[Informal poll] Compensating recognized delegates, parameters

At the launch of delegation, compensation was discussed but never formalized. A set of informal polls confirmed that there is broad interest in exploring the next step. Specifically, the question “Should delegates be compensated for this time?” received 77% support from 31 votes. However, exactly how compensation works is an important question!

I offered the only detailed compensation proposal, so far. The formulae are applied per-blockchain block. There are three main inputs: (a) the MKR weight of a delegate (cells B8:B13), (b) the total DAI distributed total across all delegates per month (cell B3), and (c) the target minimum number of delegates to receive compensation (cell B5).

The way it works is that we take the square root of the MKR weight (see quadratic funding) and then we use these numbers to determine the compensation proportion. Compensation could stream to delegates every block or we could approximate the same using a time-weighted MKR weight. We also set a maximum proportion determined by 1/B5 (for example, if B5 is 3 then the maximum proportion is 1/3). There is an iterative process (columns E-H) to proportionally redistribute any proportion that is larger than 1/B5 to other delegates. Once we have our proportions (column H) then we allocate the total monthly proportion (B3) accordingly.

Note that compensation would not be extended to shadow delegates.

Is this compensation system gamable? Yes, but the potential gain is small compared to the capital required. GovAlpha can work to formalize and reinforce the requirements to become a recognized delegate (maybe a minimum time required as a delegate with at least X% communication score). If sybil attacks are suspected then an upgrade of the delegation system could require that delegates submit a MKR bond (maybe 1-10 MKR?) to become recognized. With a MKR bond, delegates could be punished by an executive vote to burn the bonded MKR.

In terms of implementation, this compensation algorithm could be done entirely off-chain in a spreadsheet and DAI sent once a month, similar to how SourceCred works. DAI could be sent by the GovAlpha CU. No additional software development is needed unless we decide to formalize this payment algorithm as an on-chain contract.

How should delegates be compensated?

0 voters

If there is too little total MKR delegated, summing across all delegates, then the system doesn’t have enough information to produce sensible outcomes.

How much total MKR should be delegated before starting compensation?
  • 1000 MKR
  • 10,000 MKR
  • Something else (post your specific proposal below in the comments)
  • Abstain

0 voters

We should cultivate a top tier of XX well-compensated delegates ordered by largest MKR voting weight (see cell B5 in the compensation proposal spreadsheet)
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • Abstain
  • Something else (post your specific proposal below in the comments)

0 voters

The range of options below were inspired by LongForWisdom and Planet_x. Please consider these amounts as the total per-delegate compensation of this proposal plus SourceCred payouts.

How much compensation should a top-tier delegate receive per month?
  • 1k (12k per year)
  • 2k (24k per year)
  • 3k (36k per year)
  • 4k (48k per year)
  • 5k (60k per year)
  • 6k (72k per year)
  • 7k (84k per year)
  • 8k (96k per year)
  • 9k (108k per year)
  • 10k (120k per year)
  • Abstain
  • Something else (post your specific proposal below in the comments)

0 voters

Next step

This poll will be open for two weeks and close on Sep 20.

If none of the polls obtain a plurality of something else then the amount of DAI per month (cell B3) will be calculated as (delegate compensation per-month - approximate SourceCred earnings) * (number of delegates). If none of the polls result in something else then the complete proposal will proceed to a formal Signal Request. Further protections against sybil attacks can be added incrementally.

7 Likes

@Joshua_Pritikin - I feel your proposal is slightly too complicated. I would prefer a flat compensation if the amount of MKR delegated is above a certain level.

I don’t think a signal request is the appropriate mechanism for a decision of this level of importance. There should be a MIP that defines clearly how this works in practice.

I think these polls are a good start, and should inform the direction and contents of that MIP dependent on their outcome, but yeah signal requests are not great when you want to define things very clearly, durably, and in a way which that can be easily referenced in the future.

What we could do is something we’ve done in the past, and use a signal request to define a time-limited trial of some solution, and then pass a MIP later on. This tends to be worthwhile because it allows us to test a solution before codifying it explicitly. In this case though, I would still want a draft MIP that clearly defined processes and amounts before moving to a signal request.

TLDR, if there is appetite to pursue this immediately (which is still not something I recommend, given the recency of the delegation system.) Then I would like to see this process:

  • Polls like this one help to understand community sentiment.
  • A MIP is drafted (but not submitted) which details a system for compensating delegates.
  • A Signal Request and subsequent on-chain poll confirms a trial of delegate compensation according to MIP
  • We run a trial of the system for 3 months.
  • MIP is adjusted if necessary, given feedback from delegates, the community and MKR Holders.
  • MIP is submitted and hopefully passes.

On an implementation standpoint, GovAlpha has not budgeted for this expense. If we wanted to do delegate compensation in DAI, GovAlpha can do the calculations and then we can include the transfers as part of an executive vote once per month (assuming a signal request and on-chain poll confirms the desire for a trial period.)

Taking a brief look at the spreadsheet, I’m pretty sure you can calculate the capped-amounts correctly without the whole multi-stage approach. I’ll play around with it a bit later, and see if I can construct a more elegant version.


I do like the quadratic distribution approach generally, but I do wonder if it’s worthwhile to distribute this DAI value in MKR or vested MKR rather than DAI in order to promote better incentive alignment over longer periods. Distributing 3 month cliff-vested MKR each month would lead to rolling compensation unlocks that both helps capture the price-change of MKR and gives MKR Holders more of an assurance with respect to the alignment of the delegates.

9 Likes

Zooming-out and thinking about the frameworks of such MIP–when I think about a congressional team that helps/supports their Politician (Senate/Parliament) vote, legislate, debate, etc., I was wondering if you believe the current role of a Delegate is defined as a wide-spectrum of duties pertaining, but not limited to:

  • Due-diligence
  • Reporting ( provide ongoing transparency)
  • At-times, R&D
  • Monitoring the entire DeFi ecosystem
  • Evaluating
  • Providing guidance
  • Monitoring Community activity
  • Developing relationships
  • Arguably a MakerDAO Spokesperson
  • Other duties**

When you compare the support–an individual Senator gets from his/her congressional staff–its composed of a large operation of many individuals:

  • Caseworkers performing research
  • Legislative Correspondent (LC)
  • Legislative Director (LD)
  • Chief of Staff (COS)
  • District/State Director
  • Communications Director (CD)
  • Office Manager
  • Counsel
  • Professional Staff
  • Associate Staff
  • Economists or other specialists
  • And Many other Teams, including clerical staff that I won’t list here.

So, I guess my question to you LFW, should a possible MIP include a support/operations staff for Delegates–in simple words–should your idea of a MIP have a framework pertaining to a Core Unit? A CU that provides operational support for Delegates?

And if this MIP becomes a proposed CU, should Delegates be separated from such–and be stipend by the CU?

(I can imagine a future where there will be other CUs, that will also provide operational support/opinions to swing Delegates in a different direction)

Also–in your opinion, should the role of a “Recognized Delegate” stay with the Polychain, Gaunlet, Pantera, Dharma, University Institutions (MIT, Stanford), of the world? Most (I think) are willing to do it for no-pay, or a small stipend.

If your opinion is “yes”, should the community create a CU (pretty much the CU I have described above) that provides an “opinion” as to what is the best direction for MakerDAO? Because if the above mentioned folks take 90% of Delegate duties (MKR), I don’t see how we can keep growing our community, IMO.

@prose11 copying Payton here, as he has worked in legislative office before and he can attest to the support a politician receives from his/her staff.

2 Likes

At some point, maybe. As I mentioned before, I think it makes more sense to start small and react to the evolution of the system over time. So far, (with the exception of Gauntlet) each of the Recognised Delegates is operating as an individual.

I don’t really think that there should be a delegate CU, or anything of that nature, especially at this stage.

No, not necessarily. I think it makes more sense to have individual delegates than an organization in general. The more people involved, the more diffuse the responsibility.

I’m appreciative that people are considering the future of Maker, and of delegates, but I don’t believe that it’s a good idea to over-build processes or structures before there is a clear need for them. We’ve fallen into this trap a few times in the past.

We don’t need to figure out the ‘final form’ of delegation and governance in the Maker Protocol. We do need to make iterative improvements and mitigate any issues that we’re seeing. Over time the system will evolve to something that better fits our requirements, without us having to guess those future requirements in the present.

5 Likes

Fine, that’s not all that different from my proposal. The question is, how would we set the threshold?

Your suggestion implies that you don’t value marginal delegates. How would we encourage weak delegates to become stronger?

Works for me. I really enjoyed reading your take.

I encourage everybody to participate in the above polls because the questions will need to be answered anyway for the next step, whether it was a signal request or will be a MIP.

4 Likes

I also wonder if 8k DAI/month for the best 10 delegates would not be an acceptable solution (or 20 delegates with 5k/month which seems better as a start). That’s $1M/year for governance which seems fair at this stage as an expense (but governance can decide what is a fair number). Could be also limited by a minimum MKR delegated (I would say 100MKR to start, 1k end of the year then 10k mid next year in order to seed the delegate ecosystem). Having uniformity helps diversity and levels the playing field.

5 Likes

It occurs to me that the current proposal is underspecified if there are fewer recognized delegates than the target goal number. The option of 10 delegates is currently leading in the poll, but there are only 7 recognized delegates. One way to handle this would be to pay all of the delegates as if they are top-tier until there are more than 10.

1 Like

So as a counterpoint to this, having compensation scale by amount of MKR delegated incentivises delegates to reach out to MKR Holders themselves, and generally try to increase their delegated stake which may serve to increase the # of governance-active tokens.

I support having a maximum compensation to prevent centralization concerns, but I feel like it may be worthwhile to retain this growth and networking incentive for smaller delegates.

4 Likes

Made some spreadsheet changes here: Delegate Reward Proposal - Google Sheets

They are fairly minor, it supports the following options:

  1. Exponent
  2. Max annual cost for DAO.
  3. Max annual comp for delegates.
  4. Amount of comp transferred from delegates over the max.

It also no longer requires multiple rounds to calculate compensation.

Something I realized is that you can’t have both a maximum annual cost for the DAO, and also have minimum comp for the delegates alongside a non-fixed number of delegates. This is obvious when you lay it out, but bears considering. We ran into the same constraint when thinking about MKR vesting, with some arguing for a fixed cap, and others arguing for guarantees to core units.

3 Likes

This would be simple and a reasonable for now.

However is this good in the long(er) term?

I am thinking that this would make it difficult for new delegates to sustain their costs up to the point when they reach the necessary level. In other words, only rewarding the e.g. top 10 delegates seems to possibly have the side effect of creating an established elite of delegates, not open to (sometimes much needed) change.

2 Likes

Much clearer thanks for the edit to the sheet/model Long!

Although I originally thought we could move quickly for this, I support this process instead. Mocked up a quick map of the decisions we still need consensus on, plus the path to a MIP. We can use tomorrow’s call to refine these more and adjust.

3 Likes

Dai can has a vesting cliff too.

Ugh, this is going to take forever to hammer out. Is there any way we can reduce the number of decisions?

1 Like

Based on LFW’s spreadsheet, I added in a way to incorporate the tentative poll results, Delegate Reward Proposal v1.1 - Google Sheets

Specifically,

  • A maximum of 10 well-compensated delegates (cell B4)
  • A minimum of 10k MKR delegated before the funds available for delegate compensation are disbursed (cell B6)
  • 4k DAI per month (cell B7)

The way the 10k MKR threshold is used in the spreadsheet is not entirely consistent with the poll wording. I came up with the new formulae today, about a week after the poll started. (Oops!)

Given that we don’t have consensus on whether to continue or discontinue SourceCred payouts, I want to withdraw my preference in interest of getting consensus on the rest of the proposal. We can always circle back to SourceCred later.

3 Likes

Just to make sure I understand it correctly—the threshold is 10,000 MKR per Delegate, and Under 9,999.99+16 digits, a delegate does not receive compensation, correct? Or maybe I misunderstood (need to revisit this thread).

If so, I think it might be difficult to get 100,000 MKR delegated to delegates— I think in the last 3 years I’ve only witnessed the Hat exceed 100K MKR in maybe two/three occasions.

1 Like

Nope, compensation is still proportional to the square root of the MKR weight. The 10k MKR threshold only determines the size of the pool of funds to distribute. The size of the pool of funds is sum of column C plus 1 times B7. The “plus 1” is there so that there are some funds available for delegates with less than the maximum per-delegate compensation.

1 Like