Practical endgame on issuance policy by Anders Elowsson | Devcon SEA

[Music] the stage is yours thank you and welcome to this presentation of a practical endgame on Ison policy which is also the title of my recent East research post and we begin in a hypothetical future scenario a decade from now where the cost of staking has come down and staking frictions have been overcome now people still have different perspectives of the cost of staking and Stak under varing conditions so there's really no agreement on the exact yield that is sufficient for staking people have different reservation yields that is the lowest yield they find acceptable for staking they are at this point indifferent to staking or holding the token let's plot the hypothe hypothetical distribution of reservation yields we see here we have Stak in yield on the x-axis and we have amount of stake on the y- axis a few might be willing to stake at the negative yield perhaps to attack or defend etherum many more are willing to stake as the yield becomes positive and at 1% staking yield perhaps a lot of token holders rushing in this hypothetical scenario at two 2.5% half the E is St and as the yield reaches beond 4 5 or 6% not much more steak comes in because almost all e is already staked and we can explore this further by Computing the cumulative steak here in blue as it raes to one the cumulative stake can be referred to as the deposit Dru that is the proportion of all steak that is deposited for steaking and we can also call it the deposit size then we are referring to the quantity of steak the total reaching 120 million if all eat the steak now let's remove the negative part of the yield and flip AIS and we have now created from a distribution of reservation un what we generally refer to as the stake supply curve so if you remember at the 2.5% yeld half the it gets St we add also the demand curve specific by the protocol and let's add also 300K of Mev each year we then reach at the intersection of demand and Supply what we refer to as the stake in equilibrium at the black square but currently we are here and this blacker isn't exactly known to us so that's the situation we have right now we are now ready to discuss the motivation for a reduction ini ISS if you note that over half circulating Supply could be sted in a a plausible scenario but it's this General agreement amongst researches that 30 million e is perfectly sufficient for retaining security and here we would be at more than twice of that so in on this range here what's the cost why don't we want to pay more for security than what is strictly needed well for one we can approach cost rather generally speaking of you know the hardware that you need to pay uh the taxes you need to pay for you know you you have a liquidity you need to finance the upkeep of of your software Etc and it's then natural to approach the cost as the reservation yield if you remember that's the lowest yield at which you are willing to stti and then since the supply curve captures the marginal reservation yields as shown previously you can compute the total cost for staking in aggate for the protocol as the integral of the supply curve and so in this scenario oh there a picture can I remove this yeah and so in this scenario the the total cost of taking becomes $ 820,000 e for this specific supply curve that's $2.6 billion there's also a surplus uh above the supply Cod so so the marginal Staker with a low reservation Shield will have a high Surplus and uh the marginal Stak will have with with a high reservation yield will have no no Surplus at all let's now reduce isance with the red reward curve this then pushes down the staking equilibrium down to the red circle here this gives us a cost reduction benefiting all token holders uh so we're removing essentially hard cost you know taxes we're removing liquidity for users and the risk premium and benefiting all token holders in the form of a lower amount of newly mined eat and that's around $1.4 billion at the current token price we would save in this hypothetical scenario there's also serface shift from stakers to allil Holders but this just shifts value run from one set of juices to another the cost reduction that's the welfare gain for and there a wonderful opportunity we have here to reduce the cost of our uses you know uh a simple way to frame it would be don't buy things you don't need you know this is not a sustainable model and we of course need to focus on facilitating sustainable economic activity then some people say well the yield makes ethereum attractive you know it's fun to stake but we have to remember that all eat that is issu we also dute the token holders so there's nothing gained you know in aggregate we shouldn't try to trick our users into believing so we need to focus on facilitating sustainable economic activity or some people say well we need inflation otherwise people just hoard the tokens well that's not true first of all consider the fact that if you issue a high amount of token then you compel users to lock up the token staking instead of having them available had you know for paying your transactions furthermore consider the fact that of course if you if you if you focus on what we're trying to do you know the the the need to transact is universal it's not really affected and if we ensure that ethereum facilitates sustainable economic activity and that e remain sound then the best option ofil producers is to be on ethereum transacting using e and so that's what we have to strive for and then some people say well apps rely on the Yi you know framing the yield as somehow productive it's not and it's not our role unfortunately to support applications that do not facilitate sustainable economic activity that's not our job and if we do so you know a high isance would always over time turn into perverse subsidies there's also a macro perspective to consider High s participation puts pressure on the consensus layer here we have the current scenario around 35 million e St but imagine a scenario where almost all e is sted and then something breaks say there's a discouragement attack the censorship or there's U some client BG anything else in this scenario ethereum relies on the social layer as a final recourse but what happens if the social layer is sort of wrapped up everyone has stake in a game well in this case the social layer sort of becomes overloaded and we have come to a point you know where we can lead to very aggregated failure modes we would then very much prefer the bottom scenario where we have a neutral social layer available highest St position also can lead to of course higher attestation load and further more it you know if uh it put pressure on the app layer when one or or few lsds comes to dominate this money ethereum then is deprived of uh a trustless sound assets which is one of the most important crucial features of the Central blockchain and and you can imagine if if you have control of the money you sort of can control uses and apps in in various ways that we wouldn't really like furthermore consider the fact that if apps build on top of the LSD and then apps build on top of the apps building on top of the LSD and you're creating this sort of mono and then something breaks then you are sort of threatening you know the whole the whole system here which is something we really would like to avoid so in conclusion even though nominally the protocol might seem maximally secure if everyone is taking it turns out that it's actually a bit insecure both at the consensus layer and at the up layer and we would prefer the bottom scenario here however there was a potential downside for adduction initials for example sud stakers do not have the economic of scales of large staking service provider so it's perfectly possible that a relatively fewer solar stakers might be willing to stake in the scenario where the is where the yield Falls very low there are arguments in the other directions as well but still let's explore this a little bit so we go back to the first plot of reservation yields and let's split up these reservation yields into two classes we have delegate Stiers and we have solar sters so loosely defined here now there are more delegated Stiers than solar Stiers so I'm just normalizing these two plots as well so one scenario is that if we lean on economical scale is that the resolation yield among delegated St might have a peak in distribution a bit lower say at 1% whereas the peak equal solar stakers is at 2% well in this case we would indeed see that the proportion of solar stakers at the lower isi will be slightly lower the argument in the other direction is that at a very high stake partici position and LSD comes to dominated money so there's you know no loss in liquidity if you own the LSD big seconding surface providers also get better economic of scale and you know it may be perceived as that the risk of staking with them Falls because they become you know too big to fail there's no social layer there anymore so that's a counterargument and that could happen at any year depending on the exact shape and slope of the supply curve M also becomes a much bigger concern for us as lower ISS it increases solar Stak is relative yield viability H you know because they cannot effortly pull the rewards furthermore it degrades the incentive to attest properly since me will be the only thing that counts so there's a lack of incentives to attest and we can then consider various way how to resolve this you know we can try to increase the attestation penalties all right so now we increase the attestation penalties but then if you are staking service provider and you drop attestation of your competitors then then you can you can sort of force them out so then we have to have proposed penalties as well all right now we added proposed penalties but then if you're a soul ster and you make one bad proposal you can get punish very hard because we have low additions as well you know and you might be in the red for several years all right and so then there are ways to try to soften this but there's a lot of things going on at the same time here that we're working on and it also further complicates the design of orbit as a you know the sing slot finality design because then there's also a way for stakers to avoid void having to attest as much as possible taking as low risk as possible and so the lack of incentive to attest can then lead the S to deconsolidate and then we need to raise the consolidation incentives and this can of course further raise fairness concerns and what we would like to ship is murn because this reduces all these Tre issues so that's something that that is really important for etherum but we don't have it and we have to deal with the S we have we're now ready to discuss a practical endgame on isan's policy so here we have the current reward curve with isan Rising all the way up to 8 1.

8 million e uh with the current reward curve here we have the redward curve shown earlier it forms by taking the equation for the current word curve and adding the cube deposit size to the denominator and I refer to this as a practical end game it's practical because it's something we can strive toward watch without murn it's practical because we should be able to guarantee Sol stakers to not have to pay for staking and it's practical because it allows us to design consensus incentives in an easy way and and it's also practical and it's an end game because it allows us to probabilistically stop the grow in stake so that's why I favor the red reward cover or something close to it this shaded range is approximately where I would like to be then gives us room for a higher or a lower worker depending on you know how the design works out and feedback from the community others might prefer differentes for focusing on stopping the grow in Stak or increasing the yield for solar St and this can be discussed but you know if we go too high you can imagine if we say go to 110 million each state and provide anything about the bare minimum to avoid Sol sters pay you know then this would feel like a misbalance of this different Traders we're trying to to to balance and then there might be a movement to further redu ISS and we would like to make a One Singular change to isan's policy and that's why I find that this will be slightly above where we want to be and you note that none of the reward cures nor the Shaded range goes above an interest rate of 0.5% that's sufficient to ignore security to to ensure security and it's at the upand of any tradeoff I think think we should ever consider so regardless of the exact end game that we choose we could safely commit to never issuing more than 0.5% of the circul supply each year it feels like a rather attractive proposition you know tell your friends shoose ethereum shoose e there will never be more than 0.5% of the circulating Supply issued each year and a cap on ISS it's perfectly viable or Insurance security whereas a cap on the circulating Supply is not regardless of you are in proof of stake or proof of work and from my perspective we could just ship you know 0.5% cap solidifying it and then come back to addition a few years but there seems to be from the community ASP we want one single change so that's not something I would push for there are other ways to construct a reward curve I'm showing here other other Alternatives in in the it research post like sigo shapes for example uh with the equation prided there let's now look at the impact of an insurance reduction on stakers we have here the staking yield and we see that we are uh showing it with Mev now we add the supply curve and we are able to reduce the equum from 60 million down to around 40 million state but of course we are here and so given some reason we might end up in this region here we must also consider a very low supply curve and what would happen in this scenario and here we have that bad situation for the M micro perp when all almost all it is St and we could then bring it down to around 65 million e with the red reward C in this scenario and of course shiping meone as well we come down to almost 50 million e State and it might seem like a bad scenario but it turns out that it's actually better for SE to be at the Red Square there than at the black square and that happens because of this equation uh it's from my post minimum viation that sort of started this whole conversation and it captures the attainable change in the proportion of the circulating Supply so it's sort of like a proportional yield or it's been referred to as the real yield or the change in ownership we have the yield that you attain on top of the fraction line and we had the inflation rate at the bottom of the the uh fraction line that could be negative you know if there's a if there's a high burn rate but otherwise positive and we can use this equation to draw uh something I refer to as isopr proportional lines and at these lines across these lines the proportion of the circulating Supply that you attain remains constant so it doesn't matter if you're at 1 1.

5% yield 2.5% yield or 3.5% yield from the perspective of the stakers once you account for delusion effect if everyone almost Stakes you get the same proportion of the circular the supply let's draw more of these lines it's like a topographical map you know with counter lines indicating equal elevation so at this map you want as a St to be as a high elevation as possible this equium at the red uh area is not so good even though the yield is high because you're at low elevation any equium above the dashed curve is at a better higher elevation so the orange area here is better for you as a ster even though the yield is lower and here we have the supply curve from previously and we see that if you start here at the black Square you have 110 million each Stak than say a 1.8% Stak yield well it might feel good as a Staker but you follow this is a proportion line the way to the other equum that's possible on the disip curve and you see that you're equally well off as a ster at a you know 0.25% is on SE and 20 million State because you not suffering from delusion and the optimal situation here is actually at this at this red square here where you at the highest elevation is possible on the map but this is just Surplus shifting around you know it's not the motivation forance the the motivation is for inance reduction the motivation is the cost reduction but it's important to understand this as well it's very important to emphasize this and just as a concluding remark a checklist on the Practical and what needs to be done there needs to be a theoretical uh foundation and we have shipped it there's a cost reduction and there's the macro perspective we need to have curve parameterizations and they're available in the post and elsewhere uh we need to design orbit ssf consolidation incentives and I I posted yesterday uh a post on that so I would put it as yellow because it's a draft post but it's sort of outlines approximately where we would like to be we also need to redesign eum's consensus incentives a bit if we push is down close to Zer and we have been working on this a bit but as you noted the discussion about this R senses but there's a few things quirks to work out so I put at the orange if we push met Buton both of these becomes Greener because you know then the issue isn't that big anymore I also working on a probabilistic equ analysis that I will post uh in the first quarter of next year and that will uh I so I will put it at Orange and that would be like a foundation for my perspective of why we can do it uh from the perspective of you know decentralization and we also need to build community support and I'll put it at Yellow by turning the other check marks green we can also build community support better you know and we build community support of course by striving for an ethereum that facilitates sustainable economic activity that is what we're working on in all these different parts of the roadm that is how we strengthen e and ethereum and that is a suitable North Star in my opinion thank you [Music] [Applause] thank you so much endis now time for questions first one what do you think about reward corbs with Dynamic parameters yes uh I I have written about this actually there's a fq and I talk about the issue if if the question relates to the issue of trying to know dynamically I adjust the work the point is that it's not a way to to address the issues that we have in the long term in the long term it doesn't matter you know we need to focus on the one reward curve because that's how we prioritize this these different aspects you know so I think that there's there's the idea of you know having some Dynamic solution there's there's there's a threat in it that you think that you're solving something but you're not solving it you're not solving it in the run run the only reason why we would like to have a dynamic curve is if we have if we go back here oh I can yeah but if you have an equilibrium where you have to lower the isan yield so much in the beginning then of course and I describe it a bit in fq then of of course you could consider a scenario where you have like you tick down the reward curve gradually over a year you know so that you don't make people uh get to lose uh the staking so quickly making like it becomes like very bad B effects on everyone if you just shook it because otherwise the assumption is of course that some are going to leave so the yeld is going to go up again but but but if you shook it down and then uh the year becomes very low there's a risk that the solar Stiers are going to live before the year comes up again next one how does the beam chain proposed road map affects progress on issuance policy well so my interpretation of uh how how it affects is that is it's essentially um something that is sort of part of the idea is that we we try to ship asan policy as one of these gradual improvements it's not it's not like part of the beam Fork it's supposed to be done earlier so I would say that it's sort of part of the overall role map there yeah thank you what would what would you say is a good governance process to agree on a new issuance curve uh rough consensus that was a quick one um why does me burn affect our needs for consolidation incentives in orbit S ssf no it's not M burn that effect it's m so if you design orbit s like a Bas basic idea of how you can design orbit ssf is that you push out smaller stakers so that they don't attest as frequently while you at the same time give them the same yield so that's it's sort of fairness still but the situation is such that if you push out them from attesting you cannot at the same time push them out from proposing that's a a a lot of bigger concern because if we still have me in the situation then we cannot make it fair if we push down the ISS and then you still gain a lot of the value you gain almost all the value from proposing and attaining this me you know there's no M burn in place then everyone would like to be deconsolidated to remove risk but they still get to propose as often as today and so that's an issue great thank you so much the next we still got more time is there yield where you might expect institution to lose interest in staking or might find alternatives to generate yield long question does it compete with some traditional Financial yield threshold I mean the beauty of having a reward curve is that if there's a loss in interest in staking in general the yield goes up you know that's the whole principle so I mean it's hard for me to cons the the option would be here maybe implic implied is that that everyone wants to stake but no institution other than one or a few are still willing to provide services and I find it a bit perhaps tricky to contemplate that this would be the case but of course we we have to allow various institutions to compete you know we we cannot avoid that there's going to be a comp competition it's we would like of course for for there to not be but it's not something we can avoid should we go for the last one there's no V but I think since it's already here let's answer it there's already so much Financial investment by influential VC funded actors even at current levels do you think it's possible to make a change that drastically affects the bottom line yes okay do you can you a little bit more about the yes uh why do you think well I don't know I don't think that VC funded actors is is like the biggest issue for us it's I mean the biggest issue of course is if solo stakers would feel like it's it's a disappointment for them so Visa funded actors is the least of our concerns I would say in this great our time is up now thank you so much