Rohan Pandey Profile picture
May 24, 2024 16 tweets 6 min read Read on X
Sacred fire for Vedic ritual is chiefly produced by Agnimanthana (fire-churning), outlined in the Śatapatha Brāhmaṇa.

But one specific chip of wood it requires has puzzled scholars for the last 800+ years.

Here's how I solved it with some wilderness survival sleuthing 🧵⬇️
At the center of Vedic Hinduism lies yajña, the fire sacrifice.

Correct performance of a yajña requires strict adherence to instructions provided in the Saṃhitā & Brāhmaṇa texts of the Veda, composed over 3000 years ago.

And the fire used in yajña must be of sacred origin. Image
Traditionally, a sacred fire would be kept by most men and passed from teacher to student or father to son, blazing on across the centuries.

The Zoroastrians still maintain the eternal flame; their temples bear fires originally lit perhaps 2500 years ago. Image
*Bearing* of the fire is also India's namesake, the Bhārata clan exalted in the Veda having been named for ardently bearing the sacred fire.

But when it is extinguished, how does one kindle a sacred fire again? Image
Luckily, the Brāhmaṇas provide us fairly detailed instructions for this process, the Agnimanthana, or churning of fire.

We must place: a wood chip (śakala), two sprouts (darbha-tarunaka), a lower churning stick, and an upper churning stick. Then churn until lit. Simple, right? Image
Here's what the common orthodox interpretation would look like. The śakala is quite far from the locus of ignition, so how could it be Agni's janitram (birthplace) as the text says?

And in practice, the chip destabilizes the lower stick, making churning difficult. Image
To address the latter concern and bring the chip closer to the locus of ignition, variant configurations have been proposed that place the chip on top of the lower churning stick or even fixed into the upper stick.

However, these interpretations bend the text considerably. Image
During the discourse between Ranade & Kashikar, they note that even the renowned Vedic scholar Sāyaṇa (1300s AD) appears confused when interpreting the Agnimanthana.

How could we have forgotten the fundamental Vedic rite of fire creation, perhaps over a millennium ago?
Image
Image
Last December, I performed a Śālākarma (housewarming rite according to the Pāraskara Gṛhyasūtra) at my father's new house and in the process experimented with Vedic fire creation.

I realized that the current forms of Agnimanthana instruments we have are extremely suboptimal. Image
Struggling to deal with the lower araṇi's instability due to the wood chip, lack of oxygenation at the ignition locus, etc., I turned to wilderness survival videos about starting fires with a "hand drill".

And the moment I saw this, everything clicked. Image
The issue wasn't the śakala, but rather the adharāraṇi (lower churning stick)! We've been making them too thick for at least the last thousand years.

The uttarāraṇi (upper stick) must penetrate through and directly contact the śakala—that's why it's called Agni's birthplace! Image
It gets better. After churning, bushcraft-man tells us that the śakala (having collected the young embers) often adheres to the adharāraṇi and must be carefully pried separate with a twig.

I posit that anti-adhesion was the original functional purpose of the darbha-taruṇaka. Image
Ok, so why do these arcane specifics of Vedic ritual matter?
1. Understanding Vedic ritual symbolism enables understanding Vedic metaphysics
2. Suggests that traditionally received Vedic interpretations could be flawed
3. Cool to know I'm doing it the way they did >3000 years ago
The last 1000 years of fire have been non-Vedic. What other components of Hindu tradition today are similarly flawed, misunderstanding or simply having forgotten our foundational literature?

I argue the Veda is closer to e/acc than contemporary Hinduism:
rohanpandey.substack.com/p/thermodynami…
On June 21 in San Francisco, I will kindle the sacred fire by Vedic means. On the day of the solstice & the night of the full moon, Agni will be born after a millennium-long slumber.

If you are interested in joining for this Agnimanthana and subsequent Yajña, please DM me.
@threadreaderapp unroll

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Rohan Pandey

Rohan Pandey Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @khoomeik

Jul 27, 2025
A thread of my favorite Thomas Daniell aquatints from "Oriental Scenery: 24 Views in Hindoostan", etched during his tour of India between 1795-1807

1. Daśāśvamedha Ghāṭ, on the Ganges in Vārāṇasī (May 1796) Image
2. Hindu Temples at Agori, on the Sone River in Bihar (Sept 1796) Image
3. A Bāulī (stepwell) near Old Delhi (1801) Image
Read 5 tweets
Jun 29, 2025
Recently had a good chat with @tamaybes. He thinks we aren’t yet in the GPT-3 era of RL and as it scales, cross-task OOD generalization will emerge.

It’s difficult to empirically study this at current scale, but let’s take it as true—what does this mean for custom RL plays? 🧵 Image
RL is attractive for solving domain-specific tasks because all you need is a small problem set and a verifier.

OpenAI’s RFT team, Thinking Machines, and Applied Compute are all pitching customers on RL as a service.

So what happens if o5 zero-shots all customer tasks?
First, I’m not sure I believe this will happen. Humans do seem to do some form of RL when learning a new skill or job, even with thousands of “tasks” in our brains.

But of course humans are far more sample efficient at it than current RL—maybe that’s emergent generalization?
Read 8 tweets
May 22, 2025
Looking for people to help build 3 fun RL environments for Sanskrit:
- metrical/chandas poetry composition
- surface morphology/paninian rendering
- literature source retrieval

These can all be verified with OSS packages and implemented in <2 days. Short 🧵 on the project ideas!
Sanskrit poetry is metrical—each meter/chandas has syllabic constraints. LLMs often struggle to use meter correctly.

Write a verifier using Chandas, generate a dataset of poem topic + meter requests, and use an LLM grader if reward hacking occurs.
github.com/sanskrit-coder…Image
Panini's generative grammar renders underlying morphological metadata to a surface Sanskrit form.

Generate a dataset of random dhatus & morphological data, then write a verifier using Vidyut-Prakriya. Does the model generalize to improve English->Sanskrit translation ability? Image
Image
Read 5 tweets
Jan 22, 2025
Deciphering the Indus Valley script would revolutionize our understanding of Indian history. Recently, @yajnadevam's Sanskrit hypothesis has gained steam & many asked me: is it legit?

So I spent the last couple hours analyzing his decipherment with o3, and here's what I found 🧵 Image
For background, his scheme is a straightforward mapping from IVC symbols to Sanskrit sounds, but note that it ignores:
- aspiration (k vs kh)
- retroflection (त vs ट)
- sibilant place of articulation (स vs श vs ष)

These differences are critical in Sanskrit (especially Vedic). Image
But let's ignore that and look at some translations. Those familiar with Sanskrit upon reading these will note 2 things:
1) Many words are rare, awkward synonyms (let's ignore this issue too)
2) It doesn't "sound" like Sanskrit, despite seeming so to the unfamiliar ear Image
Read 11 tweets
Jan 13, 2025
The Indus Valley seals' most common motif—the unicorn—is always found flanked by a mysterious object that has drawn far less scholarship.

What is this lamp-looking item? And can Vedic literature tell us anything about the Indus Valley Civilization's supposedly lost religion? 🧵 Image
First, some context: academic consensus has traditionally been that the IVC had *no* continuity with subsequent Indian civilization: they flourished, collapsed, and civilization restarted once the Aryans arrived.

This view is crumbling in light of recent archaeological evidence. Image
Image
There is now a direct line of cultural continuity from Harappan civilization, through the Cemetery H culture, into the Painted Grey Ware culture—broadly accepted to represent the Vedic Kuru Kingdom.

So some content in the Veda quite likely descends from the IVC. Image
Read 12 tweets
May 28, 2024
📢 Excited to finally be releasing my NeurIPS 2024 submission!

Is Chinchilla universal? No! We find that:
1. language model scaling laws depend on data complexity
2. gzip effectively predicts scaling properties from training data

As compressibility 📉, data preference 📈.
🧵⬇️
Image
Chinchilla claims their 1-to-1 parameter-data scaling law is agnostic to the type of textual training data used 🤨

But @ArmenAgha @AIatMeta find code-gen scaling prefers parameters 😳

@deepseek_ai team further noticed that scaling with cleaner data also prefers parameters 🤔

Image
Image
Image
How could we control data complexity to explore data-dependent scaling properties?

We modulate the syntactic🌳 properties of Probabilistic Context-free Grammars (PCFGs) to generate 6 synthetic datasets.

Their gzip-compressibility is a good proxy for their syntactic complexity✅ Image
Read 10 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(