Supporting the Livepeer Network’s Top Performers With Confluence
A couple months ago, it became clear that rising ETH transaction fees were making it difficult for high-performing node operators on the Livepeer network — those who reliably transcode video in real-time or faster with high availability — to cost-effectively make the daily reward calls on the network required to earn LPT each day. This is an issue that affects the Livepeer network as a whole, because these are the exact types of participants who should be growing their ownership, and using the inflationary LPT to help offset the cost of node operations. In recognition of the pressing nature of this problem, the core team resolved to find a way to “activate” more high performing nodes to be able to profitably call reward. We publicly shared this as a community goal as well, since we knew the wider Livepeer ecosystem could help us address this challenge.
The suggested mechanisms that could be used to activate more nodes ranged from proactively channeling staked LPT toward high performers, to updating the Explorer or other products to encourage delegators to back them, to scaling work to reduce the gas costs associated with protocol actions. The community stepped up bigtime to make progress towards this goal, helping one another market high performing nodes, donating stake towards good candidates, and more. While this progress was great, we ultimately fell short of our goal, despite the enormous efforts of both the team and community. There are currently 44 high performing nodes, 17 of which can profitably call rewards with a 25% reward cut. Our goal was to increase the number of profitable high performers to 66.
So where do we go from here? It seems evident that the scale of the problem posed by high gas fees requires a protocol-level solution. The good news is that the migration to layer 2 will allow us to address this root problem.
The core team collected data to quantify the number of high performers and the number of nodes that could actively call reward, and built a dashboard that could track this ratio over time. This went well. Unfortunately the data was pretty telling. To profitably call reward, nodes would have to have a high reward cut of 25% and attract more than 50,000 LPT worth of stake. With the high cost of moving stake around, a few nodes could be proactively activated by grants and large holders, but it was prohibitive for the average token-holding community member to make an impact here. The scale of product work necessary to influence stake movements also felt a bit futile in the face of such high gas costs.
The feedback from the wider community was equally clear: high gas fees are the biggest barrier to their success.
All of this has led to the recognition that the big impact would come through scaling, particularly the Confluence network upgrade, which will migrate staking to a layer-2 solution, and make node operations and staking actions far more cost-effective. This big unlock will also make community efforts and product-related efforts, which up until now felt like running uphill, worthwhile.
As such, the core team’s focus shifted to making rapid progress on this scaling update. They have created a pre-proposal, informed by initial prototyping, and are actively working on completing the implementation, running benchmarks to estimate gas cost savings and finalizing the LIP (Livepeer Improvement Proposal) that will move through the remainder of the governance process. The next goal after the completion of these milestones is to launch a public testnet that the community can participate in, both to help with testing and to get familiar with the Confluence changes. The latest discussion on roadmap and milestones can be found in this forum thread.
In the meantime, we encourage community members to continue helping to activate high performers where possible. Participate in programs like the Titan Node Start Up Grant, stake towards those scoring well on the performance leaderboard, and actively participate in the testing and governance work required to get Livepeer scaled on layer 2.
The ideal end-state following the upgrade is that stake can move quickly and cost-effectively with limited gas costs, enabling token holders to chase yield and return by supporting and directing work towards a global network of high-performing node operators. It should be in every token holder’s long-term interests for high performers to be activated, transcoding video at scale, and earning LPT. While there is much work that can be done through community and product, the upcoming Confluence upgrade will be the key that unlocks this gateway to success.