Reducing the minimum dissolve delay to 3 months

I have no issue with the dissolve delays being reduced, but the way it scales is ridiculous. It should scale exponentially in my oppinion. If you make the 3m delay pay 7% apr, almost no one will stake for longer. Plus you are kind of spitting in the face of the 8y gang. And the 8y dissolve delay accounts discounts on idgeek will become even bigger.

Hi all, wanted to give an update from DFINITY side:

DFINITY supports the idea of proposal 131010 and will vote to adopt it later today.

We would like to set expectations however that the proposed change is more effort than suggested in the proposal and that it might not be possible to prioritize this feature right away.

  • The change requires more work than changing a constant. As a minimum, it requires an analysis and security review of whether pieces of the governance code rely on the minimum dissolve delay for voting being 6 months. This is more work than searching for the constant in the code - most work is to ensure that no guarantees are based on an implicit assumption that the constant never changes.
  • All frontends that integrate with the NNS would have to do the same - for example, it could be that some dapps that allow staking hardcoded the 6 months for giving feedback to users who set their dissolve delay. Therefore, the rollout will have to be communicated and planned well.
  • Since this is more than just a few hours of work, we will have to evaluate the priority of this feature compared to others. There are already some ongoing NNS-projects as well as upcoming planned features that will likely be considered higher priority, such as the periodic confirmation of following.

I wish everyone a great day!

8 Likes

Lower dissolve delay and up the APY for 8year neurons while you’re at it… :sunglasses:

1 Like

Each reduction in units of staked time will reduce the security of the network by an exponential amount. Cutting the minimum in half increases risk by a factor of 2*(x^n) where n>1. Unfortunately, there is no way to tell where we are in that decay curve. Black swans everywhere.

What gives confidence that this is a good choice? I’m not saying 3 months isn’t safe, just that it would be nice to see some significant thought and attention given to reducing security our weakest point in half(and increasing risk by over 2x) all at once.

I guess we can all keep fighting over the APY while no one notices that our network just got exponentially weaker. :face_with_diagonal_mouth:

Sorry to be the alarmist and pesimist in the room…but someone has to do it or we go to 0. :grimacing:

5 Likes

How did it become weaker? If more people stake, more people have voting power and its more decentralized. Can you elaborate?

@skilesare can you share some mathematical support for your claim that halving the minimum “increases risk by a factor of 2*(x^n)”? It’s a pretty bold assertion to make and I’ve seen no support whatsoever for the statement.

The argument is epistemological from the assumption that a cohort including 3 month horizoned decision makers will approve more unknown unknown proposals than a 6 month cohort. The falls out from the fundamental assumptions of fat tails asymmetries. More volatility leads to fatter tails. If you want to pick up pennies in front of a steamroller, at least admit the steam roller eventually catches up. We are moving halfway closer to the steamroller without any reasonable justification.

I think you may have a fundamental misunderstanding of what principals the NNS is built on if you feel the statement has no support.

If we want to be ETH2 then ditch the NNS and let node providers compete on execution and concensus slashing. Let them decide the replica direction and installed canisters by social concensus. That isn’t the game we’re pitched. The NNS’s only defense against liquidation of value by the mob is the amount of skin in the game and the time horizon that skin is trapped for.

Without qed proof that 3 months is safe(not possible with the breath of proposals the NNS supports) the the law of fat tails should apply and halving risk security should increase thr occurrence of catastrophic scenarios more than 2x.

Maybe 1 month is safe to less than one catastrophic event in 5 lifetimes, or maybe it less…much less. We don’t know and that’s what fat tails dictate. We have no mitigation strategy in place…except for the lock up length. In a catastrophic scenario where no voters can get out for six months we have six months to figure things out, which means we have exponentially more options than in a three month scenario because when you shorten the period you take entire possible universes trees off of the table.

The math: https://codowd.com/bigdata/misc/Taleb_Statistical_Consequences_of_Fat_Tails.pdf

3 Likes

Isn’t the security of the IC reliant on DFINITY, ICA, and the team holding more than 50% of the VP (without delegations)? I don’t understand why you are so concerned. If anything, we should consider removing stuff. Of course, you won’t find this explicitly mentioned anywhere.

Ultimately, probably there are offchain mechanisms in place to recover from a 51% governance attack. Maybe @lara can confirm and explain this.

:rooster::egg::rooster::egg:

Do what is right…prepar like you need to be ready for tomorrow …because it will happen…

…and then more people will care and more barbarians will be at the gate then you ever expected.

If i’m following, you’re saying the more people pile into the short end of the curve - and the higher the price gets, the value locked is being used as a pool to support the volume? Almost like a fractional banking scenerio?

We have 15% of rewards, there is no reason to assume the pie chart will change in a meaningful way.

I’m missing your pie chart example what does the number of rewards have to do with it?

The 8 year gang can earn and liquidate the sum of their current stake in around 6 years completely ceding the pie. I’m looking at least 8 years down the road. Perhaps others aren’t.(and that is the point).

Precedents take years to establish and only a moment to pull down.

The precedent was “if you can’t commit to this network for 6 months we don’t want your vote.” Now it is “we lower our threshold as soon as asked because we think it will help in the short term.”

2 Likes

Were you referring to me? If so, that was not my point. My point was that as you lower the future time-decision point of the network, the volatility of future universe increases due more people making short term decisions than before. An influx of 3 monthers will dilute the longer focused people(increasing volatility) AND open the network up to a new set of governance attacks.

Yeah, a lot of unnatural volatility is never good - in terms of artifically tinkering with things not intended as the original vision. I’m not a dev, so I can’t comment on most technical aspects of ecosystems until I learrn more about them.

I just think messing around with yields will most likely start a “tug of war” effect on the curve until everyone’s happy. I’m not saying it should never happen, having new people invested is a huge positive, but this new implementaion feels a little rushed.

The 8-year gang is probably mostly Dom and his friends. That kind of long term holders. They don’t need to liquidate with that amount of rewards. That’s what protects the network.

This proposal is more relevant for all the Anon69s who can now lock $100 worth of tokens for 3 months instead of 6 and get VP.

Then we will likely have WaterNeuron and similar using 3-month neurons. Probably, that VP as a whole will end up locked for more time.

I think you might be gravely mistaken that it is only Dom and his friends staking 8years, maybe in the amount of ICP but not in the total ICP holders… Hard to tell of course.

But i like what @skilesare is quoting here and that is what most long term holders are feeling:

"The precedent was “if you can’t commit to this network for 6 months we don’t want your vote.” Now it is “we lower our threshold as soon as asked because we think it will help in the short term.”

You set up a 20 year roadmap and there is constant discussion about rewards and the time you can stake.
If you not ok with the conditions then just wait until Defi kicks of on the ICP, if you don’t trust DeFi and prefer to stake in the NNS then you agree on the set conditions.

1 Like