976

This proposal reduces the size of the Osmosis active validator set from 120 to 100. This reduction is expected to enhance block processing performance, reduce consensus overhead, and simplify network operations while maintaining effective decentralization.

Motivation

Following [Hidden link], which reduced the validator set from 150 to 120, the Osmosis chain has experienced measurable performance improvements. These enhancements have contributed to faster block finality, more efficient network communication, and reduced hardware demands on validators, benefits that improve both user experience and the long-term scalability of the chain.

• Block times decreased from ~1.33s to ~1.22s.
• Consensus gossip traffic dropped significantly, reducing peering overhead.
• CPU load on validators decreased, particularly from the staking module logic, which consumes a significant percentage of block execution time, even in low-activity blocks.
• Validator reliability improved, with fewer missed blocks observed since the reduction.

Decentralization Impact

CometBFT consensus requires 67% voting power to approve each block or software upgrade and 33% to veto.

A common concern with reducing the validator set is that it could lower the potential decentralization of the chain by requiring a smaller number of validators to reach the 67% threshold needed for block approval, and by potentially granting veto rights to fewer validators who comprise the top 33.3%.

Despite reducing the number of active validators from 150 to 120, decentralization has not suffered. In fact, it has since improved with the Nakamoto Coefficient increasing from 10 to 12, meaning it now takes a broader range of validators to reach 33.3% voting power and veto a block than when the validator set was at 150.

Chart: Nakamoto Index for Osmosis consensus over the last year. Source: [Hidden link]

Decentralization is a trade-off in terms of performance, with smaller validator sets producing blocks faster, more reliably, and at a lower CPU cost.

Validator distribution remains broad with a long tail:

- **Top 33.3% of stake**: 12 validators
- **Top 67%**: 40 validators 
- **Bottom 33.3%**: 80 validators 

These figures show that trimming the bottom portion of the set can reduce technical overhead, which is dictated by the number of validators present, without heavily impacting the number of validators required for each consensus level, which is dictated by the voting power distribution instead.

Proposed Change

This proposal would adjust the staking parameter directly to reduce the Validator count to 100.

This reduction excludes validators representing the bottom 2% of active voting power, the same amount as the cut from 150 to 120. This threshold has a minimal impact on consensus while delivering tangible performance improvements similar to the previous reduction.

Chart: Current Validator set with consensus boundaries and proposed tail reduction marked

Projected validator set distribution after this proposal:

• Top 33.3% of stake: 11 validators (-1)
• Top 67%: 34 validators (-6)
• Bottom 33.3%: 66 validators (-14)

Forum post: [Hidden link]