This post is a call for feedback from the community on a proposal to increase the # of active transcoders in the Livepeer protocol from 10 to 15 on Friday, July 27th during round 1048, taking effect in round 1049 on Monday, July 30th during round 1051, taking effect in round 1052.
Background
During Livepeer’s alpha release (Snowmelt phase), the number of nodes who could be active on the network in a given round was intentionally kept at 10, a small number, so that the network could be tested. This small number also allowed the early network to achieve a full set of transcoders who were actively paying attention and participating in the protocol, as well as responding to frequent upgrade requests and network issues.
The goal is of course to decentralize far beyond just 10 nodes doing encoding work in the network. And although there are some Ethereum scalability limits at the moment to increasing this number dramatically, we’d like to take the first step of increasing it by 50% to 15 active transcoders.
Provide opportunity for more participants to learn the role of transcoder and compete to use inflation funding to help develop Livepeer initiatives.
There are many newly active users and groups in the community who have prepared to run transcoder nodes effectively who just haven’t had an active slot available yet.
The upcoming Merkle Mine claim period will unlock a bunch of LPT to new users, who will now have more diverse options to stake.
Potential drawbacks of the change
It will be 50% more expensive in terms of Ethereum gas to initialize each round.
The transcoder dashboard doesn’t yet have a great sense of identity, linking transcoders to their campaigns or identities. This makes it difficult for users to decide whom to delegate their LPT towards. Choosing between 15 requires more research than choosing between 10. We’d hope to address this via product work on the dashboard and welcome contributions.
Moving beyond 15
While we can push slightly beyond 15 in future updates after the next stability milestone release for Livepeer, there are two additional limiting factors:
Transcoder identies to help users delegate (as referenced above). We have started an ENS integration so transcoders can actually have display names, but need more product work.
Ethereum scalability concerns - the protocol needs an architecture update, otherwise doing all the accounting around who’s active and who’s not becomes too expensive in terms of Ethereum gas. We’re working on these updates for future versions of the protocol.
Open discussion
Have any concerns of questions about this parameter upgrade? Share them here. As part of a shift from core team parameter setting to decentralized governance we welcome all input and discussion before making any updates to the protocol.
I agree with the sentiment behind the change, but I believe with the current Ethereum network congestion and skyrocketing gas costs it could be wise to hold off on making this change at this point in time. The average transaction cost is up 700% over the last quarter and I think anything that can be done to keep these costs lower would be a good idea.
This would be more to be conservative in case gas prices go even higher over the next few months. Just my thoughts, however I am not one of the transcoders so I would be keen to hear their thoughts
Thanks for bringing this topic up. Just to clarify this change would not increase costs for a transcoder or delegator on a day to day basis. It only increases the cost of the once daily initializeRound txn which is a little bit of a tax on the network. Someone has to pay it, and right now we run a daemon that just does it. But we need to work out the incentives for why a transcoder would call it in the future or how it could be split up.
I support this proposal, but I think 2 days in between the creation of the proposal and the time of the proposed parameter change might be a little too short even at this early stage of the network. While increasing the time for receiving feedback might not be as pressing for this specific proposal, it could be beneficial to generally make sure that there is always ample time for parameter change related proposal reviews unless there are urgent circumstances. How about shifting to Monday July 30th (round 1051, taking effect in round 1052) to give people the weekend?
Thanks for the suggestion to add more time for community review. Seems like a good idea to make sure people have a chance to review and to set a good precedent that future non-emergency param updates proposals will be given at least a few days for review.
I’d like to update the proposal to have the increase take effect on Monday instead of Friday.
As a currently inactive transcoder, I’d like to see more on the network, however recognize I’m biased. Setting that bias aside, a 50% increase in transcoders, from 10-15 does feel like a beneficial move in the direction of de-, rather than re-cenntralization.
For transcoders and how much LPT supply are delegated to each would be beneficial. At the same time, I don’t know that anyone has a good idea of “how much” decentralization is “good”. We’ll have to figure that out as the project evolves.
Another useful metric would be from the perspective of network performance. What QoS/SLA do we aspire to provide as individual transcoders and as a collective group, for the Livepeer network, e.g. as good as the most popular commercial streaming services, better?
Thanks to everyone for the comments during the open feedback period. Between numerous shows of support for the proposal between here and the Discord channel, as well as being able to clarify and address any concerns, we’ll be moving ahead with the update this round.
I look forward to more debates on future parameter updates, as well as beginning to move towards more decentralized forms of invoking updates based on collective decision making.
As an update, the number of active transcoders has been increased from 10 to 15. As soon as round 1053 begins at Ethereum block 6065280, there should be 15 active transcoders in the pool.