Nockchain
Stay Updated
Nockchain is the
zk-Proof of Work
blockchain

Pining for Proofs

While Big Tech goes for the jugular, Proof-of-Stake declines.

Photo by Jasper Morrison

We're now officially in the trenches, with the whole team working wartime schedules until the Nockchain dumbnet is live.

This week, a few observations on tech news, crypto markets, and the Zorpian lifestyle (while the engineers are busy doing real work).

In this issue:

  • Amazon Goes Nuclear
  • Uniswap Leaves Ethereum L1
  • GSM-Symbolic: Do LLMs Reason?
  • How to Smell Like Pine (Pine-Maxxing)

Amazon is Going Nuclear

As we theorized last month, Big Tech is gunning for the World Stack.

And software is losing its appeal relative to physical infrastructure.

Amazon has announced a $500 million investment in X-Energy, a leader in advanced nuclear reactor technology, to develop small modular reactors (SMRs). This investment aims to meet growing energy demands, presumably as part of the AI-driven mania for physical compute. Amazon plans to purchase electricity from the first project, with potential for over 5 gigawatts of new power projects nationally.

One day you start selling books and then, before you know it, you're building private nuclear energy capacity to train AI models.

Uniswap Leaves Ethereum L1

We've said before that we do not see a bright future for Proof-of-Stake systems.

Justin Bons argues that Ethereum is facing significant challenges due to the migration of apps to competing L1s and L2s, exemplified most strikingly by Uniswap moving to its own Ethereum L2 (Unichain). Uniswap was one of the first and most significant protocols built on Ethereum. Bons contends that Ethereum's "L2 scaling" strategy has led to decreased fee revenue and compromised its original vision, making it increasingly irrelevant in the face of technologically superior competitors. ETH 2.0's execution sharding for L1 scaling has been abandoned.

It seems that the "global computer" refuses to run on a pre-mined asset of nominal value (not backed by physics).

GSM-Symbolic: Do LLMs Reason?

There is an emerging debate around how we describe the work of LLMs.

Research from Apple suggests that Large Language Models (LLMs) do not perform genuine logical reasoning in mathematical tasks. The study introduces GSM-Symbolic, an improved benchmark that reveals significant performance drops in LLMs when numerical values are altered or irrelevant clauses are added to questions.

As technology continues to accelerate, expect philosophical questions to bear increasingly significant political loads. Do LLMs really think, or are they just simulating thought? Your posture to such questions will increasingly shape your life outcomes. Yet different vested interests will be incentivized to have you answer one way or another. Choose wisely!

How to Smell Like Pine (Pine-Maxxing)

Last month the Zorp group chat tackled a pressing question: What is the best way to smell more like pine needles? Brian recommends Yatagan by Caron (~$50), Polo Green by Ralph Lauren (classic 80s scent), and Fille en Aiguilles by Serge Lutens (on the high-end) as the best options. Other things we discovered... Oud refers to fermented agarwood, often synthetic in Western perfumes; eau de parfum is defined by a higher concentration of fragrance oils than eau de toilette, and pine tar soap can provide a natural campfire scent.

SUBSCRIBE