Network Economics Update - 5/13 - Transcoder Performance


#1

The Livepeer protocol has been humming along for another week on mainnet.

  • It’s up to a 0.33% participation rate with 33,181 LPT staked up from 16,173 LPT staked last week.
  • It has a full active transcoder pool with 10 active and 1 reserve transcoders, up from 9 active last week.

Let’s take a deeper look at transcoder performance.

This chart looks at whether or not each active transcoder has successfully invoked the reward() method each of the past 4 rounds. As a reminder, calling reward() is critical, and it’s currently the transcoder’s #1 priority, as this is how new token gets generated and reallocated amongst the transcoder and all its delegators. Read more about reward here.

  • The 4 transcoders with the highest stake have reliably called reward.
  • Two transcoders, 0xbe9a, and 0xbD91, have not reliably invoked reward() at all, and are missing out on an opportunity for both themselves and their delegators.
  • A couple transcoders have been slightly less reliable. When you dig into the Etherscan transactions you can see that sometimes this is a result of transactions failing due to not submitting with enough gas, or not setting a high enough gas price and txns sitting pending. These are easily correctable. We live and we learn :slight_smile:
  • Transcoder 0x88E1 is not in the active set with enough stake, but perhaps they should be delegated towards and given a chance, considering that there are a couple transcoders who are not reliably invoking reward()?

One thing that’s being worked on is making this information more visible in the explorer transcoder dashboard. See this issue on LivepeerJS for progress. Beyond that, it will also be super important to make visible how many segments of video transcoders are processing, and how many faults they’ve been slashed for in the protocol. Those stats will pick up in importance as network usage increases through the alpha.