Bittensor recently introduced the Conviction mechanism, which requires participants to lock their TAO tokens for a specified period, such as several months or years. The locked tokens, combined with the duration of the lock, generate a 'Conviction' score, serving as a quantifiable indicator of participants' long-term commitment to the network.
Unlike the previously static ownership of subnets, the updated model introduces dynamic voting rights. This cryptographic design ensures that the locked stake and its duration can be verified within the protocol, shifting towards transparent on-chain accountability and reducing reliance on subjective trust.
Founder Const addressed concerns raised during a Discord meeting regarding the sudden departure of a key subnet operator, acknowledging the negative impact on TAO holders and apologizing to affected community members.
Bittensor, as a decentralized network, aims to incentivize peer-to-peer machine learning. Through token economics, it coordinates validators, miners, and subnet operators, laying the groundwork for open collaborative AI development that transcends centralized control.
The initial implementation of the locked staking system will primarily occur in some of Bittensor's more mature subnets, including Subnet 3, Subnet 39, and Subnet 81. Other significant subnets may also join this phase, while newer subnets will enjoy a grace period before the rules are enforced.
Once the mechanism is implemented, voting and the transfer of subnet ownership will be based on Conviction scores. Participants with the highest Conviction scores will receive the owner keys for the subnet, making control a competitive and fluid process rather than a fixed state. This measure aims to curb short-term participation and sudden exits, preventing capital outflows or investor losses. Publicly locked stakes will serve as a visible measure of each subnet team's confidence and seriousness.
The broader goal behind the Conviction mechanism is to address the governance challenges commonly faced by Web3 projects, particularly those managing valuable infrastructure and community resources. Const described this update as establishing a federation of independent and trustworthy decentralized AI companies, with governance rooted in time-based network participation. It aims to promote greater alignment and capital formation among subnet owners, miners, validators, and stakers.
Planned enhancements include extending this governance style to layer and chain levels, provided the current implementation proves effective. This evolution is seen as a structural response to past governance issues, addressing the vulnerabilities of the protocol.