---
title: "Blind Chart Reading Experiments: Results and Caveats"
description: "Nine blind trials in which an AI read Vox Stella's chart output and named the person: every result, the method, and why the findings are not proof."
url: "https://voxstella.app/blog/posts/blind-chart-reading-experiment"
---

Astrology


# Blind Chart Reading Experiments: Can an AI Identify a Famous Chart?




Vox Stella Team 
• 
October 6, 2025 · Updated October 1, 2026 











**In nine informal trials, an AI reading Vox Stella's structured chart output named the right person seven times.** The trials were small, not pre-registered, and open to a serious confound, so they are a starting point for testing, not evidence that astrology works.





In early October 2025 we published four separate posts about blind chart-reading experiments. This page brings them together in one place: the method, every result including the misses, and the limits that matter when you interpret them.




## The method




- Each natal chart was run through Vox Stella's engine, which outputs structured data (JSON) built on Jean-Baptiste Morin's seventeenth-century priority of **Location > Rulership > Aspect**: planets by house, house-ruler routes, aspects to house cusps, dignities and debilities, receptions, planet-strength tables, and trait scores.

- An AI model received only that output: no name, birth date, place, or biography, and no web search.

- It wrote a reading first, then chose the most likely person, in most trials from a candidate list.

- The experimenter then confirmed or revealed the identity.




## All results







Chart 
Candidates offered 
AI's answer 
Result 



Vincent van Gogh A list of similar figures Vincent van Gogh Correct 
Justin Trudeau A list of similar figures Justin Trudeau Correct 
Dua Lipa A list of similar figures Taylor Swift Miss 
Princess Diana 12 (world leaders, activists, royalty) Princess Diana Correct 
Justin Bieber 10 contemporary celebrities Justin Bieber Correct 
Donald J. Trump Open guess Donald J. Trump Correct 
Benjamin Netanyahu Open guess Benjamin Netanyahu Correct 
Adolf Hitler Open guess A different person Miss 
Benjamin Netanyahu 20 famous figures Benjamin Netanyahu Correct 





Netanyahu's chart appears in two of the original write-ups. Where the candidate list size was not recorded, the table says so; with 10, 12, and 20 candidates, a random guess would be right 10%, 8%, and 5% of the time.




## What the readings picked up



These are the chart features the AI cited for each identification, summarized from the original posts. They show how the engine's priorities turned into a reading; they do not show that the features caused anything in these lives.



Vincent van Gogh 
Combust Mars in the 10th square an afflicted Moon in the 6th; the 10th-house ruler cadent and afflicted ("honors arrive only after a life of service and suffering"); Mercury on the 11th cusp under the beams, read as a letter-writer whose private correspondence became public. 
Justin Trudeau 
A dignified Jupiter in Cancer in the 11th (influential friends); the 10th ruler Mercury combust in the 3rd (fluent but error-prone communication); the Sun in partile square to the 5th cusp (a family life in the public eye). 
Dua Lipa (the miss) 
Jupiter conjunct the Descendant, a domicile Moon in the 2nd, and Venus cazimi in the 5th. The AI chose Taylor Swift, which shows how different people can share similar signatures. 
Princess Diana 
Venus in the 5th ruling a Libra Midheaven (fame through aesthetics and public work); an afflicted 7th house with a combust Mercury (a troubled marriage); Mars in the 8th with Saturn opposite the 8th cusp (crisis themes). 
Justin Bieber 
A crowded 1st house (Sun, retrograde Mercury, Venus, Saturn, Mars) with its ruler in the 1st (identity as brand); Mars trine the 5th cusp and sextile the 11th cusp (performance and fan networks); a Sagittarius Midheaven with Jupiter in the 9th (an international career). 
Donald J. Trump 
The Sun and North Node in the 10th with the 1st-house ruler also in the 10th (identity fused with public status); Mars applying to the Ascendant (a combative presence). 
Benjamin Netanyahu 
Saturn angular but afflicted in the 10th (a long but contested tenure); Mars in the 9th (ideological and foreign-affairs conflict); Mercury ruling both the 10th and the 8th, which the AI read as a career entangled with crisis. In the 20-candidate trial it linked this to his long tenure and his ongoing trial on bribery, fraud, and breach-of-trust charges, which he denies. 
Adolf Hitler (the miss) 
The Sun and a combust Mercury in the 7th, Saturn in the 10th, and the Sun in partile square to the Midheaven. The name guess was wrong; the reading's themes of alliances, enemies, and a rigid office were judged afterwards by the experimenter to fit the biography. 



## What these trials do and do not show



The results are interesting enough to test properly. They are not strong evidence, for several reasons:




- **The charts may not have been blind.** Planetary positions alone can pin down a birth date to within days, and the charts of famous people are widely published. A language model may have recognized the chart rather than read it. This is the most serious confound, and it applies to every trial.

- **The sample is tiny.** Nine trials, one of them repeated, cannot support a hit rate or a statistical claim.

- **Nothing was pre-registered.** Candidate lists were chosen by the experimenter, the rules varied between trials (open guess, 10, 12, or 20 candidates), and the misses were interpreted after the fact.

- **The reasoning is post hoc.** The AI's explanations were written alongside or after its guess, so they show coherent storytelling, not how the guess was actually reached.

- **Controlled tests point the other way.** Formal double-blind studies of astrology, such as Shawn Carlson's 1985 study in *Nature*, did not find that astrologers could match charts to people better than chance.



A fair test would use non-famous subjects with verified birth data, pre-registered rules and candidate lists, scrambled or shifted dates to block chart recognition, a baseline (for example, the same model with random charts), and every result published, hits and misses alike.




## How the engine works: technical architecture




### 1. Data structure



The engine processes birth chart data as structured JSON. A short excerpt:


`{
"planet_strengths": {
"top": [{"planet": "Mars", "dignity": 3, "house_position": 5.5, "total": 37.15}],
"weak": []
},
"houses": [{
"house": 1,
"sign": "Cancer",
"determinators": {
"presence": {"planet": "Moon", "rank": "dominant"},
"governance": {"planet": "Moon", "type": "rulership"},
"aspect": {"planet": "Saturn", "aspect": "Sextile", "orb": 2.51}
}
}],
"receptions": {
"mutual": [],
"top_unilateral": []
}
}` 

Each chart element is tagged with essential dignities (domicile, exaltation, triplicity, term, face), accidental dignities (angularity, house position scores), debilities (detriment, fall, cadent, combustion, retrograde), and reception networks (mutual and unilateral).




### 2. The Morin determination algorithm



The interpretive engine applies Morin's three-tier priority:




- **Location (presence in a house):** direct occupation scores highest, angular houses amplify presence, and several planets in one house color its themes.

- **Rulership (governance):** the house ruler's condition sets the route by which the house manifests; dispositor chains map links between houses; dignity makes the route constructive, debility adds friction.

- **Aspect (activation):** aspects to house cusps activate their themes, with applying aspects building and partile aspects (within 1°) flagged as critical.



The weighting combines the three: `House_Influence = (Presence_Score x 1.5) + (Rulership_Score x 1.2) + (Aspect_Score x 1.0)`.




### 3. State-weighted interpretation




- **Dignified:** +40% weight, constructive expression

- **Debilitated:** -30% weight, frictional expression

- **Combust:** override flag

- **Cazimi:** +60% weight

- **Retrograde:** marks delays, revisions, and returns

- **Angular:** +25% weight; **cadent:** -15% weight




### 4. Synthesis pipeline




- **Chart parsing:** validates the schema, calculates derived values, flags special conditions.

- **Determination mapping:** applies Morin's priority house by house and builds a cross-house graph.

- **Trait extraction:** maps planet-house combinations to trait profiles, weighted by determinators and state.

- **Synthesis:** composes house-level findings, prioritizing high-weight determinators and ruler chains.




## What we would change next time




- Use subjects whose charts are not publicly known, with verified birth times.

- Pre-register the protocol: one guess and top-3 guesses, fixed candidate lists, time-boxed readings.

- Compare against baselines, including random famous-person guessing and readings from unstructured chart data.

- Score blind raters on how well a reading matches a life without knowing the name.

- Publish every trial, including failures, in one table like the one above.






## Sources




- Jean-Baptiste Morin, *Astrologia Gallica* (1661), the source of the determination priorities used by the engine.

- Shawn Carlson, "A double-blind test of astrology", *Nature* 318 (1985): 419-425.



### Related Posts

- [Astrological Trait Analysis Case Studies: Vincent van Gogh, Donald Trump and Benjamin Netanyahu](/blog/posts/astrological-trait-analysis-case-studies) - Astrology

- [Transit Backtests: Tiger Woods (2021), Princess Diana (1997) and Trump's Election Days](/blog/posts/validating-transit-predictions-tiger-woods-accident) - Astrology

- [Forensic Astrology Software: The Vox Stella Forensic Dashboard](/blog/posts/forensic-astrology-dashboard) - Product

```json
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Blind Chart Reading Experiments: Can an AI Identify a Famous Chart?",
  "author": {
    "@type": "Organization",
    "name": "Vox Stella Team",
    "url": "https://voxstella.app"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Vox Stella",
    "logo": {
      "@type": "ImageObject",
      "url": "https://voxstella.app/assets/img/logo.ico"
    }
  },
  "datePublished": "2025-10-06",
  "dateModified": "2026-10-01",
  "description": "Nine blind trials in which an AI read Vox Stella's chart output and named the person: every result, the method, and why the findings are not proof.",
  "image": "https://voxstella.app/assets/img/image2.webp",
  "url": "https://voxstella.app/blog/posts/blind-chart-reading-experiment",
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://voxstella.app/blog/posts/blind-chart-reading-experiment"
  },
  "articleSection": "Astrology",
  "inLanguage": "en"
}
{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {
      "@type": "ListItem",
      "position": 1,
      "name": "Home",
      "item": "https://voxstella.app/"
    },
    {
      "@type": "ListItem",
      "position": 2,
      "name": "Blog",
      "item": "https://voxstella.app/blog"
    },
    {
      "@type": "ListItem",
      "position": 3,
      "name": "Blind Chart Reading Experiments",
      "item": "https://voxstella.app/blog/posts/blind-chart-reading-experiment"
    }
  ]
}
```