Current Value: 1814000000000000 nanoseconds (21 days)
New Value: 604800000000000 nanoseconds (7 days)
Summary
Reduce the Unstaking Period for nodes from 21 days to 1 day given the recent issues seen with Silvergate and any downstream effects yet to be fully revealed.
Motivation
With the recent issues with Silvergate and the scramble of many other crypto ecosystems rushing to find banking, the possibility of a contagion that takes out many existing exchanges is now something that should be on the minds of anyone wishing to invest tokens into providing infrastructure. Should such a crisis occur, the current 21 day event unstaking period would leave too many node runners unable to sufficiently react to the circumstances. In the current market, this is going to drive down any desire for individuals to make the commitment to acquire more POKT to stake on the network and run nodes.
Rationale
1 day was selected, as it still gives enough time for the network to see and properly respond to unstaking events, but also is short enough that it would not overly problematic in a fast moving crisis that would likely leave a short window of opportunity for existing node runners to act accordingly.
Dissenting Opinions
1 day is too short; it compromises the integrity of the network by allowing nodes to unstake so soon
The UnstakingTime was set in a period of no open liquidity, and on a network with no established traction. Given the availability of liquidity through exchanges and such a robust network of node providers and relay chains, the network is at a place where such aggressive lockups are no longer needed to protect the basic functionality of the network. 24 hours would still give everyone opportunity to react to the unstaking events.
Great PUP, absolutely supporting the general idea.
However, I think this is way too aggressive. I’m all for the idea and definitely supporting it, but I think it should be done gradually. Maybe at some point in the future it might be lowered to 0 days (instant unstaking).
For now, I believe the best balance between pros and cons of such change is to put it to 7 days instead of current 21 days. There are many other benefits from decreasing the time besides those mentioned here.
ByBit is one of just three exchanges with a POKT/USDT pair that makes their reserves data available to CoinMarketCap.
While I understand the implications this has in the broader node staking provider ecosystem, being too conservative with these changes has the potential to leave everyone with POKT staked in nodes - big or small - independently run or using a service - all stuck in the same situation.
I am all for doing a change of this type. Not sure how many days is ideal, but I’d check out other chains’ unbonding periods for some reference points.
I should add that PNI now runs 1200 nodes across 6 providers, staking for about half of the chains the network supports. We have no plans to unstake those nodes anytime soon, so there will be stability there from at least our nodes.
I believe 21 days was originally decided on as a way to avoid destabilization of the nascent network. We are years beyond that, so I am all for making a change here. I’ve spoken about it internally before, so I’m happy this is being brought up for discussion.
Aligned on lowering the unbonding period and the rationale. Will add that Cosmos 21 day unstaking is rational for their network due to the interoperable nature of IBC but it’s arbitrary for us to follow it just because we use Tendermint. We need to find the right value for Pocket specific utility.
Interesting point and recent historical context about locked withdrawals on exchanges. Really flips the script on LV. I can’t say with certainty that 24 hours is the right amount of time since we have to be mindful of demand availability. Would be great to see deeper analysis on how small the window can be against network capacity requirements and scenarios to ensure consistent service. We are overprovisioned now but want to be mindful of potential future demand.
Thanks for the proposal and welcome back to the forum!
PS want to add that consideration’s for validator security and consensus should be included in an analysis. It may require a secondary implementation of either longer unbonding periods for validators or adjusting the total number of validators on the network.
In short- in favour of shortening the un-bonding period.
But not exactly for the reason stated. I personally would be market-condition agnostic on this subject.
A couple of months ago, someone reached out to take feedback on 21 day un-staking and my response was-
In the beginning, its fine to play safe and conservative but with tenure the frictions have to gradually go down in favour of the users, and taken to zero wherever feasible. If I were to use the lender or depositor analogy- the shorter the withdrawal period, greater the confidence. Same goes for transmitters.
The range is wide in the market (to cite a random few), although I would assume they are generally validators-
Sol- instant
Kusama- 8 days
Matic- 9 days
DOT- 30 days
Helium- 5 months
This is arbitrary but one-third reduction to 7 days is a big improvement. Doesn’t really have to be a day.
Can you clarify here what you think “sufficiently reacting” and “acting accordingly” is? Because it sounds like you’re suggesting that should a liquidity crisis occur in exchanges, you want noderunners to be able to rapidly participate in a run on the bank, which only further exacerbates liquidity issues.
Being in the situation of having staked nodes? Again, I’m not following the concern here.
This was EXACTLY the concern regarding the Transfer Stakes proposal fleshed out some time ago; flash attacks on governance. That proposal was halted given the easy to flesh out schema of flash staking validators to block consensus or an upgrade, given that validators hold the final veto power in our ecosystem.
There are significant security concerns around a move of this type.
Absolutely correct, and proven by the extended chain halts due to issues achieving consensus.
I very much want to hear the answer to what the noderunners are supposed to be doing in reaction.
But let’s talk the market while we’re at it:
If the goal is to make it easier for whales to flood the market on a 2x price move and insta crash the price gain by scalping, this is definitely a solid strategy. I can instant unstake 50% of my farm, sell for profit, dumping the price back to the floor, rebuy at the floor, stake, rinse and repeat. I’ll constantly have more than seven figures staked, and it will only be a day or two to restake after taking my profits. If another run happens in that window, I can unstake the other half my farm to catch that upside too. Given the volume of tokens staked, I only need 3-4 cent gains in the current market to make that a highly profitable recurring strategy.
Small node runners and independents don’t have that flexibility, so they would get hammered by the constant small band volatility without being able to profit from it. And every morning would be a prisoner’s dilemma checking the unstake numbers to see who is about to make a big move, and selling down into the market ahead of that move.
We’ve already seen this pattern in the market over the last handful of months, and increasing the ability to scalp quickly with larger numbers while also maintaining the staking operation generating the tokens to dump on the market seems like a ready made recipe to centralize control to whales quickly.
I agree with @Caesar 's sentiments from the high level, with the added caveat that none of those except Helium are infra staking, and Helium has triple the unstake time. Pocket’s product type needs to be factored into those considerations.
If the goal is to make it easier for whales to flood the market on a 2x price move and insta crash the price gain by scalping, this is definitely a solid strategy. I can instant unstake 50% of my farm, sell for profit, dumping the price back to the floor, rebuy at the floor, stake, rinse and repeat. I’ll constantly have more than seven figures staked, and it will only be a day or two to restake after taking my profits. If another run happens in that window, I can unstake the other half my farm to catch that upside too. Given the volume of tokens staked, I only need 3-4 cent gains in the current market to make that a highly profitable recurring strategy.
Small node runners and independents don’t have that flexibility, so they would get hammered by the constant small band volatility without being able to profit from it. And every morning would be a prisoner’s dilemma checking the unstake numbers to see who is about to make a big move, and selling down into the market ahead of that move.
This was never the intention. I was not considering the effects that token concentration would have had here. I was simply considering the timelines of the slow trickle of exchanges slowly closing shop.
I’ve seen a few compromises of 7 days. Does that time frame still run into these same issues? Plenty happy to amend this PUP to that timeframe if not.
Being in the situation of having staked nodes? Again, I’m not following the concern here.
Being in the situation of having staked nodes with no access to the same on/off ramps that retail expected to be available when deciding to stake those nodes. I simply find it personally hard to swallow the risk right now of choosing to lock up a bunch of capital to stake nodes should the public exchange liquidity vanish in the time it takes me to personally unstake.
Thanks for bringing data to the thread! Adding for consideration currencies that have you lock for x days before any unstaking can be initiated ($CRO) as another strategy to explore when thinking how to approach analyzing what works for POKT use case.
I myself am a big proponent of benchmarking, specially when there aren’t too many determining factors that are objective by nature. Another way could be to check what all other “web 3 infrastructure projects” are doing. That makes the comparison niche and focussed.
I’m not sure, and I think we need to model out some impacts to have a good sense of what’s feasible, but more importantly:
…it sounds like my original assumption about the outcome was the correct one here; that this is a mechanism to allow dumping in a run on the bank in the face of perceived collapsing liquidity. And bank runs due to market panic frequently are a self fulfilling prophecy in completing the collapse of liquidity. I disagree with the premise that allowing faster dumping is a net positive.
POKTs inability to be liquid on either side of the market is dysfunctional. High liquidity means that the price a buyer is willing to pay and sell move closer together and is a mark of a healthy market.
Counterpoint to strict LV is would you support a bank not letting you withdraw to prevent a bankrun? A common reason why people support crypto.
Our economics, from staking to buying, are objectively unattractive.
I encourage people on this thread to reevaluate parameters set based on a fork of a 7 year old platform rather than doubling down on what the market has rejected.
Validator security is a concern that needs to be separated from preventing entries and exits from the market. I see no reason why service nodes require this level of “accountability” and disagree that this is the right way to incentivize for redundancy. Hence my suggestion to only hold validators accountable to high unbonding times while providing more flexibility to nodes providing service and those who custody their funds with them.
I absolutely agree with the need to discuss marketplace dynamics.
The framing of this, though, in relation to fears about CEX liquidity, is the wrong framing in my opinion, and sets up the very scenario I was concerned with on first pass.
I don’t think treating this from a purely financial instrument perspective is ideal given some of the other complications that come into play there, but “Suspension of Convertibility” (a terribly fancy way of saying suspending withdrawals) is something baked into our banking system whether we support it or not, and has been used, or threatened to be used a number of times to prevent bank runs. Humans are irrational beings, and a business or institution may be liquidity crushed by a rumor when that business is actually healthy enough to maintain normal operations unless hit by a bank run.
I support exploring all of this more. The general framing (and clearly intended outcome in an already exhausted market) is my core concern.
The churn cycle I outlined preventing any chance of a healthy recovery would be my biggest concerns from a #2 perspective, since we’ve seen that cycle play out already countless times.
#1 is a massive risk from the validator pool perspective, so any change like this would likely require a separation of the servicer and validator cooldown time, introducing a new parameter and a consensus breaking change, and necessitating engineering resources to build and test a new branch. Diversion of those resources away from v1 seems like a bad idea on all sides.
In general, I am all for deregulation and movement toward free-market economy… in this case shortening and removing unbonding time. But “movement toward” does not need to imply “all at once.” Ie, maybe 7 days as a logical next step. Or maybe even 14 days as an intermediate step to 7 days…
But to quote a question posted on a POKT-related telegram thread on this subject, “why now?” We are years into the project at 21 days. Can we continue another year and consider update to this parameter as part of a wholesale parameter-set re-evaluation that takes place during transition to v1? Tying the discussion to market events such as Silvergate or Bybit or whatever is irrational. The matter should be decided on its own merit, independent of market conditions, as mentioned by @Caesar .
I agree with @RawthiL that recent history shows that 24 hrs is too short to react. 7 days is more than plenty; 3 days is probably sufficient. That being said, it can reasonably be counter-argued that there is no need to make allowance for a reaction time in the first place. The network is very robust to extreme episodes of syncronized staking/unstaking. That is why I opposed @JackALaing slow-rollout guidance he issued to node providers last July in the wake of PIP-22 approval. I am pretty sure the network would not even blink if half the Servicers (and current validators!) unstaked all in a single block.
Appreciate the feedback and show of commitment to stay staked. That being said, while we can factor that kind of sentiment into the law of probabilities, the DAO cannot and will not make governance decisions based on trust that one actor or another will or will not take certain actions.
Not my area but would love if someone in the know could comment. Tendermint or not, it seemed there was sentiment expressed in the community toward interop with IBC. If that would require us to adopt 21-day unbonding, that would be a good thing to know.
I cannot think of any scenario where 7 day unbonding time would create negative consequence on either validator security or consensus. If we staged a decrease in unbonding time, we could start there while exploring and mitigating any issue that might appear at the 1-day or less level.
I disagree with this statement on two counts. First small node runners and independents are not at all disadvantaged with respect to flexibility. It is an equal playing field in this regard. If anything the small and independent are more agile in ability to move in and out their nodes. “Keeping half staked” etc is a strawman… if it is more profitable for a small or independent to scalp, than run a node, the having to unstake all vs half is hardly an impediment. Further, if the are adept at market timing they might even being more profitable at the scalping game than large noderunners as they can sell high and buy low with little to no slippage. The large guys retain the edge of course in ability to move the market the direction they like, but that is another story.
The real disadvantaged in terms of flexibility are the clients of custodial node runners - , as they have an extra layer of bureaucracy to deal with in unstaking that independents and the node runners themselves don’t have to deal with. This has nothing to do with large vs small.
Second, I do not see how small-band volatility created by those who intermix node running and scalping hammers node runners who stay staked. Those who stay staked ride through the daily or weekly volatility like a tanker riding through the waves of the ocean. If anything they would see a slight average increase in reward levels as some of their node running competition is off grid a certain percentage of the month chasing the scalp.
As to the bigger picture question of the effect of lowering unbonding time on price volatility, I’d have to give it some thought. It is a complex topic. My gut feeling is that it lowering unbonding time has little overall effect on volatility. It may even have an opposite effect if release from unbonding constraints removes synchronization of movement that might occur currently. I have no confidence one way or the other atm; would really need to give it some thought.
Each person makes decisions based on their risk tolerance etc. And everyone plays by the same rules (in this case 21-day unbonding). At any time one can make the personal decision to stop node running so as to can keep POKT in liquid form rather than locked form according to the dictates of their personal risk/reward balance
This is a worthy topic to explore post v1. Differentiating unbonding times is a consensus-breaking protocol change. I doubt it rises to the level of importance to garner serious willingness to execute during remainder of v0. Universally changing unbonding time, on the other hand, is a simple parameter change that could be accomplished now if desired.
To add my thoughts, why should a retail staker be punished if a provider goes down or has bad performance, or raises their rates (whether fair or not). IMO the 21 day unstaking period was only ever to protect early investors and staking businesses close to the inner circle, but that’s just my opinion. There is an argument that for anyone trying to establish a business around pokt relays, that having too quick an unstake time might make it difficult to predict revenue and runway, etc. Pocket doesn’t owe anyone that luxury. Another argument for longer bonding times - to create friction and limit consumer options - feels to me more like ponzinomics (if we keep them staked and make it inconvenient, more people stay staked and number go up). We should not be catering to ponzinomics. 1 day is too short, but about a week feels like the sweet spot to me.
Given that 7 days has come up organically, and I admittedly selected 1 day arbitrarily hoping to start this discussion, I edited the proposed value based off of everyone’s feedback.