Standard Tier Module
GTI - GeoNum Trust Index (Content)
Score any content's integrity in your browser - and get an embeddable, tamper-evident verification badge.
See it run - a worked example, 100% in this browser tab
The problem
Everyone wants to gauge whether content is original, formulaic, or padded with unverifiable claims - but the tools that promise it are black-box 'AI detectors' that are unreliable, uninspectable, and upload your text to a server to do it.
The local-first solution
GTI computes the measurable, information-theoretic signals of a piece of text - distinctiveness, originality, vocabulary richness, burstiness, non-repetition - plus a claim-checkability readout, deterministically and on-device. The score is bound to the content's hash, so anyone can reproduce it and any edit is tamper-evident, and it emits an embeddable shields-style badge.
What it does
Distinctiveness/entropy, originality (compressibility), vocabulary richness, burstiness, non-repetition
Composite GTI score (0-10) with a low / medium / high band
Claim-checkability readout: quantitative claims stated without a basis, flagged for review
Content-hash binding: the score is reproducible and tamper-evident
Embeddable shields-style badge + a public verify page
GeoNum trust verdict on the numeric core
Honest scope
GTI is a deterministic content-integrity / distinctiveness signal, NOT a verdict on truth, accuracy, or human-vs-AI authorship - no reliable AI detector exists, so it surfaces the measurable structural signals only and leaves the judgment to you. Its only trust claim is reproducibility plus tamper-evidence.
Authorities cited
- Shannon, C. E. (1948). A Mathematical Theory of Communication. Bell System Technical Journal 27, 379-423, 623-656. DOI 10.1002/j.1538-7305.1948.tb01338.x. - Order-0 entropy H = -sum p log2 p, the information-density basis.
- Ziv, J., Lempel, A. (1977). A Universal Algorithm for Sequential Data Compression. IEEE Trans. Information Theory IT-23(3), 337-343. DOI 10.1109/TIT.1977.1055714. - The LZ77 sliding-window back-reference compressor used for the originality (1 - compression ratio) signal.
- Guiraud, P. (1960). Problemes et methodes de la statistique linguistique. Reidel. - Root type-token ratio (types / sqrt(tokens)), the length-corrected lexical-richness index.
- Goldberg, D. (1991). What Every Computer Scientist Should Know About Floating-Point Arithmetic. ACM Computing Surveys 23(1), 5-48. DOI 10.1145/103162.103163. - Conditioning of summation, the basis for the GeoNum drift readout.
- NOTE on the honesty frame: there is no peer-reviewed, reliable detector of AI-generated or false text; published evaluations of AI-text detectors report high error rates and bias. GTI therefore makes NO authorship or factuality claim - it reports only deterministic structural measures and a reproducible / tamper-evident content hash.
Score your content and get a badge
Paste any text, get its GTI plus an embeddable badge that links to a verify page - all in the browser, nothing uploaded. Put the badge on your post, article, or repo.