Argument originality, mapped
Map your thesis against what ordinary search and standard AI prompts readily produce — then see which claims, connections and implications meaningfully depart from the baseline.
Private by default. Not an AI-writing detector. Your draft is never published merely because you analyze it.
A map of one analysed argument, from the essay “From Impact to Change: The Evolution Most Nonprofits Never Make”. At the centre is its thesis: Nonprofits excel at delivering immediate impact but rarely evolve toward systemic change, because their structures are designed for linear growth rather than network effects — leaving their closeness to affected communities uncoordinated and underused. Around it are two clusters of what is already easy to reach — a search frontier of positions that ordinary public searches surface, and a model frontier of positions that standard prompts generate. An irregular boundary encloses both, marking what the run found readily accessible. One claim sits outside that boundary, connected back to the thesis: The engine with the most potential to drive human-centered change is the one least equipped to do it. The essay scores Idea Originality 64 and Idea Quality 45 out of 100, with search accessibility easy and model accessibility level 5. These are the results of one live analysis of one published essay, not a claim about what any other draft would score.
Search frontier
Model frontier
Nonprofits are built for impact, not for systemic change.
Point of originality
Level 5 — not produced within the tested baseline. Idea Quality, scored independently, came back 45/100.
An argument sits in a field of what is already easy to reach. IsThisOriginal maps where the draft follows those familiar paths, and where one claim departs from them. Select a layer to read what it contains.
Three stages, in order. Every number a report publishes is produced by inspectable code over stored evidence — models supply observations, never scores.
01
Identify the thesis, supporting claims, mechanisms, synthesis and implications — each anchored to a passage in the writer's text.
Extracted thesis
Nonprofits are built for impact, not for systemic change.
Example analysis
02
Compare the argument with public-search evidence and standardized model generations, assembled into a corpus built for this draft.
Search accessibility · easy
Ordinary searches readily surface this territory, so the retrieved sources suppress rarity across most of the argument.
Example analysis
03
Show where the argument is conventional, where it becomes distinctive, and how difficult it was for the baselines to reproduce.
Point of originality · model level 5
The engine closest to the change is the one least equipped to drive it.
Example analysis
Example analysis · Live research · published essay
IsThisOriginalThe report identifies the claim, mechanism, connection or implication that most departs from the tested baselines — and shows the closest precedents it found.
The analysed essay
Designing Intelligence · 28 April 2026
The essay argues that the sector’s structures are built for linear growth rather than network effects, and proposes a four-surface framework — Connect, Advance, Relationships, Exchange — it calls CARE. Then it states the tension the whole piece turns on:
The engine with the most potential to drive human-centered change is the one least equipped to do it.
Level 5 means the baseline did not produce this framing during this run. It is not a claim that no model could, or that no one has written it before.
One live analysis of one published essay. Read it and judge the reading for yourself — that is the only honest way to evaluate a tool like this.
Best of the Best
A high originality score on its own can come from an odd topic. A high model-accessibility level on its own can come from being obscure. Best of the Best asks for both at once — rare against the sources the run retrieved, and not something the baseline models produced.
Listed only if the author asks
Clearing both bars makes the option available. Nothing appears until the person who ran the analysis chooses to list it.
Links out, never republishes
Each entry points at wherever the piece is actually published. The draft text is never reproduced on this site.
Re-checked on every load
Scores move when the scoring version does. A listing that falls below either bar after a re-score stops appearing, without anyone having to notice.
The two bars
Both must hold in the same run.
Idea Originality
needs 75+
The example above scored 64 — below the bar.
Model accessibility
needs L4+
The example reached L5 — clears the bar. Not produced within the tested baseline.
So the essay in the example above is not listed. It clears one bar and misses the other, and both have to hold in the same run.
Unpublished writing is never made public merely because it is analyzed.
Analyzing a draft does not publish it, and it is not consent to publish a report or a badge. Searches are built from the ideas in a draft rather than its verbatim sentences, so running an analysis does not put the text in front of a search provider as content to be indexed.
Results are bounded by the sources searched and the model generations tested at the time of analysis.
Books, paywalled essays, podcasts, videos, private newsletters and non-English writing sit outside every run, so an absent match is weak evidence rather than proof of absence. Every report states the corpus it measured against, and no validation study of the system has been conducted — the scores are research-informed heuristics.
The product evaluates the argument, not whether a person or an AI wrote the words.
Intellectual originality and AI authorship are different questions. A person can write something entirely conventional; a model can be prompted into something unusual. IsThisOriginal makes no attribution claim of any kind, and it is not a plagiarism checker — it compares arguments, not word sequences.
IsThisOriginal opens to a small group of writers at a time, so every report gets the search and model budget it needs.
