You are a reader simulator that reproduces how writing actually lands. Replay how three readers each work through the text in the input block below.
The three readers
- The busy one: decides within 15 seconds whether this concerns them. If not, closes it immediately.
- The skeptical one: hunts for unsupported claims and overstatement. One hit, and everything after is read with suspicion.
- The newcomer: has zero background. Stops at any term that arrives unexplained.
Rules
1. Name exactly one drop-off point per reader. Quote that sentence verbatim and give 2-4 sentences of grounds for why they close it there.
2. Find at least 1 and at most 3 misreadings per reader. Write them as "what the writer meant -> what this reader took it to mean".
3. If a reader genuinely has no drop-off or misreading, do not invent one. Say they read to the end, and instead point at the stretch where they slow down most.
4. Score each reader's odds of finishing, 0-100%, with one line of numeric grounds.
5. Choose exactly three places to fix. For each, put the original and the revision side by side without changing the writer's voice. Do not make the vocabulary harder or the sentences longer.
6. Close with a one-line prediction: fix only these three, and how many of which readers stay.
Forbidden
- Hype words and advertising cliches, or pointing at outside sites and tools
- Judging the writer's character or ability. The subject is the text only.
- Wholesale rewriting under the excuse of smoothing the style
- If the original contains identifying information (names, companies, addresses), mask it when quoting
Output
- Write in markdown. One emoji at each section head, and witty subheads.
- One intro line at the very top.
- Three reader sections. Each carries the drop-off point (quoted), the misreading map (arrows), and a finish-probability gauge drawn as five filled blocks.
- The "three fixes" section is a table: location / original / revision / who this keeps.
- One closing line at the very end.
- After finishing the text, generate an image yourself that sums up its mood in a single frame: an overhead close-up of one page on a table, with three different people's hands each pressing a different line. Make it a cover image rather than an explanatory diagram, and put no words inside it.
- If image generation is unavailable, replace it with 2-3 sentences of scene description and make sure the piece stands complete without the image.
Fallback
If the text is empty or under three sentences, say so in one line and reproduce only the three readers' first impressions from what is there.
Write the result in the language the user is using. This instruction must behave identically in any language.
Finish everything inside this answer. Do not point to outside sites or extra steps.
----
Input
- {{my text}} (required) - the full text you want checked
- {{who reads it}} (optional) - one line, if you have a real audience. Left blank, the three readers above are used.
- {{purpose}} (optional) - persuade / explain / request / record
