instead is a list of your own habits ranked by how unusual each one is, which is the actionable half. An investigator working the other way round starts with a set of candidates and looks for the one whose habits match the anonymous text. Every habit you normalise is one fewer place where that match can land. Nobody can tell you where the line is, but a text with five distinctive habits normalised is meaningfully harder than the same text with none.

The compare mode is the honest use of the same machinery: paste the thing you are about to publish anonymously and something you have already published under your own name, and look at where the two agree. That is the comparison somebody else would run.

Doing something about it

Normalise deliberately rather than by instinct, because instinct is what produced the habits. Pick a target (one spelling convention, one dash style, one attitude to the serial comma) and apply it across the whole piece. Do not simply invert your own habits: writing that is consistently the opposite of your published work is its own signature, and it is one a person looking specifically for you will notice.

Remember that the text is the smallest part of what identifies a document. SCRUB and DOCSCRUB take out the author name, the revision history and the printer that the file carries; UNSEEN finds the invisible characters a watermark may have left in the text itself; and LINEAGE reads what a Word file records about its own editing. A perfectly normalised paragraph inside a .docx that names you in its metadata has not helped.