[RFC] Lower the 100% delegator bond requirement

As an ATOM holder, I am presented with a dilemma for my holdings. I can delegate these holdings to help secure the network or I can keep the tokens liquid and use them to transact on the network. I’d like to propose adjustments to delegator bond requirements that will alleviate this dilemma thus allowing a holder to participate in delegation, network security, and governance while still retaining optionality for transacting with ATOMs.

A short story of how delegation works today: I own 100 ATOMs and decide I’d like to be involved in network security and governance. I delegate these atoms to a few validators to help decentralization. After a few days, I have a friend that wants some ATOMs to learn more and get involved. I go to my ledger to send them some ATOMs only to discover that I have to wait 21 days before I can send any of my ATOMs! I am aware that as a delegator, there is some portion of my ATOMs that could be slashed (and accepted this risk when I delegated) - but I didn’t imagine that delegating would prevented from using some of my ATOMs on the network. I start the un-bonding process and now question if it makes sense to participate in network decentralization and governance with any of my ATOMs or keep them available for transacting.

I believe that for most ATOM holders the above realization will be non-obvious and unpleasant. Certainly a bond and lock-up period are important aspects for proof-of-stake networks - these make up the stake. However, the bond does not need to represent 100% of the delegator’s balance. For example, today, with a slashing rate of 5%, if I delegate my 100 ATOMs to a validator I am at risk of having 5 ATOMs slashed should the validator double sign. From my perspective as an ATOM holder and delegator, the other 95 ATOMs should be liquid for immediate use since they are never at risk of being slashed.

I’d like to hear thoughts and counter arguments on the above example and experience. I don’t imply that the bond must be lowered to 5% but that it should be lowered from 100%. Lowering the bond requirement would provide a more pleasant and welcoming experience for ATOM holders to become delegators to help secure and govern the network.

2 Likes

Current design of unbonding penaltimate delegators who frequently/suddenly unbond their stake, result in more stable staking status for the network.

The penalty is done by 2 ways, illiquidity(3 weeks) and lose of reward during unbonding.

If those penalty disappear, the staking status of ATOM will be unstable and volatile, correlated to ATOM market price, because the traders have nothing to lose to suddenly liquidate most of their ATOMs.

Reward and illiquidity are a pair I believe. Holders should give up either one. If a holder wants both, he/she can half stake, half unstake, result in half reward and half liquidity.

I don’t agree to give up stable staking status of ATOM to provide reward and liquidity at the same time. Because prime utility of ATOM is to stake, not to trade, the stability of staking status should override the demand of liquid trading.

3 Likes

IMO an inevitable evolution of the Cosmos Hub protocol is going to be enabling ATOMs to collateralize more computations.

As things like the InterChain collateralization protocol mature I would expect validators where maximum slashing might exceed 100% of bonded ATOMs

Traders would lose the rewards on a larger stake. We also haven’t seen this volatility play out on other PoS networks so I think this is unlikely (though would love to see if there are counterexamples).

Looking at this proposal another way, the security of the network comes from the value of the network x slashing cost. Locking up tokens doesn’t contribute to that value, while liquidity is likely to by allowing more trading of those funds. For that reason I also support this proposal for anything not slashable.

Volatile staking status is not a binary event, but a relative term of prudentiality. And dPoS history is very short compared to even short history of PoW. Non-existence of counterexample does not prove it is unlikely.

When we invest cash or stock in a bank account(additional interest) or a pension fund(tax benefit), it is very reasonable to trade reward vs liquidity. It is very natural structure to bring up “sound” mid/long term investment and to achieve stable market. I see 3 weeks as a minimal structure for this. Different dPoS also provides higher rate for longer illiquidity.

In short, fast liquidity needs panelty, longer illiquidity needs incentives.
In Cosmos, the source of delegation reward is from dilution of liquid(short term trading) ATOMs. If most of delegated ATOM can be unbonded right away, most short term traders also delegate most of their ATOM and frequently bond/unbond their holdings. Then, the source of reward(dilution of short term traders) will reduce therefore lack of staking incentives.

I see short term price gain from allowing more trading of big funds, but in longer term, I suspect the effect because capital incentive driven funds are not a good source of decentralization and
possibility of fast/sudden exodus of delegated ATOM can harm prudentiality of the network and market price.

This is a great example!

With that context, I think a happy medium would be economically ideal. We’re all incentivized now to keep some unbonded tokens for liquidity, which reduces the amount staked i.e. security. Allowing some % of tokens to be redeemed without an extended unbonding period would help ensure a healthy market while maximizing staked coins.

That may however introduce complexity that’s not worth it right now for a new network.

Max_immediate_unbonding_amount_per_day = staked amount / 21
Above might be an intuitive approach for acquire both stability and some immediate liquidity.
Delegator can immediately cash out at least each one day amount.

I would suggest you dive into Sunny’s writing. Paper clarify the rationale behind the existing model. I personally think that reasoning is sound enough.

@zaki It seem right that any change in slashing beyond the 5% today would require a governance proposal. I also don’t mean to imply that whatever rate is set today for the bond is immutable. Just like this aims to adjust the bond with a proposal - future changes to slashing amounts can just as easily request an adjustment to the bond to accommodate their slashing requirements. This sends a clearer message when requirements change.

@xhipster This link is not loading for me. Since it is on ipfs - there may no longer be a node hosting it (or the node is offline). Can you please link to a location that will load?

I’ve just suggesting that I expect that this feature will eventually become unnecessarily because it’s in the interest of Atom holders to optimize their validator to be fully collateralized.

I am fine with the current settings. My rationale is that atoms are designed intentionally to be illiquid. Atom holders, no matter their purposes for holding atoms, need to do their own planning before making the decision to stake or to trade atoms. If they want to stake but foresee the need for some liquidity, they should plan it beforehand.

Suppose we try this thought experiment. Let’s say that the cosmos network launched with a bond lockup of 10%. I now propose raising the bond requirement to 100% and use some of the counter arguments used here as the argument for why we should raise the bond to 100%.

What arguments do you put forth as to why raising to 100%? Before shrugging this off - do think about it from the perspective of a flipped status-quo. I find that helps when evaluating tradeoffs like the one this post proposes.

I should raise another concern about this topic.

If majority of staked Atom can be immediately unbonded, it is a great environment for big centralized exchanges to bond most of their client’s Atom. I expect many centralized exchanges will adopt it because they can comply with customer’s sudden withdraw.

A lot of huge exchanges will be top validators of Cosmos network, causing worst scenario of validator centralization.

Therefore, illiquidity of staked Atom is an intentional inconvinience for the sake of stability&decentralization of the network. We really need to think about why we should care much about “short term trading demands” by sacrificing such important shields for the network.

https://github.com/cosmos/cosmos/blob/master/Cosmos_Token_Model.pdf

I read the above referenced paper. The cosmos mainnet launch has already diverged in economics from the recommendation of the paper by using a single token for staking and fees. Pointing this out to highlight that the paper is only an initial set of ideas that should evolve with new feedback and information.