Should we change the stability fee poll cadence?

This has been discussed previously several times. I’m creating this thread in an attempt to gather consensus and to move us forward to a decision on what the stability fee cadence should be. I’ve intentionally left out specific options for the initial stage. Two weeks was always somewhat arbitrary so we should have a chance to discuss other options.


Advantages of a longer cadence

  • It allows us to observe the results of stability fee changes without confounding those results with more changes.
  • Infrequent measured changes look better from an outside perspective than rapid changes with questionable scientific backing.
  • Some MKR holders cited the frequent votes as a contributor to voter apathy, a longer cadence would reduce this.
  • Gives MKR Token Holders more time to push through the executive vote before it is potentially replaced by a new executive.
  • Allows more time for discussion and consensus building on each rate change.

Advantages of a shorter cadence

  • It allows us to more quickly react to changing market conditions without resorting to exceptional procedures.
  • In an ideal world, the stability fee can better reflect the market conditions at any given time.
  • It could be argued that a shorter cadence keeps voters involved with Maker as there is a vote occurring every week.

If you have additional advantages for either option, please express them in a comment and I will add them to these lists.

I have described my plan for the progression of this (and future) proposals here. Please feel free to provide feedback, the process is open to change.

We’ll start with an initial sentiment poll to determine whether there is appetite to change the cadence from the current cadence of 1 week. Please use this week to signal and to propose and discuss specific options. If we have a majority (>50%) for change after 1 week has passed, I will add a poll to gather consensus on the proposed options.

This initial sentiment poll will remain open throughout the process.

  • Yes, we should change the stability fee poll cadence.
  • No, we should not change the stability fee poll cadence.
  • Abstain (I just want to see results)
  • Abstain (I have no opinion)
  • Abstain (I don’t feel I am knowledgeable on the subject)
  • Abstain (I disagree with the poll options)
  • Abstain (I have a different objection)

0 voters

2 Likes

2 weeks is good for me

1 Like

I am fully in favor of moving to a 2-week cadence.

The main reason for this is that 1 week is simply not enough time to:

  • properly gauge market response to any single stability fee change, or
  • give the world enough time to find out about the rate change.

2 weeks may also not be enough time, however it makes sense to move this cadence in 1 week increments until we find the sweet spot.

The benefits listed above for a shorter cadence all assume the following:

  • the world is able to hear news of a stability fee change nearly instantly and can react within a day or two.
  • that the first few days of action after a stability fee change are representative of a longer-term trend that may (or may not) occur.
  • that any change in peg behavior that may occur immediately after a stability fee change on a 1-week cadence would be instantly attributable to the most recent fee update.

None of the above can be proven to be true. Therefore, a shorter cadence should not be considered unless other advantages can be presented.

1 Like

Have the foundation risk team expressed a desire for longer or shorter cadences? Sorry haven’t listened to all the governance calls lately.

Also

Have we considered some kind of forward guidance on SF changes? E.g. if the peg is off by more than x we change the peg with Y. If the peg is of by 2x we change the peg 2y etc. Or something like that. Obviously it would be the executive vote that determines the change no matter the forward guidance but maybe a forward guidance could be somewhat ratified by a on chain poll.

Benefits of forward guidance imo would be that SF changes would be somewhat anticipated by the market. Hence the effects of a SF change would manifest faster in the data. This I assume would be beneficial especially if we move to longer SF poll cadences which is why I am raising this question now.

Formulating and finding consensus on such forward guidance would also be a nice preparation for automating SF changes one day. This forward guidance would kind of be the spec of such a SF change algorithm.

I don’t really disagree, but I couldn’t think of any others. If anyone from the short cadence lobby wants to throw up some more advantages to shorter cadences in general, I’d be happy to add them to the OP.

Maybe, this decision process is fairly expensive in terms of time. We should try to find a good spot now or potentially we could agree to ‘trial’ 2, 3 or 4 week cadences for a few months each, and see which works better.

I would guess that @Vishesh would be in favour. Not sure about @cyrus and @Primoz

This came up previously in this thread: Signal request/ Poll: Define standard stability fee changes based on recent price of DAI There was some discussion, but I think it’s more complicated than it sounds. I’m undecided on it.

Thus far this is seems fairly contentious at 6 to 6. Do any of the voters for status-quo want to share their arguments as to why a shorter (one week) cadence is preferable to a longer one?

@tbone @NikKunkel @swakya @jernejml @Aaron_Bartsch @lix

To me the arguments currently look rather one-sided in favour of moving to a longer cadence, it would be great to get an idea of your reasoning so that I can either add it to the initial post and/or argue the points. In the principles of scientific governance, could you guys share why you believe what you believe?

Is anyone considering moving directly to a three or four week cadence?

1 Like

I think it outweighs all other arguments combined.

4 Likes

Can’t say I thought very deeply before voting here. I’m guided by the idea that voting should be as continuous as possible. Ideally we’d have a module that sets the fee as the vote-weighted average (with a fixed delay) and anyone gets “SF module” votes they can change at any time. Since we don’t have that, the faster the cadence the better the continuous approximation.

Of all the arguments for a longer cadence, the one about having time to observe the effect of changes is the most persuasive to me. We are in a transition phase anyway (and probably will be for a while!) so I’m completely open to experimentation. I’m happy with us going for 2 weeks.

1 Like

I used to think we needed slower changes too. I changed my mind during the long stent below 1 we had late January - May of 2019. Voting every two weeks is not nearly frequent enough to respond to demand for leverage due to ETH volatility. If feasible, I would vote every day. I realize this wouldn’t work due to extreme governance fatigue, so settling at once per week is already making a trade off to allow inefficient decentralized governance.

Hypothetically speaking, lets use a centralized perpetual swap contract (bitmex, deribit) funding rate as crude representation of the SF in SCD - they need to change funding rates every minute to keep their contracts at 1. What would happen if they could only change once per day? Once per week? I realize they have completely different requirements than we do, but I think you get my point. They would become unusable due to the volatility of demand for leverage. Voting once per week is already making a significant sacrifice to allow decentralized governance at the expense of a less stable DAI. We should not be extending this sacrifice to give us better observations, less governance fatigue, or more time to build consensus; keeping DAI at 1 is far more important.

3 Likes

This is my view as well. We need to be able to react to changing market conditions. Changes in demand and supply can fluctuate much quicker than our ability to respond in a long-cadence world. We’re endangering the peg by limiting ourselves to such slow reaction. Additionally, we risk impeding Dai supply growth.

Take today as an example. We just had an executive vote to lower the stability fee last week. However rates on competing lending platforms are lower than ours, and the Dai peg is running high. Theoretically we should be lowering the stability fee, to stabilize our peg and capture Dai borrowing demand that we currently lose to competing platforms. However, in a world with a long-cadence we would be relegating ourselves to not adjusting the stability fee until next week. Meanwhile Dai would not be at the peg, and we would have lost a prime opportunity to scale the Dai supply.

4 Likes

Would it be feasible to have a dynamic cadence depending e.g. on a average of the peg? E.g. If the peg is of by X% then weekly cadence. If the peg is of by 2X% then change SF twice a week. If the peg is within X% then no changes or monthly?

Having a dynamic cadence of some kind would also have a forward guidance effect to it. As it states to CDP owners that if the peg is of we will adjust fast.

Thanks @lix and @NikKunkel for providing your reasoning against the change, I’d love to see more add their thoughts or voice their support for this reasoning.

This has now been active for a week and there is no majority for change from the status quo, so we shouldn’t move on to specifics.

We’ve previously seen more than 16 voters and this issue seems quite contentious, so I’m inclined to leave the poll open for another week to allow further discussion and give more people a chance to respond and argue their points of view.

I think this should be a general rule going forward on contentious issues, possibly something along the lines of: ‘If the leading option has less than 66% agreement after seven days, the first stage is extended by another week.’

All this said, I am also pro-change, so I recognise their could be motivated reasoning occurring here. Please judge me accordingly.


I would suggest the following process for moving forward on this issue:

  • Poll remains open for another week for discussion and trying to reach a majority for change. My reasoning is that this is clearly contentious and we’ve previously seen more than 16 people vote on a specific poll.
  • If the poll reaches a majority for change after that week, we’ll move to stage 2.
  • If the poll fails to reach a majority after the next week, I’ll close the poll and we should consider the matter settled for a period of at least 1 month.

I plan to assume agreement with this process for the time being, but please express your opinion on it below.

  • Yes, I agree with the bulleted process above.
  • No, I disagree with the bulleted process above.
  • Abstain (please give a reason)

0 voters


@DLB @ChrisBlec @Jeffshaw @kwadrax @Mitote

Pinging the other pro-change voters for action, if you can’t convince or rally a majority over the next seven days, then this process will end without change.


I’m still of the opinion the default cadence of the stability fee poll should increase to two weeks. I think there have been instances in the past and in the future where one week is not nearly enough time to observe the stability fee adjustment. Maybe we can add the option for governance to institute an emergency poll if market conditions dictate we intervene sooner than every two weeks for a short period of time. I get that eventually, we want to have a system that continuously adjusts (every minute), but we really need to learn about how the system is working right now and increasing the cadence will allow us to learn more how the system is working to changes in the stability fee.

1 Like