The Approach

How humanity approaches
the divine

114,858 verses · 39 traditions · 4,500 years

A corpus of humanity's sacred writing, read by machine under two lenses and calibrated against texts that are not sacred at all. The findings that survived are here; so are the beautiful theories that did not. What remains, across every century and continent, is less a shared belief than a shared road: separation, descent, surrender at the bottom, return.

Scroll to explore
drag to rotate · hover to read
Three ways to explore
01

By Tradition

Choose any of 39 traditions. Browse its verses. Find what resonates across every other tradition in the corpus.

Open a tradition →
02

By Axis Profile

Set the 13 transformation sliders to describe a spiritual quality (surrender, clarity, stillness) and find every verse that embodies it.

Search by axis →
03

Confirmed Resonances

Browse cross-tradition verse pairs that score highly under both a neural embedding and a 13-axis transformation framework.

View resonances →
Four studies

The Journey

Do itinerary texts trace a descent into darkness and an emergence? Sensory trajectories through narrative order.

The Temples

Ritual architecture as staged sensory restriction, from Göbekli Tepe to Chartres, compared against the texts.

The Origins

Creation narratives isolated as a genre: recurring motifs, and whether similarity decays with distance.

The Awareness

Where machine self-description lands among 114,858 verses of human contemplative writing.

What the data says

Five findings

Finding 01

Rumi and Guru Nanak: 0.991

The highest tradition similarity in the corpus. Sufi Persian poetry and early Sikh devotional writing: the pair history would predict it, and the ranking found it on its own.

Finding 02

Gita ↔ Tao: same meaning, different mechanics

Semantically nearly identical (0.977). But their transformation profiles diverge sharply (0.623). Same neighborhood in meaning space, different roads.

Finding 03

Philokalia ↔ Zhuangzi

Greek Christian asceticism and Taoist philosophy had no direct historical contact, yet they are near neighbors under both lenses. Both lenses read the same translations, so this is convergence, not independent proof.

Finding 04

Dark Night ↔ Dhammapada

The largest embedding/axis gap in the corpus: 0.915 semantic similarity, negative axis similarity. Same contemplative vocabulary, opposite position on the path. One text speaks from the destination, the other from the painful middle.

Finding 05

The gradient, with caveats

In 732 of 741 tradition pairs, embedding similarity exceeds axis similarity. Shared vocabulary, different transformational work, or partly an artifact of how embeddings compress text. The data supports the question, not yet the conclusion.

Read the full analysis →

The Approach · universalaxes.org · A computational study of sacred texts · Payton May & Claude