On short documents, the free-rounds loop is often all-or-nothing

Charles Li · HumanPen · short-doc-all-or-nothing

On a long thesis chapter, HumanPen’s homepage loop—paid rewrite of flagged continuous prose, re-import if still ≥20%, free rounds on leftover flags until the report shows * under 20%—can move through several leftover spans. On a short document (a few hundred words of qualifying prose: abstract, response paper, tight methods blurb), that same loop often feels all-or-nothing.

Why: after you rewrite the flagged continuous-prose spans, few continuous-prose spans remain. There is little middle ground left for “partial progress.” Either the short spine clears into the product’s * under 20% stop condition on the next import, or the thin remaining prose still sits high. That is product behaviour on short files—not a promise that the detector will pass you.

Distinct from free-rounds-until-under-20 (the billing loop itself), asterisk-under-20-means (what the glyph displays), and pay-by-rewritten-words (what enters the bill). Here the literacy move is short-doc all-or-nothing behaviour of the free-rounds loop.

This note explains the homepage workflow honestly. It does not guarantee a next percentage, school clearance, or “undetectable” output.

Better order on a short file

1. Import the full Turnitin / iThenticate AI writing report; confirm attempt identity and coverage.
2. Count how many continuous-prose spans are actually flagged. On short docs that list is often tiny.
3. Rewrite those flagged spans (paid on rewritten words). Freeze citations, numbers, formulas, tables, and layout.
4. Re-check on your school surface; re-import if still ≥20%. Free rounds apply to remaining flagged spans—which on a short file may be nearly none, or one stubborn block.
5. Expect blunt outcomes: * under 20% stop, or still-high on a thin spine—not a long ladder of mid-range percentages.

Why short files feel binary

Long files keep leftover flagged blocks after a first pass, so free rounds can chip away. Short files often have only one or two qualifying continuous-prose windows. Once those are rewritten, the free-rounds / under-20% loop has almost nothing left to iterate—so the product experience collapses toward all-or-nothing. That matches how little prose the report had to work with; it is not a moral cliff and not a score guarantee.

What this is not

• Not a guarantee the next AI percentage will drop, clear, or match another tool.
• Not “undetectable,” bypass, or whole-file wash advice.
• Not a substitute for your course integrity process if one is opened.

Where HumanPen fits

HumanPen at humanpen.net is built for document-level revision against an imported detector report: rewrite flagged continuous prose, keep citations and layout, then free rounds on leftover flags until * under 20% on the product framing. On short documents, expect that loop to feel binary because few spans remain. No promised next percentage.

Product: humanpen.net. Charles Li · HumanPen.