🧠 STARGATE: CIA Remote Viewing Archive (Raw PDFs + Metadata)

Overview

STARGATE is the most comprehensive open-access archive of declassified CIA documents related to psychic research, remote viewing (RV), and anomalous cognition.

This dataset consolidates over 12,000 scanned PDF files, drawn from decades of classified government programs designed to investigate and operationalize extrasensory perception (ESP) in intelligence-gathering contexts.

Programs Included

• Grill Flame (1973–1978) - DIA's early remote viewing experiments
• Center Lane (1978–1985) - Army Intelligence psychic espionage program
• Sun Streak (1985–1991) - Continuation under new codename
• Star Gate (1991–1995) - Final incarnation before congressional termination
• Related Programs: Gondola Wish, Scanate, Inscom Center Lane

---

🎯 Use Cases

1. Natural Language Processing (NLP)

• Information extraction - Extract dates, locations, agents, targets from session transcripts
• Entity recognition - Identify patterns in how remote viewers describe unknown locations
• Summarization - Generate abstracts of lengthy technical reports
• Text classification - Categorize documents by program, methodology, success rate

2. Historical & Intelligence Research

• Cold War intelligence tactics - How did the U.S. approach psychic espionage?
• Scientific methodology - What protocols were used? How did they evolve?
• Congressional oversight - Why was the program terminated in 1995?
• Cross-agency collaboration - DIA, CIA, Army Intelligence, SRI International

3. Psychological Analysis

• Cognitive bias in perception and reporting
• Training methodologies for perceptual enhancement
• Statistical evaluation of claimed psychic phenomena
• Interviewer-subject dynamics in session transcripts

4. Machine Learning Training

• OCR benchmarking - Scanned 1970s-1990s government documents (challenging quality)
• Document classification - Multi-class categorization (session transcript, technical report, administrative memo, etc.)
• Anomaly detection - Identify unusual patterns or outliers in session data
• Time-series analysis - Success rates, funding levels, program activity over decades

---

📊 Dataset Structure

STARGATE/
├── pdfs/                    # 12,000+ raw scanned PDF files
│   ├── STAR-GATE-01.pdf
│   ├── STAR-GATE-02.pdf
│   └── ...
├── metadata.csv             # Document metadata (title, date, program, classification)
├── processed/               # (Future) OCR-extracted text versions
└── README.md

Metadata Fields

Field	Description	Example
document_id	Unique identifier	STAR-GATE-00042
title	Document title	"Remote Viewing Session: Soviet Submarine Base"
date	Document date	1983-04-15
program	Program codename	Center Lane, Star Gate
classification	Original classification	SECRET, CONFIDENTIAL, UNCLASSIFIED
declassification_date	When it was released	2000-08-09
page_count	Number of pages	23
file_path	Relative path to PDF	pdfs/STAR-GATE-00042.pdf

---

🚀 Quick Start

Load the Dataset

from datasets import load_dataset

dataset = load_dataset("GotThatData/STARGATE")

Example: Extract Metadata for a Specific Program

import pandas as pd

# Load metadata
df = pd.read_csv("hf://datasets/GotThatData/STARGATE/metadata.csv")

# Filter for Star Gate program
star_gate_docs = df[df['program'] == 'Star Gate']
print(f"Found {len(star_gate_docs)} Star Gate documents")

# Sort by date
star_gate_docs = star_gate_docs.sort_values('date')
print(star_gate_docs[['title', 'date', 'page_count']].head(10))

Example: OCR Text Extraction (with PyMuPDF)

import fitz  # PyMuPDF

def extract_text_from_pdf(pdf_path):
    doc = fitz.open(pdf_path)
    text = ""
    for page in doc:
        text += page.get_text()
    return text

# Extract text from a document
pdf_path = "hf://datasets/GotThatData/STARGATE/pdfs/STAR-GATE-00042.pdf"
text = extract_text_from_pdf(pdf_path)
print(text[:500])  # First 500 characters

Example: Train a Document Classifier

from transformers import AutoTokenizer, AutoModelForSequenceClassification, Trainer

# Load pre-trained model
model_name = "distilbert-base-uncased"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name, num_labels=4)

# Prepare data (classification: transcript, report, memo, other)
# ... (data preparation code)

# Fine-tune
trainer = Trainer(model=model, train_dataset=train_data, eval_dataset=eval_data)
trainer.train()

---

📚 Background & Context

What Was STARGATE?

In 1972, the CIA and DIA began funding research into remote viewing - the alleged ability to perceive distant or unseen targets through extrasensory means. Programs evolved over two decades under various codenames:

1. Early Phase (1973-1978): Grill Flame - Initial DIA experiments at SRI International
2. Operational Phase (1978-1991): Center Lane & Sun Streak - Active intelligence gathering
3. Final Phase (1991-1995): Star Gate - Consolidated under DIA oversight

Why Was It Terminated?

In 1995, the CIA commissioned an independent review by the American Institutes for Research (AIR). The final report concluded:

"The information provided by remote viewing is vague and ambiguous, making it difficult, if not impossible, for the technique to yield information of sufficient quality and accuracy for actionable intelligence."

Congress defunded the program shortly after. All documents were declassified and released via the CIA's CREST database between 2000-2017.

Notable Participants

• Ingo Swann - Developed Controlled Remote Viewing (CRV) methodology
• Pat Price - Remote viewer credited with "successful" Cold War missions
• Dr. Hal Puthoff - SRI physicist who led early experiments
• Dr. Edwin May - Director of cognitive sciences division at SRI

---

🔬 Research Applications

Academic Papers Using This Dataset

(We'd love to list your work here! Open a discussion or PR)

Potential Research Questions

1. Linguistics: How did remote viewing terminology evolve across programs?
2. Psychology: What cognitive biases are evident in session transcripts?
3. History: How did Cold War tensions influence psychic espionage funding?
4. Statistics: Can we replicate the AIR study's statistical analysis with modern methods?
5. AI Ethics: What does government-funded psychic research teach us about evaluating anomalous claims in AI?

---

📝 Citation

If you use this dataset in your research, please cite:

@dataset{daugherty2025stargate,
  author = {Bryan Daugherty},
  title = {STARGATE: CIA Remote Viewing Archive},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/GotThatData/STARGATE}
}

---

🔗 Related Resources

• CIA CREST Database: https://www.cia.gov/readingroom/
• AIR Final Report (1995): Research Evaluation
• Companion Dataset: STARGATE-Processed - OCR-extracted text versions

---

📜 License

MIT License - freely available for research and commercial use.

Note: These are declassified U.S. government documents and are in the public domain. The dataset compilation and metadata are released under MIT.

---

🙏 Acknowledgments

• CIA CREST Team - For declassifying and digitizing these historical documents
• Internet Archive - For mirroring and preserving government data
• Hugging Face Community - For providing infrastructure for large-scale dataset hosting

---

🐛 Known Issues & Future Work

• OCR Quality: Scanned 1970s-1990s documents have variable quality. Some pages are illegible.
• Missing Documents: Not all STARGATE-related documents may be included (ongoing FOIA releases).
• Metadata Gaps: Some documents lack complete metadata (date, program name, etc.).

Roadmap:

• Complete OCR extraction for all PDFs → STARGATE-Processed dataset
• Add entity annotations (people, locations, programs)
• Create topic models for common themes
• Build search/query interface (Space app)

---

💬 Feedback & Contributions

Found an error? Have a feature request? Want to contribute OCR'd text or annotations?

• Discussions: Open a discussion
• Contact: YourFriends@smartledger.solutions

---

"The truth is rarely pure and never simple." — Oscar Wilde