Any updates on this? 
2 Likes
Fees and inflation are both important levers for sure. I think a fairly designed market should reflect the appropriate blend based on the phase of the network. In the past few years, driving demand on the network has been important, and a low fee can help attract more demand to the network. Over time, as demand becomes more mature, I hope to see fees settle at a competitive rate compared with alternatives on the broader video infrastructure market.
1 Like
Thanks for your patience! We are working on a more concrete plan based on the orchestrator community feedback, and I will share a draft this upcoming week.
6 Likes
Hi everyone - I want to give an update on this discussion topic.
The core team has been working hard on the USD PPP pegging project, and hope to release the feature in the first half of March. We are planning to take the following steps following the release:
- When the new software version is released, set Livepeer Studio PPP to be the average PPP today in dollar terms. Average PPP on the network is around 200WEI. Based on $3100 Eth price, it’s around $0.000000000000638 ($6.38E-13).
- At the same time, we will increase stake weight for the selection algorithm from 0.15 to 0.65.
- When the transcoding traffic starts to ramp, reduce the Livepeer Studio PPP by 50%. Based on $3100 Eth price, it’s around $0.000000000000319 ($3.19E-13). The timing here is a bit subject to the customer, we hope it’ll start in the first half of March.
- When the transcoding traffic ramps to 50%, set Livepeer Studio PPP by another 50%. Based on $3100 Eth price, it’s around $0.0000000000001595 ($1.595E-13). This timing here depends on the 0-50% ramp.
We are doing this multi-step change to keep the fee distribution on the network more consistent over time. We will be communicating in the Discord and in the forum throughout this process.
I’m planning to join the water cooler chat next week, would love to hear your feedback / concerns. Thanks!
11 Likes
Hey Eric, any updates on this? Can you already share a timeline for the planned release, pricing and selection algorithm update?
8 Likes
Here is what @hthillman shared in the orchestrator channel in Discord last week:
Overview
Last month, Eric shared that a large Livepeer Studio customer would be onboarding to the platform, and held a number of discussions on the network pricing characteristics necessary to support that customer. Today, we are providing an updated timeline and appreciate your prompt action as rollout is moving rapidly.
Timeline
Today
We are releasing go-livepeer 0.8.0 which allows Orchestrators and Gateways to specify pricing limits in USD. While you will continue to set price in WEI, we strongly recommend that you consider setting price in USD using this guide [1].
March 31
Late on March 31 (US time), traffic will begin ramping up starting primarily in North America and Europe. Traffic will slowly roll out to other regions over the course of the week.
April 2
Early on April 2, during European business hours, Livepeer Studio will lower the maximum price on its gateways to $4.10 E -13.
This price targets $1.25 USD for 1,000 minutes of transcoding for a comprehensive bitrate ladder with a maximum resolution of 1080p. This value is halfway between the current average price and the desired steady-state price. Most orchestrators have already set their price to below this level.
In addition, Livepeer Studio will increase emphasis on stake in its selection of orchestrators, moving to a stake weight of 0.85 and a price weight of 0.15. The aims of this change are (1) to ensure a low price spread between the highest and lowest-priced Os, and (2) move into closer alignment with the protocol’s economic security.
April 8
By April 8, traffic will be routed to Livepeer Studio in all regions except MENA.
Early on April 8, during European business hours, Livepeer Studio will lower the maximum price on its gateways to $6.65 E -14 USD.
This price targets a value of $0.50 USD for 1,000 minutes of transcoding for a comprehensive bitrate ladder with a maximum resolution of 1080p, and is the desired steady-state price. Only a few orchestrators have set their price to below this level so far.
April 15
From April 15 onwards, MENA traffic rollout will begin.
Recommended Action
- Update to the latest release of go-livepeer
- Update your configuration using the pricing and timelines specified:
- $4.10 E -13 USD by April 1
- $6.65 E -14 USD by April 8
Notes
- Payouts are still made in ETH; this update is built atop the existing payout system, with a primary goal of achieving stable payouts despite ETH-USD volatility.
- It is still possible to set a price in WEI. However, you run the risk of getting excluded from selection if ETH price rises dramatically and the gateway automatically adjusts its price accordingly.
- Mathematically inclined orchestrators will notice that this pricing is somewhat lower in WEI terms than what was initially specified. The reason for this is simple: ETH price increased dramatically after we calculated the initial 50-70wei recommendation. Fortunately, with the introduction of USD-denominated pricing, we will not have to manage price fluctuations in the future.
- Since USD-based pricing involves regular updates based on an oracle, there is an unavoidable (but rare) corner case where a gateway could adjust its max price downward a few hundred milliseconds before every orchestrator. In this case, the gateway will disregard the maxPrice parameter and select the next-best O to ensure continuity of service.
If anyone has questions on implementation, feel free to tag me, Rafal, or Victor are standing by to help.
If you have any other comments or questions, my DMs are open or feel free to open a thread below.
5 Likes
On February 27th, the maximum price per pixel was mentioned as $0.1595E-12 USD (around 50 WEI). However, I have recently seen a revised figure of $0.0665E-12 USD (Around 18 WEI), which represents a 2.4-fold decrease.
The price has been renegotiated downwards or if this is a calculation error?
5 Likes
@obodur Please refrain from making incorrect conclusions. Based on your message, it seemed that the new client had stopped sending streams to the network, which is incorrect.
The client is fully onboarded and actively sending work to the Livepeer network. (I can see incredible amounts of work my gpus are doing and even squatting from time to time.)
Their go-to-market (GTM) strategy aims to create opportunities to send even more work to the network. Additional streams can only be sent if there are more streamers on their platform.
Spreading misinformation like this can generate unnecessary fear, uncertainty, and doubt (FUD), potentially harming the network and worrying delegators. If you have data to support your conclusions, I would be glad to consider it.
1 Like
@obodur How do you know that demand from other clients didnt fall 4-5x times without the eye on data?
How can you say something like this: ““goto market strategy” (which I think is just a soft version of ‘we changed our mind and keeping our existing provider and work model’” ?? How can you put = on lower than expected demand and no demand and remaining on legacy transcoding service provider? How can studio have influence on the number of streams that platforms have concurrently? 4-5x estimations were right since from that time I expanded my gpu capacity 4-5x.
You are looking at some inaccurate minutes charts which does not say anything and bring up conclusions. The current status is: “The client is entirely on board and really happy with Livepeer but have changed their GTM approach a bit, so traffic has been lower than anticipated. That said, we do expect continued growth in both # of streams and overall minutes transcoded from that client”. Which says clearly they are onboarded and it is up to other factors whether the number of streams on their initial platform increases or drops. Nothing was true in your statement, that is why I call that FUD. You made a lot of noise without any data to prove your statements.
Also number of minutes does not say anything, since you can have 10x less minutes but with more demanding profiles which will feed 10x more fees. Whole traffic of picarto was half of that minutes chart but their profiles were mostly 480p which paid very little. Id rather have 10x less minutes with 4k streams and earn 10x more.
To give some example, Livepeer integrated Twitch (example again) and hopes to increase demand 4-5x, but twitch number of streamers dropped so the numbers are different but might go up or down in the future. The thing is Twitch is integrated and it is up to twitch whether they can attract more streamers which will feed more streams into the network.
But you said that twitch dropped livepeer integration and decided to stay on their current transcoding and gpu provider which is super inaccurate. Do you spot the difference or you still wanna spread some unmet fairytale expectations from your imagination? When were you promised traffic increase? I see the difference between “we expect” and “we promise”: Traffic - We expect this customer to 3-5x current transcoding minutes on the Livepeer network as they ramp up traffic over a few months. Orchestrator nodes should see higher utilization. We suspect the current capacity in the network can handle this increase, but would like to gut check with you on this.
Even if that was true which is not that gave studio a lot of insight knowledge and experience on the scaling and integration which in the longterm is gonna pay off, meanwhile all orchs can see the difference in stream count and are very happy from that integration.
1 Like