DAO proposal - Delegation strategy update & fee structure changes

I don’t have numbers on-hand but the reasoning behind this change is that 10% of Marinade stake is supposed to be controlled by the gauges. If multiple validators are above the 1.5% cap and this stake gets redistributed to all validators, then gauges only account for 9.X% of the total stake and the rest is redistributed according to the algorithmic delegation strategy.

It makes more sense that those 10% of the stake stay between validators that received votes, and it also gives a bit more value to having a gauge up and some votes in them.

It’s also a change that had been requested by some validators that were disappointed seeing this stake being shared among everyone and not only active gauges participants.

3 Likes

@dobby I’m not sure you’re aware but a lot of the validators that run at 10% stake already have stake from the Solana foundation… no one in the US can kyc for foundation stake and the waitlist is over one year. For anyone not in the foundation we can’t really run at high commission because it deters people from delegating stake… so we run at very low commission (i.e. we run at 0%) or you run at a somewhat higher commission (usually 2% to 8%) and run the risk of being overlooked by delegators. It’s not an easy spot to be in… basically all non-foundation validators tend to take a larger gamble on costs (depends how much money you have I suppose). @knoxtrades and I are both quite new as validators (but have been in ecosystem for quite a while now). Other than us @pumpkinspool and Node Monkey are newish as well but more seasoned than us. There are some other more recent validators that I can think of. The issue is that even with more Marinade stake than currently… we are still quite some ways from hitting profitability if we want to stay more competitive with lower commission. Being outside of the solana foundation delegation program definitely has a stronger natural selection system… but perhaps too hard lol.

4 Likes

The planned changes will definitely bring more stake to 0% or 8% commission validators, and I hope it’ll help more validator in those cases.

Thank you for those remarks, it’s quite interesting to consider this too indeed

As a note, if we don’t see any new questions on this post in the next 30 hours, I’ll start the process of moving this proposal on-chain :slight_smile:

2 Likes

I strongly dislike this proposal for several reasons. There are many very sensible things suggested, however, it tries to tackle too many things at once and lacks the specifics for implementation details, resulting, imo, in a complete mess.

  1. I think changing msol unstake pool split to 50/50 is reasonable (although i’m not LP in it). Given that it’s clear, concise and isolated from other issues, i wonder why not put it up for vote in a separate proposal?

  2. Improving data gathering process and making it less dependent on third parties is definitely sensible. It seems like a logical thing to do in any case. Same goes for validators dashboards, tools and so forth. Improving emergency unstake ops also sounds like a good idea. My sinister side might suggest it is put into proposal as “justification” for fee hike, as it requires some work. Regardless of baseless speculations on my part (sorry!), I do not see why it requires dao vote at all – all of this is technicallly feasible and (unless i’m missing something) not particularly hard, just requires time and effort.

  3. Scoring system. Putting up “we are going to improve scoring system” without describing specifics doesn’t make sense to me. I am on board with capping credits due to voting mods (for now), while still punishing for low amount of credits, as it is a very good metric of staking performance. I am not a fan of forcing wider geography distribution, current system is probably fine, there is no good reason of trying to force solana nodes to places with underdeveloped datacenter/bandwidth infrastructure. Overall, given that the scoring system is very important part of the pool, I would like to see a developed specification, then discussion and then a separate proposal for it. (Leaving aside target commission for now).

  4. I kind of feel bad about removing compensation. It was voted on and approved. Cancelling it doesn’t seem right. If it is so necessary to lenghten the runway now, I would prefer to pause it, returning to it in 6–12 months. In any case, I’d prefer it to be a separate proposal.

  5. Targeting 6% commission is too drastic of a change. This is somewhat contrary to the stated marinade mission of supporting wide validator set.

    • The current solana foundation (SF) stake is barely break even at 10% and the stake pool is not large enough to provide enough support for most validators at 6%. It doesn’t make economic sense for foundation nodes to change commission for foundation nodes to 6 %. As a result, 1/2 to 2/3 of the currently staked nodes will lose significant portion of the stake and stake distribution will become very top heavy. would suggest to make the target
    • Such change might require a lot of restaking, which will eat into pool APR.
    • Small foundation validators, who bought/accumulated MNDE for staking gauges have already taking 2 haircuts: mnde price is falling, the amount of locked mnde keeps increasing, reducing their resulting stake from gauges. Now they will be taking a 3rd haircut by either reducing their commission and losing on gauge income or losing significant amount of stake from the stake pool.

I would prefer to target 8%, as it is currently the sweet spot given amount of SF stake and the size of marinade stake pools. You would keep large number of SF nodes, but will still be able to nudge them to lower commission, which is good for the network and the pool long-term. As the pool grows (or, let’s say block rewards increase or small nodes no longer subsidize large nodes with voting fees), the commission can be lowered further, which is probably desirable in the long run.

  1. The necessity of tripling marinade commission. The math/financials on this are completely opaque. I don’t expect people to work for free, but I wonder if such a hike is a good idea. It does not improve competitiveness of the pool, yet validators are expected to take quite a pay cut. If current runway “as-is” is 3–5 months, maybe extending it to 6-12 months while focusing on the pool growth is a much better approach. It’s hard to offer anything specific without knowing financial. I do have to point out that 35% of MNDE supply is allocated to treasury and 30% to the team, which is about 5 times larger than current circulating MNDE supply. That is before the fees. In other words, MNDE team has way more resources at their disposal comparing to small validators who are barely keeping their heads above water as of now.

In summary, while there are some very good ideas, some of the moves are, IMO, ill-advised or maybe worth rethinking. In addition, there are so many issues jumbled into one proposal, it’s hard to have a proper discussion about it. I would suggest splitting this into several proposals and give more discussion/consideration to points 5 and 6 in particular. As it stands, i would definitely encourage people to vote NO on this.

1 Like

Hello @mrknc, and thanks for the constructive feedback. I can try to shine light on some of the points you raise, but I will join you to say that this is a big proposal that contains quite a lot of changes.
The idea behind grouping all those changes together is that they all relate to 2 big topics:

  • Marinade’s financial situation
  • The algorithmic delegation strategy

It would have been a possibility to just put out a proposal about updating the fees, but moving the fees to 6% while conserving mSOL APY can only be done by modifying the commission target for the delegation strategy.

So, given that the first proposal would have had to affect the delegation strategy anyways, it was decided that those 2 changes could be introduced together. Considering we’re already modifying the delegation strategy in this proposal, it also made sense to integrate all the improvements (like data collection or emergency unstake) directly to the same proposal, so we can treat it as one project internally.
The mSOL/SOL unstake pool also being tightly related to the need of raising Marinade’s fees, it was added as a part of this proposal too. I agree that it could have been turned into an individual proposal, but to preserve MNDE holders’ time and keep the focus on one discussion, I included it in this one.

Now let me try to answer some of your questions in the different points you make:

3/ I totally understand your thinking regarding the new scoring system and I will check with @lj1024 if we can try to bring more clarity to it. The issue is that the delegation strategy is complex and our dev team will probably have to iterate on and improve the new code (because those changes are too major to just modify the existing delegation strategy) before putting it live. This means that this proposal is written before we exactly know how the new delegation strategy will work, but can only focus on what it will try to achieve. Obviously, validators will be consulted if this new delegation strategy rolls out and results will be observed by both Marinade and the validators community to check if some other improvements/tweaks are needed after its implementation, and if it serves its initial purpose stated in this proposal.

4/ This is something I considered suggesting, but what we need to take into account here is that the Unstake incident impacted mSOL holders at the time. With time passing by and TVL growing, if we re-activate the compensation plan in 6 months or 12, the people benefitting from it probably won’t be the same as the ones that experienced the incident. I don’t necessarily agree that it should be a separate proposal (even though I see your point), in my opinion a new proposal can “erase” an old proposal if it’s stated in an obvious manner and people voting agree with it.

5/

As a result, 1/2 to 2/3 of the currently staked nodes will lose significant portion of the stake and stake distribution will become very top heavy. would suggest to make the target

I’m not sure why you’re mentionning that this change would make the stake distribution top heavy? Some smaller validators are trying to attract stake with a lower commission, losing money every epoch in the hope of reaching more stake this way, I don’t think that only top validators run below 10%, it’s more distributed than this. You also may have forgot to end the sentence but I think you meant 8%?
In any case, validators will still be able to get stake at 10%, but this proposal slowly invites validators that can do it to lower their commission, and offers a way to get stake pretty quickly by running at a very low commission %, which will be of interest for other validators.

Such change might require a lot of restaking, which will eat into pool APR.

Actually, with the partial unstake added in the last months, those changes would be realized without impacting the APY as it currently is. Marinade is already rebalancing stake with a max cap every epoch.

Small foundation validators, who bought/accumulated MNDE for staking gauges have already taking 2 haircuts: mnde price is falling, the amount of locked mnde keeps increasing, reducing their resulting stake from gauges. Now they will be taking a 3rd haircut by either reducing their commission and losing on gauge income or losing significant amount of stake from the stake pool.

You might have seen the previous version of this proposal that suggested that gauges would be affected by the commission multiplier, but it’s not the case in this proposal. Any SOL obtained through the gauge would not be affected by this proposal, even at 10% commission. Reducing their commission will be a choice that they can make in order to try to attract more Marinade stake, or not.

6/

The necessity of tripling marinade commission.

This is actually a bit more straightforward than what you could expect. Moving the fees to 6% and the commission target to 6% is actually the only way for Marinade to come back to a neutral/positive cashflow without losing on APY and reflecting those changes on mSOL users. Those numbers have been decided mostly for those reasons, and 6% is still in the range of fees already in place in other Solana stake pools. Marinade was actually the lowest fee for a long period, but the current market situation forces to revisit this. It is not excluded that a DAO vote can revisit the fees again once the treasury is in a more comfortable state.

You also mention that Marinade DAO owns 35% of the MNDE supply, which is true. But Marinade DAO is composed of all MNDE holders, not the Marinade team. I’m not sure MNDE holders are very keen on unlocking MNDE from the treasury and dump it into the market, lowering MNDE price, just for Marinade to keep paying salaries. You can actually see the state of the “expenses” treasury in the original post and take a look at the Kitchen stories to see the expenses and revenues and get a better financial overview. If you agree that selling MNDE from the treasury is not an option, you’ll see that raising the fees is the only sensible option for Marinade at this point.

I understand your desire to see this proposal broken into multiple small proposals, but I truly believe that it makes sense to group everything contained in this proposal together. I also fear that multiple small proposals would get less engagement and induce fatigue for MNDE holders that need to vote on all of them in the end.

I want to thank you again for raising your voice and bringing up all those points, I hope my answers helped you see some parts of it differently. Let me know if you still feel some points need to be detailed or discussed.

1 Like

Bullet point 3 is a great point. Data center risk works both ways - you don’t want to have it too concentrated, but at the same time, there are not very many options available to node operators as it is. At least not many that are trustworthy and reliable. I would hope to see it work out where data center risk could be more than offset by high reliability

1 Like

I understand the logic of it, but still think we are better off separating proposals and discussing them properly. Overall market outlook is bleak, and what is even more concerning, TVL is not growing nearly as fast as was hoped for (which is one of the primary reason for financial issues). This is very ambitious and complex project, MNDE holders concern at this point should be saving time, but improving chances of protocol survival and success.

Certainly devs should have the ability to tweak and iterate. However, currently there is literally zero info on the new scoring aside from “it’s going to be better”. The little I know about it (punishing high credits and using skip rate “for information purposes only”) I had to wrangle from lj1024 in discord. Why not develop specifics first before bringing it to discussion? Besides, the current system seems mostly fine, it is certainly better than a non-existent one.

Now, this is a very good point.

Poor edit, sorry. Yes, I meant 8%. The proposal will certainly help low commission validators, but not by much, probably. The economics are just not there right now. Vast majority of validators would not be around if it wasn’t for SF stake due to the high cost of voting fees. Hence most of them are running high commission and its barely sufficient right now. For new validators they have no hope to attract stake above 0-2%, so they have no choice. At 2% you need 250k sol stake to break even on voting fees. 50k stake from marinade will certainly help, but it’s not nearly enough. I certainly have admiration for people who were able to somehow bootstrap themselves, but i’ve been around for 18 months and there are very few success stories. The ecosystem is not mature enough to support hundreds of well-run low commission validators. That’s why I suggested lowering commission gradually as TVL grows.

Actually, with the partial unstake added in the last months, those changes would be realized without impacting the APY as it currently is. Marinade is already rebalancing stake with a max cap every epoch.

One would have to look at the numbers, but I wonder if this statement is true. So correct me if I’m wrong: low cap would imply it would take longer if there is a lot to rebalance, meaning bigger hit to APR due to increased marinade fee while system reaches target stake distribution. High cap would mean more rebalancing, triggering APR hit due to unstaking/restaking.

You might have seen the previous version of this proposal that suggested that gauges would be affected by the commission multiplier, but it’s not the case in this proposal. Any SOL obtained through the gauge would not be affected by this proposal, even at 10% commission. Reducing their commission will be a choice that they can make in order to try to attract more Marinade stake, or not.

I think you missed my point. I agree with gauges not being affected by commission cut for now, but they will take a hit either on gauge income or on marinade stake regardless of what they do.

Moving the fees to 6% and the commission target to 6% is actually the only way for Marinade to come back to a neutral/positive cashflow without losing on APY and reflecting those changes on mSOL users.

You also mention that Marinade DAO owns 35% of the MNDE supply, which is true. But Marinade DAO is composed of all MNDE holders, not the Marinade team. I’m not sure MNDE holders are very keen on unlocking MNDE from the treasury and dump it into the market, lowering MNDE price, just for Marinade to keep paying salaries.

This might sound crazy, but I question the notion of necessarily having a positive cash flow right now, while validators are slowly bleeding voting fees and, to a smaller degree, rising hardware costs. It is contrary to the declared mission. Also, I wasn’t suggesting dumping more MNDE into the market, lord knows there is already a lot of that and not enough liquidity. I definitely agree that 3–5 months of runway is too little. However, I would prefer extending runway to about a year without gutting large proportion of validators instead of targeting positive cashflow.

  • removing compensation program should bring about 700 - 900 sol (30–35k) a month according to the OP (although June blog post states 1246 mSOL income in June.
  • bumping management fees to 4% would double the above
  • changing fee split in unstake pool would bring additional income, although i’m unsure how much, wish you would provide some numbers
  • about 800k sitting in the treasury right now , 2/3 of which is not generating yield for some reason. Also June blog post quoted above mentioned 300k in treasure, so situation has somehow improved?
  • MNDE could be considered to generate income without dumping it on the market through gauges either by delegating it to your own validator or increasing incentives to LPs in pools in which treasury is participating in.

With all of the above Marinade can reduce the burn rate significantly and provide itself plenty of runway, while pursuing more sensible commission adjustment and keeping healthy amount of validators, which is in line with declared Marinade mission. At least that’s how I would do it.

P.S.
The real fatigue comes from considering proposals, not from voting. You think 5 page proposal doesn’t produce fatigue? =)

1 Like

Low credits don’t have anything to do with “staking performance.” Credits (vote credits) are a metric of voting performance, but in some cases validators have made modifications to exploit this beyond the possible number of credits validators can receive without making custom modifications to the software provided by Solana Labs.

There are obvious problems with the system, which is why there is interest in improving it.

I don’t know what you mean by a non-existent one.

You’re not a fan of decentralization?

2 Likes

I believe Cerba has already answered this point, but to make sure it doesn’t get lost:

The various fee changes are meant to act in concert in order to stabilize the treasury. Merely splitting this out to a proposal wouldn’t move the needle, so it is bundled with other changes that would be required to do this without impacting APY for mSOL stakers. It is the same reason why separating the compensation plan changes to a standalone proposal wouldn’t serve the same purpose.

Arguably, this could have been two proposals (a point I made myself when reviewing a draft):

  1. A proposal with the changes required to stabilize Marinade’s treasury given market conditions; and
  2. A proposal including the other strategy changes.

However, given that there are changes to the strategy included on #1, and the parties most affected by both proposals would be the same (validators), the decision was to bundle them.

I hear you and it is a valid point, which is why this is a governance proposal and not an internal change.

As a counterpoint: keep in mind that the important thing was the goal of temporarily bumping APY for mSOL holders, not merely drain the treasury of fees.

While this changes the mechanics of how that is happening, mSOL APY is expected to remain as if the compensation plan was still in place, thus not impacting mSOL holders negatively. In fact, this change would be more sustainable and long-term than a plan that’s meant to end in a few weeks.

To be clear: while I’d love if this were to take Marinade to cashflow-positive, that is neither the goal nor what it is calculated to do. The key aim for these changes is to stop the bleed, since as Cerba mentioned on his original post…

Yes, our top two goals are decentralizing both Marinade and Solana, but those become much harder to reach when you can’t pay people.

1 Like

I agree with you on this, and this is what this proposal was made for: try to guarantee the survival of the protocol.
PlayerOfBits also explained some reasons behind bundling those changes, and I think you understand why we did it, and I fully understand why you don’t like it. If you aren’t happy with everything that this proposal encompasses, you can invite people to vote “No” and/or make separate proposals. You can even make a proposal contesting one of the points of this proposal after it has passed in order to revisit it, if it passes.

However, currently there is literally zero info on the new scoring aside from “it’s going to be better”. The little I know about it (punishing high credits and using skip rate “for information purposes only”) I had to wrangle from lj1024 in discord.

The proposal actually explains that credits will move to a secondary place in the delegation strategy, and that the new important metrics will be data center location, commission and stability over epochs. While we can’t yet provide the exact parameters of the strategy, I think this is enough to understand the purpose of the change, wouldn’t you say?
As @pumpkinspool stated, the current situation is not “mostly fine” as modded validators are skewing the average and it’s affecting every validator not running those mods. The current delegation strategy does work, but those improvements are needed to make the delegation strategy more transparent, easy to understand and read, and fairer to everyone.

What I can add is that Marinade will probably gather feedback on the new delegation strategy and tweak it before we arrive to a final version. You’ll probably get a way better understanding of the new delegation strategy as it’s finalized and details can be discussed directly on Discord, between validators and the dev team to tweak the last things.

At 2% you need 250k sol stake to break even on voting fees. 50k stake from marinade will certainly help, but it’s not nearly enough. I certainly have admiration for people who were able to somehow bootstrap themselves, but i’ve been around for 18 months and there are very few success stories. The ecosystem is not mature enough to support hundreds of well-run low commission validators. That’s why I suggested lowering commission gradually as TVL grows.

I agree with you on this point. I would love for Marinade to have enough stake to fully bootstrap a large number of validators, but this can only be achieved in combination with our other efforts to bring more TVL. The current situation forces Marinade to redirect stake towards lower-commission validators, that’s unfortunately the only option we have. As TVL grows according to our plan, we can revisit the target commission and adapt the strategy to the amount of stake we have, and to the number of validators that Marinade should reach and support.
Lowering commission gradually as TVL grows would unfortunately not bring enough revenues in the next months to keep building, but it’s something that can be revisited once Marinade has achieved its objective of securing a strong treasury in USDC/mSOL to pay for expenses.

One would have to look at the numbers, but I wonder if this statement is true. So correct me if I’m wrong: low cap would imply it would take longer if there is a lot to rebalance, meaning bigger hit to APR due to increased marinade fee while system reaches target stake distribution. High cap would mean more rebalancing, triggering APR hit due to unstaking/restaking.

What I meant by this is that the stake is already being rebalanced with a max cap of 4% per epoch. This has been the case for the last months, so it’s already reflected in the 30-day APY of mSOL for example.
When applying the new delegation strategy, we’d use the same cap and maybe it’d take 10 or 20 epochs to fully rebalance everything, but the impact on APY would not be very far from the one we currently have when rebalancing. If we do raise this cap, then yes, this could lower a bit the APY for a small period, but since stake would reach lower-commission validators faster, the APY would be compensated in the next months too.
If you’re really interested in this question, maybe you can try to model the impact and check if the APY hit would be bigger than what we expect? I’ll also check with @lj1024 if we have any model already that would simulate the impact on APY.

I think you missed my point. I agree with gauges not being affected by commission cut for now, but they will take a hit either on gauge income or on marinade stake regardless of what they do.

I see what you mean and sorry for missing your point. I don’t really have an answer to bring to you regarding this, unfortunately validators running at 10% will be affected by this proposal. The gauges will allow to “lower” the impact for validators at 10% that have invested in MNDE, but in order to keep the same stake, they’ll probably have to acquire more MNDE indeed. The only thing I can say is that the drop won’t be “brutal” and they will have time to plan around and make a decision based on what is happening and how things are evolving.

This might sound crazy, but I question the notion of necessarily having a positive cash flow right now, while validators are slowly bleeding voting fees and, to a smaller degree, rising hardware costs. It is contrary to the declared mission.

The mission stated in the article you quote is “Now Marinade’s goal is to make Solana performant and decentralized.” By directing more stake towards stable validators over multiple epochs, Marinade is actually pursuing this mission. We also make the case that there is currently not enough SOL staked (overall) to support 7000 or 10000 validators, it’s simply not possible. The goal of those changes are to allow Marinade to gain more money out of its stake, and consolidate the stake to validators being performant and stable, with a bonus if they run at a lower commission. I wouldn’t be surprised if a validator in a low-populated data center being stable over 50 epochs get stake from Marinade, even at 10% commission.

The thing that you mention and that I understand is that some validators rely on Marinade to become profitable. While I’m very happy that it’s the case for some of them, it’s unfortunately not a “guarantee” that Marinade can uphold to, no matter what. The current situation pushes Marinade to a corner, and we first need to build a solid treasury in order to keep pushing the TVL higher. If Marinade itself is not profitable, validators relying on Marinade won’t stay profitable either for long.

Having a cash-flow positive balance is actually extremely important, as a “12-months runway” can quickly become a 3-months runway if TVL were to disappear, or if SOL price was to drop heavily (impacting all future revenues).

removing compensation program should bring about 700 - 900 sol (30–35k) a month according to the OP (although June blog post states 1246 mSOL income in June.
bumping management fees to 4% would double the above
changing fee split in unstake pool would bring additional income, although i’m unsure how much, wish you would provide some numbers
about 800k sitting in the treasury right now , 2/3 of which is not generating yield for some reason. Also June blog post quoted above mentioned 300k in treasure, so situation has somehow improved?
MNDE could be considered to generate income without dumping it on the market through gauges either by delegating it to your own validator or increasing incentives to LPs in pools in which treasury is participating in.

With 2% fees, the estimate is 0.01% of the TVL per month as revenue. For 7M SOL, keeping the fees to 2% would represent 700 SOL per month, or $24k5 per month (with SOL at $35). 4% would be $49k, and 6% will lead to $73k5. With expenses around $110k per month, you can see that 4% would not cut it.

Fees from the unstake pool can be calculated with the stats available here. We’re seeing between 200 and 500 mSOL in the last months depending on the unstake activity, so it can be expected that moving the fees to 50/50 will bring between 400 and 900 mSOL (or between $16k and $36k at current prices). We also need to take into consideration that the bear market has leaded to quite a lot of unstakes, and those profit cannot be considered guaranteed at all.

Regarding using the 800k in the treasury to earn yield, managing a treasury is actually quite a responsibility and requires time and attention. I’m not aware of the plans regarding those funds, but would be uncomfortable having all this money in Solana DeFi. This treasury is actually the “expense” account that received the funds from the TEP, and I imagine that a part of these funds will be moved to the DAO treasury as we steer the ship back to profitability. (300k to 800k is the result of the TEP). They are probably kept here as we might need to use them over the next months to pay for salaries, unless this proposal passes.

Regarding your last point, Marinade does not run any validators, and it would be quite unfair to use the MNDE in treasury, not even distributed yet, to skew the gauges to our advantage. All MNDE holders using them and leveraging them would probably not like this move too.

As you can see, with the changes on the delegation strategy and the unstake pool, Marinade should reach between $90k and $110k monthly revenues. This is just enough to guarantee a 6-8 months runway (with the current treasury from the TEP) while keeping operations going, but not really more than what’s needed.

Definitely agree with your point on fatigue, this proposal shows that the biggest the proposal is, the more complex it is for MNDE holders to find time and energy to participate. I’m definitely taking this lesson with me, but as explained in all the last messages, those changes made sense as a block and had a same reasoning, so I deemed worth to consolidate everything into one big proposal.

Thanks for the detailed answer, I will close this message by saying that you’ll still be free to vote no and invite people to do the same, or to vote yes and bring up a new proposal suggesting some addendum to certain parts of this proposal even if it passes.

1 Like

Hello folks,

How are we doing? Any major outstanding questions?

I wanted to make sure we weren’t missing anything before moving this forward, given the length of the exchanges.

1 Like

Let me bring (hopefully) a little bit more clarity to what we intend to do.

First, re-iteration of the planned steps:

  1. Fees, commission
  2. Proper and detailed data gathering
  3. New scoring model

I think the first point has already been covered very thoroughly, so allow me to skip that one and let’s go to the second point.

I love simple, robust things that I can rely on. That has been the case for as long as I can remember. I feel delegation strategy is neither of those things. When I joined Marinade and saw this beast spreading across 4 repositories, being run manually each epoch, I happily offered to start improving it. I understand that in the first phases of any startup-like project, the focus needs to be on speed of development, but as the project grows, I believe the foundations should be revisited and stabilized.

So far, we have done a good job doing that: the code base has been reduced, process automated and overall simplified, new features were added with ease on the development side.
Yet, there is still plenty of legacy relicts.

And the current data structure? It is difficult even for me to see which field is moving average, which is average of averages, which data represents the end result of the epoch, what exactly is erased from the first scoring run by the second etc.); there is too few datapoints in time (effectively once per epoch), the storage is iffy, the data gathering process is tightly coupled with scoring/emergency unstaking; the data sources we use could be improved, … When I want to check some validators, the easiest way is to pull from githug and run my saved queries in my favourite SQL client. Now imagine making this data accessible to e.g. Decentraizer… And this all is going to take some time (= resources) to fix, so of course it is necessary to get support for this idea through the proposal. Maybe MNDE holders don’t feel, so strongly as I do about these issues.

Intermezzo, dashboards: The visibility of what is about to happen is very low (read 0). I want validators to see in real time who is about to receive stake and who is going to get partially unstaked;

Now to the the last part. Indeed, there is very little info on how exactly the new scoring is going to work and I do not want to give a hard promise on the exact methodology before we are actually able to gather all the data that we need to even test the ideas we have.

When the new scoring system is implemented, it is going to run side by side with the old one, so no worries, we do not plan to surprise anyone with some unknown enigma that would eat all of your stake. We will still be in the phase of development and will be able to receive the feedback while still in testing/development phase.

Reminder: 1) there is very little variance in credits (the values for performing validators are all very close to the average); 2) to make network stable and decentralized it needs different things: performance, unique geo-locations, timely upgrades, long term stability, new validators - (all of them of course still only under condition of reasonable performance over time - we do not care that you are top 50 in credit if you are down for 2 epochs). We still want to help bootstrap new validators with small stake but at the same time reward good perfomers (we need to remain competitive within the staking pool space). All that could be covered by selecting our favourite 400 - 500 validators (the count may change based on TVL, obviously) using some of the variants of Pareto Efficiency - and we intend to do that.

Again, to re-iterate, we will run this in a testing mode first, of course. Maybe Pareto will not work out well (very difficult to say now) and we will instead adapt current scoring system on the new data (with better capability to see “red flags” and with better capability to see some new angles to who is validating well). But without this proposal, I will not be able to explore these new possibilities/approaches. I hope this makes sense to everyone that needed further info/answers.

I intended to keep it short. Oh well…

2 Likes

In the absence of further questions, I’ve moved the proposal on-chain!
Let’s get your Chefs NFTs to the voting booth :slight_smile:

1 Like

Solend tentatively voted yes but I wanted to personally voice 2 concerns

  1. (minor) moving up to 6% is fine, but i think should still respect the payback by keeping at 4% for 30 more days then bump to 6. kind of semantic but would feel like being more true to the intent of the original plan
  2. (moderate) I worry if the changes to the unstake pool will significantly lower liquidity. Liquidity is important for maintaining peg. I hope this could be monitored after the change and reverted if needed.
1 Like

Hey @nope, thanks for the comments! Let us crunch the numbers on the first one and address it separately, but I wanted to clarify this first:

By maintaining the peg do you mean the rate people are likely to get if they want to instant unstake? I don’t believe there’s a reason why liquidity on the pool should affect the mSOL-SOL relationship, which comes from accumulating the staking rewards.

I think I see where the issue might be coming from: I’ve just realized that @Cerba referred to it only as the unstake pool on the proposal, but it is about instant unstaking.

There are no fees to merely unstake your SOL, nor any specific pool for it - Marinade would just unstake it at the end of the epoch.

1 Like

Hello nope, thanks for the late comment!

To answer to your first point, we ran some calculations and we estimate that roughly 50% of the planned compensation plan has been executed already. This is in huge part due to the recent lenght of epochs on Solana that has increased quite a lot, as you can see below (x being the epoch and y being the number of hours they lasted for). Longer epochs lead to a lower APY for staking rewards, which affected the estimation we were able to make at that time.

In this context, it would take more than 40 days to fully complete the compensation plan, whereas the vote about the compensation plan had already put a maximum end date to September, 2. Assuming the change goes live by August 10, this means we will only be removing 20 days.

Given the importance of stabilizing Marinade’s finances and the support that this current proposal has received on-chain (and that states that it would overwrite the compensation plan), we feel that we should carry on with the proposal as it is and not try to modify it in the last hours of its existence.
If you strongly believe that Marinade still has a “debt” to its users, I think this topic is worth rediscussing, maybe first on our Discord and if it’s an opinion shared by many, on the forum once again.

1 Like