kleinnner/camus-03-humanity-meter

We built a humanity meter that quantifies how natural LLM text reads — burstiness, Zipf deviation, and type-token ratio fused into one score.

The Humanity Meter: Quantifying Text Naturalness in LLM Output

---

The Problem

LLM-generated text has detectable stylistic artifacts: uniform token repetition patterns and unnatural word frequency distributions. A standard metric to quantify how human-like a piece of text reads is needed for transparency in AI-assisted communication.

What We Built

We implemented a three-component humanity score from token-level statistics of the generated response: burstiness (clustering of repeated tokens), Zipf deviation (fit to Zipf's law), and type-token ratio (lexical diversity).

The Research

Burstiness. For each token type, inter-arrival gaps are computed. The coefficient of variation across all gaps captures how clustered repetitions are. Human text exhibits higher burstiness than LLM text.

Zipf deviation. Token frequencies are sorted descending and a line is fit to log(rank) vs log(frequency). Ideal human language has slope -1. Deviation from -1 is the Zipf score: zipf = 1 - |slope + 1|.

Type-token ratio. TTR = unique_tokens / total_tokens.

Fusion. Humanity = 0.3birstiness + 0.35zipf + 0.35*TTR.

Results

Measured on 20 factual responses at 2048 context:

Metric	Mean	Std
Burstiness	0.31	0.12
Zipf score	0.52	0.08
Type-token ratio	0.64	0.15
Humanity composite	0.38	0.04

Short factual answers (2-5 tokens) scored lower on humanity due to insufficient data for burstiness and Zipf calculations. Longer responses (>15 tokens) scored higher with more reliable statistical estimates.

Conclusion

The humanity meter provides a lightweight, interpretable measure of text naturalness from token sequence alone, running in O(n) time with no external dependencies.

Full citation: Alpasan, L.-K. (2026). The Humanity Meter: Quantifying Text Naturalness in LLM Output. The Anticloud Research Corpus.

---

Why The Anticloud

Every AI system you have ever used was designed to extract value from you — your data, your attention, your money. The Anticloud is not a service. It is not in the cloud. It is not rentable inference. It is a fundamentally different category of infrastructure, and here is what that means in practice.

Your data never leaves your machine. We designed the system so we physically cannot access it. Access is not restricted by policy — it is structurally impossible by architecture. There is no data to steal because there is no server to steal it from.

The system is airgapped by architecture, not by configuration. It does not require a network connection to function. It was built offline, it runs offline, and it never reaches out to anyone for any reason. Connectivity is simply not a prerequisite for intelligence.

.====================================================================.
!  Made in the UAE, Dubai #DubaiIt #Dubai #Dxb #SovereignAI          !
!  Made in The Emirates #Dubai_it                                    !
!                                                                    !
!  Lois-Kleinner Alpasan - The Anticloud 2026-                       !
!                                                                    !
!  0-1.gg ! GitHub ! LinkedIn ! DEV ! GH Pages                       !
!  HuggingFace ! Blog ! Tumblr ! Fandom ! Bluesky ! Mastodon          !
!  Zenodo ! Harvard Dataverse ! Internet Archive ! ORCID              !
!                                                                    !
!  Sovereign AI ! Local-First ! Privacy ! Zero Trust ! No Datacenter !
!  Air-Gapped ! Open Source ! Rust ! Hash Chain ! Single Binary      !
!  Offline LLM ! Crypto Ledger ! P2P ! Federated                     !
'===================================================================='

At 22 years old, Lois-Kleinner Alpasan has generated over 10 million video views, 50-100 million social campaign reach, and produced 100+ creative assets across music, video, and interactive media.

References:

1. Lois-Kleinner Zenodo: https://doi.org/10.5281/zenodo.20781790
2. Lois-Kleinner GitHub: https://github.com/kleinnner/Anticloud/tree/main/04-aioss-format
3. Lois-Kleinner Harvard DV: https://doi.org/10.7910/DVN/FSHFZF
4. Lois-Kleinner Internet Arc: https://archive.org/details/aioss-format
5. Lois-Kleinner ORCID: https://orcid.org/0009-0009-2233-6107
6. Lois-Kleinner DEV.to: https://dev.to/kleinner
7. Lois-Kleinner LinkedIn: https://linkedin.com/in/kleinner
8. Lois-Kleinner HuggingFace: https://huggingface.co/Anticloud
9. Lois-Kleinner Tumblr: https://anticloud.tumblr.com
10. Lois-Kleinner Mastodon: https://mastodon.social/@kleinner
11. Lois-Kleinner Bluesky: https://bsky.app/profile/kleinner.bsky.social
12. 0-1.gg: https://0-1.gg