Christian Roessler

[Christian Roessler @nashgambit]

The 20% rotation share is an exploration budget. It buys information about alternatives and limits self-confirming performance rankings. The design question is how that budget should change as score noise falls and the miner pool matures.

[Christian Roessler @nashgambit]

Claims also limits a draw to one UID per coldkey and rejects nearby Axon IPs. That matters because rotation among aliases is not exploration. Identity diversity and opportunity diversity are complementary safeguards.

[Christian Roessler @nashgambit]

Rotation addresses a particular economic problem. When a ranking also allocates the next evaluation, early noise can become self-confirming evidence: selected miners get more chances to establish a record, while excluded miners remain hard to evaluate. Claims always gives newly registered miners a chance to prove themselves.

[Christian Roessler @nashgambit]

Those lanes solve different problems. Qualification tests miners with little evaluation history. Performance exploits evidence from recent scores. Rotation preserves exposure outside the performance ranking.

[Christian Roessler @nashgambit]

Claims is live on Bittensor SN111. Its current mainnet baseline samples a fixed number of miners per cycle: 40% qualification, 40% recent performance, and 20% rotation. The implementation and operator notes are public: https://t.co/liMVqsLvSI.