The initial idea would be to allocate 10% of Marinade total stake
I think that eventually this approach could replace the algo formula or move it to a suggesting or auto-balancing position somehow. But I would proceed slowly and with caution, not to lose sight of the decentralizing effort.
One idea could be linking the directly governance-allocated stake to MNDE float when it gets over 10% to ensure a steady increase.
… aaaaand I’m clearly not operating at 100% myself yet - not sure how I missed it. Thanks @gekonn!
I’d be hesitant to fully replace the algo formula, since it’s unbiased, but it’s something we can figure out once we see how well the approach works and the impact it has.
I believe there needs to be a cutoff. Once validator no longer needs to reach outside of marinade the incentivization would be detrimental to the newest validators joining the pool.
There is an obvious need to incentivize smaller validators, but I see a situation where there is a nakamoto coefficient within the marinade pool of validators.
NOTE: “everyone will have the same incentives to accumulate (or not) MNDE and will have time to do so” – is not true. Those with more liquidity have much more time/power to increase their holdings. This is a separate topic, but I am requesting the language we use to stop describing linear earnings as equal to all; we do not have the same starting positions.
The ability to command stake at scale will be extremely valuable in a few years from now, especially wrt. to MEV. Huge fan of this proposal and have been lobbying for it for a while now.
I’d like to address Jesse’s comment as well: if you are a holder of MNDE, even if just a small amount like I hold for instance, you will be able to come together with other community members to directly command stake. The average DAO community of a few dozen people that are really engaged will be able to collaborate and delegate the stake to their communities’ validator, so by no means required for anyone to be a whale and pick up the bill all on their own.
Incredible proposal!
However; can you help me understand this:
But how does this guaranteed stake increase the number of validators on Solana? In the problem statement we call out the primary limitation as cost, but the validator would still need to purchase MNDE to get their portion of stake.
The assumption here is that purchasing x% of MNDE and locking to acquire y% of stake is a cheaper + better alternative than existing options. Is that correct?
Interesting proposal, a lot to unpack.
First of all, I think it’s ok to acknowledge that in part this proposal aims to increase the utility and thus value of MNDE, that’s perfectly fine.
I disagree with the sentiment that this benefits small validators who are struggling to make a profit, however. Validators who are struggling to make a profit are likely at the limit of their cash flow (or they’d have purchased more SOL for self stake etc) and don’t have large excess liquidity to pump into MNDE. In fact large validators who have been profitable for some time are more likely to be sitting on reserves enabling them to benefit from this.
Despite this I think it’s an interesting and worthwhile proposal, but I don’t think all Marinade stake should be governed this way. This is essentially a distribution based on economic strength and wealth of a validator, while the current delegation formula is merit-based. A healthy mix of those two will be a great result. I think anything higher than 20-30% of total stake being controlled by MNDE voting could lead to negative results and an erosion of Marinade’s brand of championing decentralization and small validators, where it becomes simply a pool for the rich.
After reading replies here and in the validators channel on Marinade Discord, let me bring some clarity on some points:
An additional criteria may need to be added to guarantee a minimal performance for the validator in terms of APY, to avoid directing stake to validators that have very low credits. If you have a clear idea on how this criteria could be implemented, please express it here.
But this proposal is actually a way for validators that do not fall into the “top 300” in terms of score to start getting stake from Marinade and build their way towards more stake in the end.
Raising this 10% of total stake number and progressively lowering the % controlled by the algo formula would only be done via on-chain votes and based on the results of this first implementation, so we can all take into account all the possible implications before acting.
The assumption is that via this system, it would be more profitable for validators to buy and lock MNDE and direct stake to their validator than to buy SOL and stake it to their validator, yes. While also having the additional benefit of spreading MNDE to validators, that would own and govern Marinade, its delegation strategy, its treasury, etc. making the Marinade DAO more decentralized in the end.
I admit that it doesn’t address directly (and not as much as some may have hoped) the profitability of small validators, but it opens a way to convince MNDE holders (or to become one) and guarantee stake your way, resulting in a more sticky stake and a new access to profitability (in addition to the ones already at their disposition that I list in the problem statement).
The problem of “becoming profitable” is, to my opinion, linked to the low number of options available to attract stake, and this proposal aims at offering another and a new way of doing so.
Thank you for all those comments, you’re all helping moving this proposal forward and fixing its flaws and I thank you all for this 
Edit:
Out of curiosity, I ran a test with random numbers:
If we imagine a validator that has 500 SOL self-staked, which represents its initial investment in its validator, and that has no delegated stake, we can observe those results:
As a comparison, selling those 500 SOL for approximately 555 000 MNDE would allow to control 33,333 SOL to delegate (if we take the numbers from my proposal). This investment would have way better returns:
And this doesn’t take into account the other advantages that having MNDE can carry, today or in the future.
(TL,DR: there’s no fair, there’s only fairer)
Let’s also open the fairness angle early: is this a plain rich getting more influential? is this moving away from merit only? In my opinion, it only looks like that if you ignore the context. And the context is:
Existing state: competing on validator score among the top 300 is already a very tight game. The difference can often come to randomness, not an action to be done to be a validator. This is an opportunity to improve - randomness is not fair and it’s not merit, it’s just random.
Being early: buying 5000 Solana to self stake for $2 back then or for $100 now requires a different level of conviction. MNDE could be a new round of influence being distributed, a new chance for people. Current liquidity miners, retro-drop recipients, contributors, GM praising discord members, AMA questioning marinators are already frontrunning the big guys. Big money is not the only way, leading to A+B:
Curation: The proposal needs to start with a simple system - there’s a need to build fast, small and stable first to verify the approach …, but you can already see a few years into the future if Solana and Marinade are successful in decentralizing: there will be thousands of validators and this gauges system will have a discovery problem. How will the DAO solve it I am curious; I see hope in curating preselected delegation bundles: delegating to a (?DAO curated?) bundle of validators situated in developing countries, where the additional income be spread around, or bundle validators run by individuals or organizations with altruistic goals…
Lastly: Validator game isn’t fair already. Pathos warning: Someone was born rich and could easily get the top hardware, someone was able to get a privileged spot in a data center, someone was just super early to Solana or knows the right people… That’s life - obviously smart and rich will have all sorts of advantages. The goal is not to strip away that advantage here by force, but instead enable anyone to involve themselves in a conversation and have an opportunity to demonstrate the merit.
So, in the perspective of fairness we should be asking the questions:
As long as we have Yes and Yes, we should try, learn and improve.
Agreed. I do have a general concern over what can be perceived as fairness, but it’s not the “rich gets richer” issue: it’s a factor of fundamental differentiation.
If I only want to make money, I’ll take my SOL to the highest bidder. The reason I have mSOL is the promise to make Solana more decentralized through an unbiased, algorithmic distribution.
I’d be very careful with the perception around the stake percentage that gets controlled by voting, merely because if at any point it creates the impression that Marinade is no longer impartial, it will lose value in people’s minds.
Yes, we can explain why it hasn’t happened, provide evidence in terms of costs and effects, but if you have to explain yourself then chances are you have already lost.
Indeed. But if this is anyone’s concern, the flip side is that someone that well-funded and well-connected likely doesn’t need to bother with suborning or subverting Marinade. They can probably find other easier ways to grow their stash. 
Happy to give this a shot with a non-controlling percentage, like the suggested 10%, then slowly take it from there.
Well done, Chefs! I’m a big fan of pretty much everything I’ve read so far from the team on Marinade’s governance construct and the thought process behind it. And let’s be honest here - the conversations we are having on fairness and providing support for newer and/or smaller validators don’t really happen much in other forums/discord chats. If we as a community can stay anchored around these values, I think it will go a long way.
Totally agree with @gekonn’s sentiment on fairness as well as the context on being early here:
I would even suggest that the best possible way to promote fairness today is to reach out to some of these smaller validators and make sure they are aware of Marinade’s mission, values, and the opportunity to provide sticky support for their operations. At current prices, you can participate in protocol governance for ~ $100 USD (and potentially compound voting power as suggested in “How Marinade’s MNDE will fuel on-chain DAO governance”)
Let’s also not lose sight of the cap that the Marinade team has established - the largest amount of the overall Marinade pool that can be allocated to a single validator is 1.5%. More info here.
I wanted to pose a few additional questions:
Are there any plans to formalize a process to get closer to full decentralization through a gauge voting system (ie steps to get to 50% delegation controlled by the algo formula / 50% governed by DAO or 25/75 or whatever that maximum number might be)?
I understand the desire to start smaller with 10% controlled by DAO voting and see how things go through a trial period, but have you considered starting at 15% or even 20%?
IMO, more power in the hands of the community from the very beginning could potentially:
Help accelerate Marinade’s TVL growth and make progress toward the goal of 30 million SOL staked in 2022 (currently 6.6 million SOL), substantially improving Solana network decentralization and providing massive support for smaller validators.
Maximize the impact of that sticky stake and predictability for smaller validators right out of the starting gate (assuming total MNDE locked in governance increases over time)
Further increase the incentive to buy & hold MNDE in the near term
Look forward to continuing this discussion!
This is a great proposal! I really appreciate all consideration that went in to the proposal along with the great points made in the discussion.
I think you could build on @dobby’s astute point of a 1.5% cap to a single validator and have a cap for “governed stake” as a control mechanism that could be adjusted to make sure that there’s more opportunity for new and small validators to become profitable
Hello Dobby, thanks for the very interesting feedback 
Let me try to answer some of those questions!
Currently, this proposal only suggests a starting point at 10% of the total stake and does not formalize any evolution of this system, but this is something that can be reflected on here, together. I think that implementing the future evolution of this system right from the start and in the initial proposal could lead to a more difficult discussion in the end, as we’d debate both the implementation and how it should evolve. I’d rather have a first vote about the implementation, and a second vote based on a suggested plan, based off data we’ll have at our disposal!
I’d be interested to see what other DAO members and validators think of a higher starting point. I wouldn’t be opposed to raising it myself, as I agree with your points, but the members and users most concerned by this proposal should also express on how they feel about this.
For these specific gauges, it would be a “Set & Forget” type of thing indeed, with the possibility to reallocate your MNDE whenever you desire to do so.
The point that both you and Ben raise on a max cap is also interesting. Should the stake controlled by MNDE be considered as the “same stake” as the stake delegated by Marinade’s delegation strategy, or should it be treated in a totally independent way?
For example, should a validator that currently has reached the 1.5% cap on the stake he gets from the delegation strategy be able to get more stake through the gauges? Or should it be forbidden by the system?
In my opinion, both should be considered individually, and another max cap should also be specifically implemented on the stake controlled through gauges too, but I’d be curious to hear your thoughts on this.
Thank you for raising this point.
I like this approach! Very reasonable and measured. No need to try to lock something into place now that might prove to be sub-optimal 3 months from now. Totally with you.
Also with you on this point. I think it’s probably best that any cap placed on the portion controlled by MNDE governance be considered independently from that of the algo formula, else the whole gauge voting system becomes less desirable. I would lean toward giving the community more power and more reason to participate and to care about governance rather than less. For that reason, I also think that any new cap introduced should be higher than 1.5% (if capped at all).
Thanks for taking the time to answer some questions and explain your current thinking. This is fun!
I think for a change like this we should start small. There are two things at play for me:
I’m comfortable with 10%, but around the 15%-20% range it starts feeling less like something we’re experimenting with and more like a change that will be harder to roll back if we realize it had some side effects we did not like.
(Likely because I currently wouldn’t want more than half the stake controlled in this way, and 20% it would be well into that percentage)
Hello, since Marinade’s Team will have 30% of MNDE tokens, and part of the team is running Solana Validator nodes, is this really a good strategy?
Valid point, if I understand you correctly(?) and important to discuss.
Do you mean that it is problematic if someone in the core team also runs a validator, or anyone contributing to the Marinade DAO?
Would love to see if you could elaborate a bit more on it. Thanks!
EDIT: The core team does not get 30% - this is for the contributors of the MarinadeDAO, just so ppl don’t think that 30% is spread to like 8 ppl which is not true. However, @meyerbro 's point is still very valid and have to be addressed imo.
I think the tricky part is the perceived self enrichment/conflict of interest where you have (as Marinade) received 6.5m SOL in stake on the premise that this is fairly and algorithmically distributed amongst as many validators as possible to aid decentralization, then create a token, give a large chunk to yourselves then let that token decide the distribution of that stake, including to validators you yourselves operate
(I’m phrasing what I believe meyerbro’s point is, and what I think the discussion of this point would center around, I don’t mean to imply any malice or ill intent on behalf of the marinade team)
Very valid point and I do agree with this point of view, that it is indeed problematic and perhaps even goes against the Marinade ethos. I would for sure support something that would deal with this possible issue.
I’m thinking what kinds of safe guards one can apply to mitigate a possible issue related to this. I do not have any clear ideas for the moment myself, so I’ll wait for someone with possibly a higher blood sugar level to share their thoughts on it too 
Tokenomics
MNDE is currently allocated as such:
35% - DAO Distribution . Used as token holders see fit via proposals, initially used for Liquidity Miningprograms.
35% - DAO Treasury . This is a treasury reserve to be used for operations, grant programs, and strategic partnerships. This treasury is controlled by Marinade’s multisig and will be unlocked and governed by MNDE holders.
Besides this reserve, the treasury also receives mSOL protocol fees:
2% of staking reward fees generated by Marinade protocol
25% of liquid unstake fees generated by the Unstake liquidity pool
30% - Team . This is an allocation for current and future contributors. The distribution has not started yet and those funds are currently controlled by Marinade’s multisig.
Thanks for raising the point, @meyerbro, it’s one I hadn’t considered (I hadn’t realized team members were validators).
As you can see in my posts, I’m aligned with you and @laine about Marinade keeping most (or potentially all) the stake distributed algorithmically. I do think there are safeguards in place which make @Cerba’s proposal rather low risk:
Would love to read your thoughts on those.
[1] eg. what happened in OpenSea, where Nate Chastain abused his position to engage in petty corruption, but his actions were against the team’s interests so he was expelled.
