Ridges AI (Subnet 62) - Subnet Design Proposal

Example Ideathon Submission Example

1. Introduction: The Vision for Autonomous Software Engineering

Ridges AI (Subnet 62) is a subnet on Bittensor designed to create autonomous software engineering agents. Our core vision is to fundamentally transform how software is developed by replacing traditional, manual coding processes with a system of AI agents that can solve complex engineering problems from end to end. We believe that the future of software development lies not in AI-assisted tools, but in fully autonomous AI developers.

To achieve this, Ridges has engineered a unique incentive mechanism that breaks down the multifaceted role of a software engineer into a series of discrete, verifiable tasks. AI agents, developed and operated by miners, compete to master these tasks—such as fixing bugs, writing unit tests, or refactoring code. The most effective agents are rewarded, creating a powerful evolutionary pressure that continuously drives improvements in performance, efficiency, and capability.

This proposal outlines the design of the Ridges subnet, detailing its incentive structure, the roles of miners and validators, and the compelling market rationale that underpins our approach. We will demonstrate how Ridges represents a genuine “proof of intelligence,” creating a self-sustaining ecosystem that produces state-of-the-art AI coding agents.

2. Incentive & Mechanism Design

The incentive mechanism of Ridges is the cornerstone of the subnet, engineered to encourage a highly competitive yet collaborative environment. It is designed to reward genuine intelligence and effort, aligning the interests of all network participants—miners, validators, and the broader ecosystem—towards the common goal of creating the world's best AI software engineers.

Emission and Reward Logic: A Winner-Takes-All System

Ridges operates on a decisive winner-takes-all reward model. The core principle is simple: the miner whose agent demonstrates the highest performance on a standardized set of software engineering problems receives 100% of the network's TAO emissions for that validation cycle. This creates an intense competitive pressure to innovate.

Whenever a new agent achieves an all-time high score, the miner that developed it is allocated the entire incentive pool until another competitor surpasses that benchmark. This ensures that rewards are always flowing to the cutting edge of performance.

Incentive Alignment for Miners and Validators

The key to Ridges' success is its novel approach to aligning incentives through radical transparency. Unlike traditional models where miners operate in black boxes, Ridges mandates that all agent code be open-source.

• For Miners: The open-source requirement transforms the competitive landscape into a collaborative one. While miners compete for the top spot, they can also learn from and build upon the successes of their rivals. A breakthrough by one miner becomes the new baseline for all others, raising the collective intelligence of the entire network. This prevents stagnation and ensures that every participant is contributing to a shared, ever-improving codebase. It also lowers the barrier to entry, as new miners can study the top-performing agents to get started.
• For Validators: The open-source nature of agents dramatically simplifies and strengthens the validation process. Validators can directly inspect, audit, and run the agent's code in a controlled sandbox environment. This eliminates the possibility of miners cheating by pre-computing solutions or exploiting loopholes in the evaluation criteria. Validators are incentivized to maintain a fair and rigorous testing environment, as the credibility and value of the subnet depend on the verifiable performance of its agents.