diff --git a/sig-economics/README.md b/sig-economics/README.md index 7d3a1018..2e112672 100644 --- a/sig-economics/README.md +++ b/sig-economics/README.md @@ -21,6 +21,7 @@ Meetings happen every [First Wednesday of the Month](https://calendar.google.com | #7 | Wednesday, August 2, 2023 10:00 AM PT (Pacific Time) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/007-2023-08-02.md) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/007-2023-08-02.md#transcript) | [Link](https://fc2diejm5vhix53dnrxh7dh2bflkuufterrrnmjb54ng4ojumtkq.arweave.net/KLQ0ESztTov3Y2xuf4z6CVaqULMkYxaxIe8abjk0ZNU) | #8 | Wednesday, September 6, 2023 10:00 AM PT (Pacific Time) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/008-2023-09-06.md) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/008-2023-09-06.md#transcript) | [Link](https://nlzzmaec3zpmorkyh3f4w3tep2hy444funhf645cqroo7x4qtusa.arweave.net/avOWAILeXsdFWD7Ly25kfo-Oc4WjTl9zooRc79-QnSQ) | #9 | Wednesday, October 4, 2023 10:00 AM PT (Pacific Time) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/009-2023-10-04.md) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/009-2023-10-04.md#transcript) | [Link](https://zq6rzise7b65e5o4uq2t2dgfvdu5fjj4p2aluihozhraqk7eyyca.arweave.net/zD0cokT4fdJ13KQ1PQzFqOnSpTx-gLog7sniCCvkxgQ) +| #10 | Wednesday, November 1, 2023 10:00 AM PT (Pacific Time) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/010-2023-11-01.md) | [Link](https://github.com/akash-network/community/blob/main/sig-economics/meetings/010-2023-11-01.md#transcript) | [Link](https://6cjhlgaiutfckbasf6sqjbypkrhyjgll463t67f7aw3af5gtghea.arweave.net/8JJ1mAikyiUEEi-lBIcPVE-EmWvntz98vwW2AvTTMcg) ## Leadership diff --git a/sig-economics/meetings/010-2023-11-01.md b/sig-economics/meetings/010-2023-11-01.md new file mode 100644 index 00000000..9adca6b1 --- /dev/null +++ b/sig-economics/meetings/010-2023-11-01.md @@ -0,0 +1,392 @@ +--- +title: "Economics SIG Meeting #10" +key: "sig-economics-meeting-10" +date: 2023-11-01T18:00:00Z +contributors: + + +- Bilal Khan +- Cheng Wang +- Deval Patel +- George Pro +- Halil Kerimoglu +- Kelvin Rodriguez +- Max +- Mousum XXX +- Pyakhurel Pasas +- Robert Del Rey +- Robert Koschig +- Rodrigo Rochin +- Scott Hewitson +- SGC | DCnorse +- Tyler Wright +- Zach Horn +- Zach Ross + + +--- + +# Akash Network - Economics Special Interest Group (SIG) - Meeting #10 + +## Agenda + +* Updates since last monthly sig-economics meeting. +* Open up to community for ideas or concerns + +## Meeting Details + +- Date: Wednesday, November 1, 2023 +- Time: 10:00 AM PT (Pacific Time) +- [Recording](https://6cjhlgaiutfckbasf6sqjbypkrhyjgll463t67f7aw3af5gtghea.arweave.net/8JJ1mAikyiUEEi-lBIcPVE-EmWvntz98vwW2AvTTMcg) +- [Transcript](#transcript) + +## Participants + +- Bilal Khan +- Cheng Wang +- Deval Patel +- George Pro +- Halil Kerimoglu +- Kelvin Rodriguez +- Max +- Mousum XXX +- Pyakhurel Pasas +- Robert Del Rey +- Robert Koschig +- Rodrigo Rochin +- Scott Hewitson +- SGC | DCnorse +- Tyler Wright +- Zach Horn +- Zach Ross + +## Notes + +**Introduction and Updates** +- Scott Hewitson initiated the meeting, gave a recap of meeting #9 and any outstanding items in discussions or on-chain related to sig-economics. + +**Discussion on Provider Incentives** +- The initial phase of the incentive model may be manual and target a select number of providers for ease of monitoring and to prevent exploitation of the reward system. +- A performance-based reward system is being considered, with uptime being on of the metrics for evaluation. +- The challenges of purely algorithmic reward distribution are acknowledged, with a human oversight component deemed a solution to manage any unforeseen issues. +- The need for a robust monitoring system for uptime is discussed, including the potential use of automated agents deployed on providers' systems. +- The importance of flexibility in the uptime measurements is raised, suggesting rewards could be adjusted based on the scheduled availability of providers to accommodate various use cases. +- The discussion emphasizes the need for a consistent and accurate measurement of uptime across tools to ensure fairness in the incentive model. + +**Community Questions** +- **Question 1**: Max asked about the difficulty of using IBC denominations of USDC and if there were any solutions or discussions of native USDC. + - **Answer**: Cheng stated improving UX in the current state is probablt not advised, once we know more about the techincal implications of native USDC it will be dissemenated to the community. +- **Question 2**: Robert Koschig asked if the provider incentives framework has been created or what stage it is at. + - **Answer**: Cheng responded by saying it is still in a draft form and will post those details in sig-economics on discord. He'd like to seek some feedback or suggestions from the community. + +**Closing Remarks** +- Scott asked for any final thoughts and promoted Discord as a platform for continued discussions. + + +### Action Items + +- Cheng will consolidate the current ideation for provider incentives and post in discord for feedback. +- All are welcome to contribute! Anyone who wants to contribute please visit sig-economics in the [Discord Server](https://discord.gg/akash) + +## Transcript + +This editable transcript was computer generated and might contain errors. People can also change the text after it was created. + +Scott Hewitson: All right, everybody today is the 10th. Sig economics meeting it's a little Milestone here. + +Scott Hewitson: So for today, we'll go through some of the things we talked about lasts meeting and then go through any discussions and Discord and then I'll pass it over to chain. It's given any major update and then we can open it up per usual. So last time we kind of talked there are some questions about the idps and community pools and we kind of said that the community pool is acting as all for now. So as IDP gets further ironed out and there's some engineering bandwidth to complete it. That's when there's no real time as of right now, but it's on the horizon. Talked a little bit about bounties formalizing that a little better Tyler and I have been working. + +Scott Hewitson: On making that a better program that's coming from the community support proposal five grand a month. Andrew had a question about tracking the community pool funds for a stable payments because there is some usdc in there. It's 45 cents worth. So I'm gonna drop a query that anybody can Use to check the amounts in the community pool. Actually did it earlier and I was looking at this so if you put it in your browser, it'll show you that's akt amounts and anything IVC. It should be usdc but if you didn't know IBC tokens are non-fungeable. So depending on the route that they come they will have different. + +Scott Hewitson: IBC numbers, I don't know what they call it IVC token order. So there's that. + +Scott Hewitson: ' Let's go through the Discord. We had folks Brought up some discussion about using usk it's a kujira's stable coin. So there's a little bit of discussion in there. I know Andrew asked if If he's here welcome to talk about it. If not, anyone can straight talk about it. Let's see. What else is in here? I know Piper Dev brought up. Some usdc gas payment feature. There's some discussions about that. So if you're on the call, you want to talk about that at all or have any thoughts on that? + +Scott Hewitson: And then yeah, I think that's it from the recap and last meeting and then talking about any updates since I know in steering committee right after this, they're probably gonna bring up the proposal that's live right now. And that was a lending pool for nitron. So we don't have to touch about that. at the moment. But if anyone feels strongly we can talk about it here as well. Yeah, that's about the recap and Cheng. I'm gonna pass it over to you. if you want to give any major updates and we'll go from there. + +Cheng Wang: Appreciate it amazing update and summary There's not too much to move on my end. Excuse me. I'm still getting over a little bit of a cold. I'm like 99% just have this lingering cough. So, excuse me, if I do cough. + +Cheng Wang: Close to the mic and blow your drums out. The big thing really is going to be that I'm looking forward to that is still under development in various stages just simply because it's just distractions and things and I'm not this old person working on it. I think I internally myself Greg where I'd eating around the provider incentives program right that we talked about before and one of the reasons for increasing the community pool tax is to get the pool ready to fund these types of efforts and one of the reasons why it's taking long is because just kind of peel back the curtains and breaking the fourth wall a little bit if you will is the complications around what + +Cheng Wang: in order to launch the incentive program. We definitely want to do it. but at the same time we don't know what is exactly right without actually going out there and testing it a little bit right? So there's some internal debate around what size we should calibrate it at how Loosely we should formatted or how tight we should button it up and it's really just around those particular parameters and what is the length of time? So for example instead of Designing a super? + +00:05:00 + +Cheng Wang: There's some thought hey should we just design something very programmatic get it deployed and have it run programmatically so that there's very little manual intervention that needs to happen. The challenge to that is you need a sink a lot of design effort and potentially engineering time into getting something like that across and there's a high likelihood of that not being the correct design simply because you don't know how the market and how people will react in the wild when things are actually implemented right kind of Mike Tyson at the same. Everyone's gotta got a plan until you get punch them out and better punch than the market reminding you what's right or wrong. + +Cheng Wang: So my thought is to get something done. That's lightweight. Let's say small amount, in the tens of thousands or when the low hundreds or something like that and instead of eyes the models that we know we want in some small quantity over a set period time. Let's say three months six months nine months, whatever the case may be right? Maybe not nine months so that's a three to six month and see how that pans out. Right with the participation rate is what people say what the feedback is all that kind of stuff and just kind of pull the trigger and do something really lightweight and get the thing going. So that's currently what we're internally hashing out. I do expect this to be hash out in the next couple. + +Cheng Wang: Weeks so that we will have something to the submit for discussion and GitHub. Of course Nothing's Gonna Be I couldn't just happened, right? So we'll submit something like that and next and in the coming weeks is my hope and engage you all as well as others who aren't on this call, but hopefully we'll be reading or listening to this at some point in time for feedback. + +Cheng Wang: So that's really the big thing provider incentives getting that I think nailed down is gonna be pretty big because I think that's one of the things we're currently missing to really ensure provider stability as demand continues to be bootstrapped. Right and there's some additional incent non-sensitive, excuse me initiatives that are coming online very very quickly you guys all like the thumper foundational model training initiative that was passed some time ago right in proposals and work on that is going to hit the ground running very soon. There's some Alpha there and as soon as those things start running we'll make the announcement and of course as those things continue to progress we will track those and make announcements and share progress as we have tangible things to share. And so yeah, look forward to that definitely keep your eyes peeled and years perked for that effort. Other than that, I don't really have + +Cheng Wang: too much else in specifics to share + +Scott Hewitson: awesome Thanks for that update Cheng, going along with that, Cheng said we're trying to figure out the best way to go about provider incentives, having more minds thinking about it in my opinion is better than less mine. So if anyone has any ideas, please drop it in the cig economics or if you got something you want to show we can spin up a working group or something but really, the idea if we go high level right it's to get percent or to get providers on the network and stay on the network when demand might not be there completely. we demands kind of been + +Scott Hewitson: It's a metric that I'm looking at all the time. And when you have varying levels of demand, providers will spin down but we really want them to stay there so that when demand comes we're ready to go and it's stable. So yeah, feel free drop in Deval what's going on? + +Deval Patel: Hey, so I got a message from Discord user called Joseph. He's looking for some kind of for demo if we can give him he has some Enterprise users who wants to use the GPU and CPU for the AI purposes and they want to like a demo to Showcase where we can run the better workflow on So if we can fasculate I'm happy to facility meeting in between Akash and him right and if you can showcase him that what kind of use case we can provide through Akash for the cheap purpose He'll be happy to run a few workloads formed Enterprise. + +00:10:00 + +Scott Hewitson: That's awesome. If you could throw all that info into a Discord and cig economics, I can follow up with you me are tired even chain can follow up and yeah, I know that'd be great. + +Scott Hewitson: Thank you. + +Scott Hewitson: Let's see. So that's kind of. What I had at the moment. Does anyone want to talk about anything in between since last meeting or anything you want to bring up any ideas really? It's an open floor. Everyone here is pretty open to good ideas. or any ideas really, yeah. + +Scott Hewitson: I know I Andrew's got a question for me somewhere. If she's on no, he's not on it. Hi, what's up? + +Tyler Wright: Yeah, I know we might talk a little bit about it later. And you talked about the proposal that's currently on right now. because we have a certain amount of time and Searing committee again, we might talk about with the later, but I would love to hear I know that Huey you do a great job of responding to these discussion throws a number of people that see the potential and what a cautious today and are obviously trying to figure out ways Whether it be lending pools or economics ESC initiatives, and I just want to hear do we have a hard stance that we want to here? I know that's on the case by case basis you're respond to these people, but I just want to hear from you and other members on the call if + +Scott Hewitson: Yeah, for Yeah, so looking at the + +Scott Hewitson: Proposal right now. It's got 41% turnout. So it's reached Quorum and it's currently a no vote. I think I put my opinions. I don't know if I put in an Insider Channel or not, but really my thoughts on this they're asking it's a small amount of akt. It's like 5,000 akt and they're just using it to bootstrap lending pool. If you haven't been on nitron, I've never used it personally. I went on defy llama and looked at their stats and kind of see how much liquidity is there there under a million and liquidity across the whole protocol or whole lending protocol, you compare that to osmosis who has got 70 80 million in liquidity. So in my mind, it's like this is a smaller player. + +Scott Hewitson: they're trying to get tokens on to their platforms that people use them because the more liquidity you have the better prices you get when you trade or when you're lending you get better prices. So yeah, I mean another point I brought up was it is a small amount. and it's incentives for 90 days. So, once those 90 days are up it's like What more value is being brought to Aka I don't know and I don't know if those people keep their deposits there, I think fiber brought it up that it needs to be above the staking rewards amount for anyone to keep money there because why would you not just stake it or there's less risk and higher return? So those are kind of all the thoughts that I had around this. + +Scott Hewitson: I think I said I was 50/50 but leaning to know but then if it did pass, it's also a smaller amount smaller we do these little experiments and see what happens a little bit. So yeah, that's kind of where I was thinking Ty is that answer your question? or Cheng over to jump into + +Cheng Wang: Yeah, I go everything you said Scott. I'm in line with you completely there. I just want to add a little bit more context like I know. Sega economics here in terms of the economics effort. this is well within our purview to talk about here. So it's perfect to bring these up and in terms of where our focus and effort needs to be to really drive the Network adoption. coming from another perspective like what is our ultimate goal here at a koschi? + +00:15:00 + +Cheng Wang: Nothing right vaporware or other kind of POS economics stuff that we're all very familiar with not that those things don't have their time and place they do right exchanges taxes Etc. They all have their place and needless to save bootstrapping those things aren't easy. So, my heart goes out to I totally understand the condition that there were in it's kind of cold sore problem we have to deal with it ourselves, right and in supply and demand you need incentives for example provider incentives to help get providers on and ensure providers stability so that we can ensure that demand can come in and see hey you go into a store and you go into a supermarket and you're buying plantains, for example, you don't want the store to be empty of plantains. You want them to be on the shelves, And so those are things that we are very aware of in this kind of suicide of marketplace problem or just bootstrapping anything. + +Cheng Wang: However, at this point in time really mainstream adoption is the key right for us and focusing all of our efforts and funds towards that maybe not all but the vast majority of our time and energy on that I think is the right approach here and it's the right Focus right? Because what ultimately is going to make I mean, what is for example people in Cosmos or outside of Cosmos room critique, let's say the cosmos Hub right what utility are they providing? what is Adam 2.0 all this kind of + +Cheng Wang: Often the feedback under critique. There's nothing like super tangible, right and for a caution Network, we're doing something incredibly tangible and as hard as hell but it's finally really gaining traction now and we have incredible stuff in the pipeline coming down where our did the internally I think what I like to say is we need to demonstrate. I mean for all of us here, we probably know in one way shape or form, right but we need to show the public time and time and time again. Anything the traditional Cloud can do a cost can do. + +Cheng Wang: We believe that we know it's possible. Right and we're marching towards that every single day. We're working on this project, but it's about demonstrating that as much as we possibly can and continuing to show the public what we're capable of so, I think the Sega economics as well as all economics efforts currently should be in support of one person's opinion should be vast majority in support of that particular effort to drive adoption more so than anything else + +Scott Hewitson: I think that was greatly said I just add one last little bit to that. So I taking his whole thing and add it and Distilling it to one sentence. I think the community pool should be used only for organic Network if things that will drive organic Network spend and lending pools trading like it's kind of secondary. so thanks for all that thanks for all that Cheng and Ty for the question. + +Scott Hewitson: Does anyone else have anything they want to bring up any questions floor is open for anything? Really? + +Scott Hewitson: Yeah, Max what's going on? + +Max: Hey guys. Yeah, like you said I just had a question about the usdc on the cash and I know you spoke a little bit Scott in the beginning that depending on the channel that you use for IPC transfers you get a different IBC denim. + +Max: I wanted to know what's the plan to? address this issue and I remember hearing about maybe native usdc on a cash and stuff like this. So yeah, what are your thoughts about the usage of USBC and also for providers the incentives to use usdc or I think there's an incentive to use akit instead of usdc. Maybe around but yeah, that's about this. + +00:20:00 + +Cheng Wang: Yeah, I would say yeah. + +Scott Hewitson: changed you want to + +Cheng Wang: Yeah. Sorry, I would say in terms I'll address the incentive thing first because that's the easiest thing to address for using it was in the design of I think when you're saying the incentive to use akto or usdc you're kind of indirectly referring to the take rate, right? Because the take rate is so much higher on usdc than akt there's an incentive to use the native token for payment processing. And I think that is definitely the case especially early on right where folks are trying to especially now with + +Cheng Wang: The price action being what it is and kind of the momentum there. I think people are viewing more of a hold on akt, right then sell on AKC. So in that regard people are saying hey, what, I would like to risk optimize for more akt if I'm a provider and also I get to keep more of it because the take rate is so much lower. So there's definitely that and that was intentional in the design. I think we talked about it too right where we want to ensure that native token always has the kind of the leg up if you will in that regard usdc and other tokens are associated will always Prefer, obviously, these parameters are all changeable via governance, right and that's the whole idea of it. But I mean the original intention was to prefer higher take rates for other tokens because again it doesn't + +Cheng Wang: Because it doesn't use akt directly and we want to ensure that the value capture as well as the value creation to akt is never broken and that chain remains strong to ensure it's Implement rather a higher take rate for these other currencies and that take room goes back to the community pool. Right? We're currently in the community pool in the future some IDP sub pool, whatever it is, that adds value continues to fund the development of a caution Network, So again, I know this already Max of course, but that's just the loop right and that's the reason why that exists going forward using us DC natively, I think I don't know what the exact. technical + +Cheng Wang: constraints are precisely to make that happen what we do know for example in arguments at least whom we're close to is kava has already done native usdt via kava. So we're trying to get scheduled time with their engineering team and our engineering team to talk and understand. + +Cheng Wang: What it requires and what that kind of looks like because supporting to just understand what the technical guts are necessary to support native usdt and Native usdc right the end result. I don't know the technical bits are but the max you're so that me probably definitely but in terms of the end result, it'll be as if you're using Akash or atom or anything else it'll just be another token within Cosmos and it'll be fungible there wouldn't be usdc via osmosis on Akash, It's like and then that is only in order to get back to you as DC. You have to take it you see on Akash send it back to osmosis. + +Cheng Wang: Use the apps of bridge back to usdc in order to get usable usdc on exchanges right or something like that or to put back into your own wallet somewhere or something like that. So, we're trying to understand what the technical implications and requirements are to support your CT natively on Akash, that's step one. And that should be happening with the next week or two that conversation the very least and then from there we'll be able to better Define what are the next steps as far as building support for it? And on either right? I'm pretty sure these are core implementation, And so that's where we currently are. So once we kind of bridge that Goldberg That we're currently in this initial really rough implementation of uscc. It'll be a lot easier and things will be fungible. I think I'm pretty sure 99% sure. + +Cheng Wang: half the time + +Max: Yeah that makes sense and perfectly answers my question and I guess follow-up question. do you think it's worth putting effort right now to improve the ux to use usdc from osmosis or not worth it, since it's not very good ux in terms of IBC and non-fundibility of + +00:25:00 + +Cheng Wang: I would if I were betting man, which I sometimes am. I would say probably not worthwhile improving the ux around this current IBC denom shenanigans. I would probably in order to make a more. Well informed decision I think will have the conversation with coffee scene is AP. Like I said next week or two once we have that conversation. I think we'll be able to judge a little bit more. With the greater degree of certainty and understanding what the right path forward is for, staple coin Not true, but direct stable coin support rather than just whole hopping around thing. Yeah. + +Max: Sounds good. Yeah, I think I agree with your bet I mean if the conversation and all the details are to be talked about next week or soon enough. definitely not worth trying to + +Max: make the IVC then stuff in the department experience easier since it might be absolutely in a few weeks. So yeah, All right. Thanks. + +Cheng Wang: Yeah, absolutely or just a cleaner implementation right instead of designing around something that is suboptimal to just have a cleaner flow in designing around that I think closer to where the end State ultimately should be right. + +Max: Yeah, for sure because right now you need to know the exact denim to put in the sdl for providers to bid on it, but even us as developers, we had a hard time figuring that out. So can't imagine for anyone else. + +Scott Hewitson: Ross normal Focus impossible + +Max: That's exactly… + +Scott Hewitson: just Okay. + +Max: what I meant. + +Scott Hewitson: I'm just playing. + +Scott Hewitson: No, that was a great question. And thanks for the thorough response string would appreciate that. Yeah, Robert what's going on? Good to see you here. you got. + +Scott Hewitson: maybe he'll jump back on. Or maybe he was trying to leave and not raise his hand. there we go. + +Scott Hewitson: Hey Robert. Were you trying to ask a question? + +Robert Koschig: Yeah. He's basically to check in back on the topic of provider and this in ideation. is there already something or you really start basically from zero? + +Cheng Wang: There are some initial thoughts Robert circulating around and like I said, we'll definitely share a collated higher conviction version of our thoughts summarize and GitHub discussions before anything goes live so door. everyone will have a crack at it. before these things go live in terms of will obviously we want to Share something that is cohesive something that is well thought out and has a reasoning explained behind it. So that when we get into the discussions, right people can really pull that apart and there's not a lack of detail. That's the reason why it's like taking a little while because there's debates of how should you do this, but I think we're coming together on a closer understanding at this point. It's just about nailing down things like + +Cheng Wang: The specific GPU models, we should incentivize in this first batch because what we want to keep in mind is this is going to be the first iteration. It's not like the end all be all. So if there's a particular model not supported that doesn't mean it won't be supported later on. You know what I mean? It's just hey, there's this one thing we want to do. Let's start and do something small but quick and so then get some learnings from it and the second iteration will be better in the third fourth Etc. So on and of course not build something over engineer something where we back ourselves into a corner and spend so much time on it and effort that it ends up having to be being changed, dramatically or completely. So yeah that's where we are and we'll get out I'll put gas on that and get that shared publicly as soon as we can I'm hoping in the next week or two is guess. But yeah, just that would be the goal. So that at least the next say economics? + +Robert Koschig: Okay. + +Cheng Wang: We can have something tangible up there that will have already been subject to some amount of public discussion that we can continue to hash out here if anything. Yeah. + +00:30:00 + +Robert Koschig: That was basically like… + +Cheng Wang: please I'll see economics. + +Robert Koschig: we can start here if we want because what comes to mind when we talk about,… + +Cheng Wang: For sure. Yeah. + +Robert Koschig: incentivizing up time one model that I like a lot is Hopper. Does it live clear? Does it the probabilistic payments? And not true if you know it, but it's basically, you don't get a reward like every block or whatever but it's like a likelihood that you get it and obviously you can say If you have a certain type of GPU that we need more than the other, you're likely to this higher and you need to collect not just one, lottery this week winning ticket but five and that's over time. So it's on average if you're up all the time because you randomly send out the tickets right to and if your life you can collect it. + +Robert Koschig: And that's why you have an incentive to be alive because if you missed it because you're not up and running and you're not gonna get it and obviously the downside of all this is This is a little bit complicated to implement. I mean, you can tell that I'm not a deaf so I can only make up a fancy incentive models. But yeah, that's one component for sure that comes to mind that that feels like one idea to kind of go after the uptime incentivization besides the other constraint. Obviously that when there is demand you kind of want to slow down the absolute bucket of + +Robert Koschig: This It's just inverse to when there's actually demanded you would just cap it or even shrink it down. So you only have a Baseline and whenever there's no demand people still get this minimum in incentives and if there's demand and this Baseline emission is zero, basically + +Cheng Wang: Yeah, I think this is Definitely closer to there are some elements of that are very in line with what we're thinking as well. So hypothetically I'm a provider that I have, 10 gpus whatever the case may be in those gpus are models that the network wants to incentivize because they're important to adoption and keeping the supply on so hypothetically when those resources come on, we'll say The idea the incentives is to create a floor right? I think that's where you're talking about here where if you're earning more than the floor you don't get incent you shouldn't also double dip incentives probably right or if you do it should be something smaller. And so the idea of incentives here is to ensure that providers aren't losing money. + +Cheng Wang: Per se by operating on a cost Network so that they can continue to sustain those operations. Whatever they're kind of like break even costs are and of course everyone's a little bit different but the way we can't calibrate it to each person's like economics situation, Every providers economics situation might differ a little bit. So we'll have to use some sort of generalistic approach to parameterize it and so yeah, I think setting a floor is a huge component of this and then anything beyond the floor is going to be just + +Cheng Wang: Gravy, I think that the provider keeps and not incentivizing anything beyond that I think there's some thoughts around it as well. do we just say instead of eyes the floor? Let's say your floor is a thousand a month right in emissions and you get the Thousand no matter what and the excess you keep is or is it if you make up to a thousand in actual deployments, then you don't get the rewards at all or some small amount, So there's some gradients there between those approaches and then the other thing is definitely in the initial implementation won't be by block period because that requires programming right and Engineering effort, so I think it'll largely be limited to some small amount. Let's say 10, 20 or 30 providers, whatever maybe or chip types maybe like 10 two or three four types and maybe 10 or 20 providers something like that depending on how that's + +Cheng Wang: Framed and the economics will have to match. Those things in order for people to come online use it Is it enough incentive? So those are kind of things because what we don't want to do is overpay, it's okay A little bit especially if it's small scale, but we want to hit that sweet spot right where incentivizes people enough to join and keep on but it's not so much where people are just going to necessarily arbit right to say, what I'm getting five grand a month for these machines that I can go out and lease on some Cloud for 200 a month. Yeah. Okay, of course, I'm gonna do that, we all want to create too much of that scenario either. + +Robert Koschig: yeah, I mean that makes it almost sense and I really like also the kind of cost markup a floor anchor so to speak that's Yeah,… + +00:35:00 + +Cheng Wang: anchor here + +Robert Koschig: really good. Obviously and if you then even tie it to a limited amount of providers because then you kind of avoid the issue of if a cash token all of a sudden goes up and you don't adjust, all of a sudden you have 500 people coming in just for the farming so to speak. Yeah, that's for the things you want to avoid. + +Cheng Wang: Yeah. + +Cheng Wang: Yeah, and I think the initial implementation will be pretty manual to it'll be like, hey these 10 providers that we're incentivizing we know their provider addresses, right? And so we're going to monitor those providers for uptime and we can easily keep track of it. and not necessarily give out the rewards per blog period but her X interval whatever right by weekly monthly whatever the case may be in order to and I think manually process the processing those things will be a little bit painful in the beginning but I think it'll give us a great degree of understanding and control so that we don't get into How their algorithmic runaway scenario where one of the snare is a yield Illustrated, for example. Happens yeah. + +Robert Koschig: Yeah, but for sure interesting right to think about the way how to incentivize that time because right that the probabilistic payment stuff is more like, you put the pressure on the Node operators to be like, yeah, if you're not up, it's not my fault. You need to collect the stuff whereas there are other networks right that more we measure the uptime nimm for example, does it they have a performance factor for the rewards? So you get rewarded according to how could your performances and the performance basically is uptime. So right over the past 24 hours if you were just the ping I think in arbitrary interval and if you've been up only 80% of the time, you get pen penalized on the rewards that you would actually get if you up fully 100% So it sounds like in the beginning to keep it simple. You're more like using this type of measurement where you will just look at the actual data window when they were up. + +Robert Koschig: Is that something you could actually do also a lot larger scale more permissionless to kind of measure up time. I mean also I don't want to steer too many, spot here. I just like the discussion. So if anybody else has other questions for sure stop us. + +Cheng Wang: No worries I think you had something to go, but he had a Henry's. + +Cheng Wang: You're 100% right? Yeah, that's a great idea great point and I think in between right? So I'll collate the existing thoughts and share them in say economics so we can get the discussion rolling totally agree. I think it's a great idea and I'll do that sometimes today for Love it. Yeah,. + +Cheng Wang: Exactly public accountability. I'm gonna do it today 50% Just Let's see. What else I think wherever you asked a question previously. around Yeah,… + +Scott Hewitson: No, I was just as we're talking about this Robert really appreciate you coming in. + +Cheng Wang: I think is there a programmatic way to monitor uptime in those things like that? + +Scott Hewitson: You've been in the past two and… + +Cheng Wang: So I think A Deval was on the word,… + +Scott Hewitson: you always have good things to say. So I really appreciate that. I was gonna ask you Cheng do we have something that we can throw into cigar economics and… + +Cheng Wang: but I think he dropped but yeah as far as praeter and stuff like that that would probably lean on their Tech to monitor this stuff. + +Scott Hewitson: Discord just maybe a rough bullets of… + +Cheng Wang: I probably believe there's that kind of capability and… + +Scott Hewitson: what we're thinking, kind of how we've been going right? + +Cheng Wang: in addition to that once we know their Address right on the provider address we can easily monitor via prater's dashboard. + +Scott Hewitson: Some internal folks right some stuff and then it goes on discussions and it's all right, that's kind of the 90 percent product and then any final tweet … + +Cheng Wang: I don't know… + +Scott Hewitson: I think there's interest and… + +Cheng Wang: how far back that information goes… + +Scott Hewitson: some smart people out there that can help steer some of these things. + +Cheng Wang: because they do show you up time. I don't know if it's a two week period or whatever it is,… + +Scott Hewitson: So if we … + +Cheng Wang: but what we could do is whatever that + +Scott Hewitson: put out a 30 or 40 percent, some bullet points of this is… + +Cheng Wang: Time period is let's say it's a seven day. + +00:40:00 + +Scott Hewitson: what we're thinking get some inputs. + +Cheng Wang: I'm just throwing out a number… + +Scott Hewitson: You never know when … + +Cheng Wang: if it's a 70 period we can do rewards every seven day period right just say hey you had 20% uptime you get nothing… + +Scott Hewitson: great ideas come out and get implemented into that final or not the final but next step of discussions or… + +Cheng Wang: because you suck or you have 90% Plus or… + +Scott Hewitson: it's kind of like, all right,… + +Cheng Wang: 99% plus up time you're grades. + +Scott Hewitson: we've already made up our mind almost + +Cheng Wang: That's what we want. Then you get it and we can do some sort of thing. That's a great idea too late some sort of gradient where beneath a certain amount of uptime you don't get any rewards and some Take a percentage that you are up you get additional rewards right to incentivize greater uptime too. I think is reliability and providers. That's… + +Scott Hewitson: awesome And… + +Cheng Wang: what that does right you want someone who's like having it on all the time not On and… + +Scott Hewitson: he doesn't do it today everyone hit him up on Discord. + +Cheng Wang: off on and off. Okay. + +Tyler Wright: There's a case where maybe somebody has a100s that they need to use during the day, but then they want to make them available on a cost Network for 10 hours and they know that 10 hours is every single night of specific time and it's scheduled. We want to make sure that they're not being penalized for something. that we would want to encourage theoretically on the marketplace. I think there's other elements of the uptime metric as it is stand today that could use some tinkering but I think that as you all think about incentives up time is certainly very important, but I think it's just one piece of the provider puzzle, obviously GPU type is another piece that you all know about but I think it's just a piece of that puzzle and determining that Precision determine for incentives. + +Cheng Wang: Only with you. good point + +Max: Yeah, Thanks for bringing that up and I was aware that there was this good discussion in the signal it takes about this. But for Maybe one solution to this is that providers could specify the schedule that they want to be online and we could base Based ourselves on this to calculate the uptime. That way if it's scheduled to be online for just a few hours during the day. We can check that and also we've done some work. instead because currently the way we check the updine is just by fetching or pinging the status endpoints from providers. + +Max: And sometimes that isn't very reliable for some reason some providers are actually online, but their status endpoint they're making some changes on their DNS config in the status point isn't online, but their provider is online and they work running. and so yeah, we made some work to work towards deploying and agent on each of the providers automatically So there would be at least one deployment on all of the providers. that runs the agent and gives us the actual API endpoints that we would query on that. + +Scott Hewitson: Max good + +Max: specific deployment and we can also do some benchmarking and + +Max: other types of metrics to gather more data about the providers Which there could be used also for the incentives? Maybe just thinking about this right now. But yeah. + +Max: Maybe we can talk about it in a separate color something. + +Tyler Wright: Yeah, yeah, if you want to come to Sig analytics, maybe we can create a separate call where you pray towards some other folks get together because to your point Max at first use case. I was talking about where people are trying to Compute online for specific schedule period of time it's like a newer thing that is being explored. But the way that uptime the second point that you're making about how up time metric is taking right now is something that I think is worthy of further discussion amongst the team so we can make sure it's the same across all tools and makes sense for everybody. + +Tyler Wright: So I'll set up some time for us to talk about that whether it's next week and Sig analytics or prior to that. + +00:45:00 + +Max: Sounds good. + +Scott Hewitson: awesome Does anyone else have anything else that they'd like to bring up to the group? We've got about 11 minutes left of the scheduled time before steering committee. + +Cheng Wang: Please do definitely considering Community if you guys have the time. + +Scott Hewitson: Yeah for Already, it looks like we don't have anything else. So action items anyone who has ideas wants to contribute to The molding or developing of the provider incentives, Please drop it into Sig economics and Discord and Cheng is also going to drop kind of what the internal teams thinking about at the moment very rough. Nothing set in stone. So Yeah,… + +Cheng Wang: Don't judge me. + +Scott Hewitson: that's the big action item this one. With that, I think we can conclude today's session and as always appreciate everybody and we'll see you here shortly. + +Cheng Wang: Thank everyone for coming and participating. + +Robert Del Rey: Stay safe guys. Thanks. + +Robert Koschig: Thanks. + +Cheng Wang: As well. Bye. + +Meeting ended after 00:46:20 +