March 2025 Monthly Summary

Roko Network Monthly Update

Month: March 2025


Executive Summary

March 2025 was a breakthrough month with significant progress on the NPoS pallet, a hardware security vendor hardware security integration, and the time validation mechanism for the Bittensor subnet. The team also received critical intelligence from Bittensor core developers about upcoming features that will impact subnet launch timing.


Major Technical Achievements

NPoS Pallet Integration

Stalker (Unforkable) achieved significant progress on the Nominated Proof of Stake pallet:

  • Rewrote staking process to use EVM memory instead of Substrate memory
  • Staking information now accessible to dApps and tooling
  • Most unit tests passing with high confidence in implementation
  • Critical Issue: A dependency library name change in Rust ecosystem broke builds
  • Forced unplanned upgrade to newer Substrate dependencies
  • Key insight: AI tools ineffective at complex Rust build issues but excel at JavaScript

Time Validation Mechanism

Brendon completed the proof of concept for time validation:

  • Addresses weight copying, delay mining, and miner overfitting issues
  • Two-method approach:
  • External API on centralized infrastructure (30% weight)
  • Mean average of miner timing values (70% weight)
  • Decision: Focus on timing services only initially, defer AI/ML integration
  • Separates concerns between timing subnet and potential GPU orchestration wrapper

a hardware security vendor Security Integration

Two support sessions with Bob from a hardware security vendor:

  • Bootware installation on SCM (Secure Compute Module) devices
  • LUKS key management using a hardware security vendor hardware
  • A/B partition strategy for safe updates
  • Image creation with private/public key pairs
  • Confirmed: Encrypted bootware runs without leaving dev mode

Infrastructure Progress

Distributed Computing Architecture

BigM presented secure CDN network vision:

  • Minis Forum mini PCs ($500-1,000 each)
  • a hardware security vendor security modules + timing cards
  • Bank-level security for distributed network
  • Capable of running 70B parameter models per node
  • Total developer network cost under $10,000

AI-Assisted Development Adoption

Team exploring AI coding assistants:

  • Mantis (Claude wrapper) for development acceleration
  • OpenManus on A100 GPU infrastructure
  • Prediction: 6-12 months until most code is AI-written

Bittensor Ecosystem Intelligence

Critical updates from Slice's conversations with core developers:

  • Start Button Feature: Subnet owners can fully prepare code before launch
  • Exploit Alert: Subnets above #64 experiencing excess emissions extraction
  • Funding Option: Const Capital offering loans for subnet registration in exchange for alpha tokens
  • Strategic Decision: Validator operation deprioritized; subnet ownership is better focus

Partnership Developments

Greenhouse Automation Integration

  • Conversations with former colleague on greenhouse hardware
  • Laura-based monitoring with edge nodes
  • Potential integration with Roko timing and C2 infrastructure
  • NDA being signed for deeper exploration

Task Bounty Platform

  • Integration exploration with contract board project
  • Audited smart contracts, seeking $5M funding
  • Roko positioned to provide grants for network application developers

Post-Quantum Cryptography Research

  • Discussion of Cosq chip vs a hardware security vendor root of trust
  • Key generation combining multiple entropy sources:
  • Hardware RNG chips
  • Oscillators
  • Network statistics
  • Cosmic rays
  • Larger keys/datasets approach rather than new algorithms

Key Metrics

MetricStatus
NPoS Pallet95%+ unit tests passing
Time ValidationPOC complete
a hardware security vendor IntegrationTwo training sessions completed
Mini PC NetworkArchitecture defined
Bittensor IntelStart button feature incoming

Challenges & Resolutions

1. Rust Dependency Breaking Change

  • Library renamed in ecosystem
  • Resolution: Upgrade to newer Substrate dependencies
  • Lesson: AI tools struggle with complex Rust issues

2. Subnet Launch Timing

  • Current day-one rush problematic
  • Resolution: Wait for start button feature

3. Resource Allocation

  • A100 GPU utilization question
  • Options: Mining (challenging vs H100s), inference via Spheron/IONet, internal tools

Looking Ahead to April

  • Resolve Rust dependency issues in NPoS integration
  • Complete time validation code implementation
  • Monitor Bittensor start button feature release
  • Set up additional a hardware security vendor node via Cloudy Storm
  • Continue greenhouse partnership discussions

Team Focus Areas

Team MemberPrimary Focus
Stalker (Unforkable)NPoS pallet, dependency resolution
BrendonTime validation implementation
Manitcora hardware security vendor configuration, greenhouse NDA
BigMSecure CDN architecture
SliceBittensor ecosystem monitoring
TeamAI tool adoption for development