What are the challenges of the Cosmos and how to tackle them?

Yulin Liu
The Startup
Published in
9 min readAug 14, 2019


Disclaimer: The views expressed in this article are those of the author and do not necessarily reflect the official position of the DFINITY Foundation.

Cosmos is a proof-of-stake (PoS) blockchain project that empowers the interoperation of multiple independent blockchains [1]. This article addresses its design of token economics and on-chain governance system. The views and concerns expressed in this article should not be taken as criticism on Cosmos but rather respectful discussions with good faith. The article was sent to the Interchain Foundation (curator of the Cosmos) for review before publication.

Token Economics

Users pay gas fee in “atom” (the native token of Cosmos). 2% of the gas fee flows to a reserve pool. The fund accumulated in the pool is saved for system upgrades and is determined by the Cosmos governance system (see the next section). The rest of the gas fee, together with newly minted atom tokens in each block, is distributed to validators in proportion to the percentage of atom tokens they stake in the system. The more atom tokens staked, the higher reward received. Validators run full nodes and secure the Cosmos network by collecting, assembling and broadcasting transactions according to the Tendermint consensus protocol.

The system starts with 100 validators and targets 300 validators in 10 years. Those who stake the most atom tokens in the system become validators. Those whose staked atom tokens are outnumbered by new entrants, lose their validator status, a.k.a. Liquid PoS.

Apart from their self-staked atom tokens, validators could also offer staking service to atom token holders and attract their delegated atom tokens. In other words, atom token holders could delegate their tokens to validators and share gas revenue and block rewards with validators without running a full node. In return, validators could charge a certain commission for the staking service.

If violation of protocol (e.g. double signing, network disruption etc) is detected, a fraction of the staked tokens is slashed. Thus, delegators should carefully choose reliable validators for staking service. Otherwise, their staked tokens get slashed proportionally as well when their validators breach the protocol.

Cosmos sets atom token inflation rate between 7–20%. High inflation rate, i.e. high block rewards, dilutes the wealth of the atom token holders who do not stake their tokens and thus incentivizes them to stake their tokens in the system. The inflation rate is automatically adjusted so that the percentage of staked token attains the target of 66.7%. Specifically, when the staking ratio is lower than the target, the inflation rate is raised so that atom token holders are more encouraged to stake their tokens. Vice versa.

Aside: Inflation is defined as the change of purchasing power of a sovereign currency. Tokens have not been widely accepted and used as a medium of exchange to purchase a variety of goods in the real economy. Therefore, the term “inflation” used in the crypto world is a misnomer. To indicate the change of token number, the correct term is token dilution/growth rate.

Governance System

Validators are required to vote on all proposals. Failing to do so in a timely manner will invoke a temporary suspension of the validator identity for one week. The on-chain governance system of Cosmos can be divided into three phases:

Phase 1: Proposing

Anyone could submit proposals to the voting system. To enter the voting phase, the proposal needs to attract a minimum deposit from the proposer or other atom token holders. Phase 1 lasts maximally two weeks.

Phase 2: Voting

Atom token holders gain voting power by staking tokens in the system. The voting power is proportional to the number of staked tokens.

Voters have five options: Yea, Yea with Force, Nay, Nay with Force and Abstain.

Token delegators could cast their own votes. Otherwise, they automatically inherit the delegatee’s vote (a.k.a. liquid democracy). Phase 2 lasts two weeks.

Phase 3: Tallying

A proposal is considered passed if and only if the following two criteria are both satisfied:

  1. More than half of the voting tokens have voted Yea and Yea with Force
  2. Less than one third of the voting tokens have voted Nay with Force

The deposit staked in the proposal will be confiscated to the reserve pool if the majority of voters consider it as a spamming proposal. Otherwise, it will be returned to the original owners. Successful proposals will be implemented by validators.

Potential Concerns

Is it necessary to impose the artificial lower bound of 7% token dilution rate?

The upper bound of 20% token dilution rate is set to prevent the system from diluting non-stakers’ wealth too quickly. While the lower bound is set, presumably, to guarantee minimum block rewards validators could receive. As the security of the Cosmos network relies on validators, it seems to be valid to draw a bottom line to assure validators that block rewards will not diminish to zero.

However, what matters to validators is not the number of newly minted atom tokens per block they receive but rather the value of the block rewards denominated in fiat currency. Even if the dilution rate is lowered to less than 7%, as long as the price of the atom token is sufficiently high, validators could earn enough fiat revenue to cover the operating cost.

One typical example is Bitcoin. The Bitcoin dilution rate has decreased from more than 100% to around 3.8% in recent years (see my article on Cryptocurrency Valuation [2]). In contrast, the number of miners and hash rate have increased significantly, making Bitcoin network more secure than ever.

Moreover, 7% dilution rate makes atom token less appealing to liquidity users. Without these daily active atom token users, what is the sense of making the Cosmos network more secure?

Staking period (21 days) is way too short. Speculators could vote for proposals that are in favor of short-run interests.

The solution is very simple. The voting power should increase not only with the amount of staked tokens but also with the staking period. The longer tokens are staked, the larger the voting power (see my article on Wonder Woman Chain [3]).

The current voting design does not scale as the system grows.

Voting is effort and energy costly; voters need to go through numerous discussions in the community, take into account different opinions, and think through them before casting votes. As the ecosystem grows, proposals on different topics will be initiated more frequently. It is neither efficient nor fair to require all validators to vote on every proposal.

One solution would be to try Assessment Voting [4a] (see also Random-Sample Voting [4b]). Simply put: for every proposal 20% of the validators are randomly selected to vote within one week. If the voting result is clear, e.g. more than 60% voted yes/no, then the voting ends. If the voting result is tight, for example in the range of 40–60%, then the rest 80% of validators vote in the second week. Votes in the two weeks are aggregated for the determination of the final outcome.

Validators chosen to vote are encouraged to vote with caution as their votes now carry more weight compared to the scenario where everyone votes (the smaller the voting group, the larger impact each vote has). Moreover, such a design saves validators from voting on every proposal and allows validators to dwell more on the proposals they are chosen to vote for. Many more proposals can be initiated more often and voted in parallel.

The voting design is susceptible to ambushes at the end of the voting period.

As the voting statistics are revealed in real time, attackers could hold their votes until the last moment and suddenly veto or adopt other actions to change the outcomes.

“Wait for Quiet” proposed by Dominic Williams [5] is designed to tackle the ambush vote. To illustrate the idea in a nutshell: the voting period is extended one more day if the result flips in the last hours so that others could have time to react, e.g. liquid token holders could stake tokens and vote.

Delegators, dApp developers and liquid token holders are underrepresented in the governance system.

As voting is time and energy consuming, most delegators would not vote for themselves [6] but rather inherit validators’ votes. The validators’ interest is then over-represented. Moreover, dApp developers and non-stakers/liquidity token holders/users are vital for the development of the Cosmos ecosystem but they are not well represented in the on-chain decision making. In the next section, I showcase the validators’ monopolistic voting power.

Monopolistic Validator Interest Group

As of the date of writing (30th July 2019), the top 20 validators own more than 70% of the voting power. Recently, one of the major validators brought up a proposal: Are validators charging 0% commission harmful to the success of the Cosmos Hub? This topic has been debated for several months since the launch of the Cosmos network.

Some validators, e.g. Sikka and SparkPool, charge 0% commission. The concern of this proposal is that other validators need to lower their commission rate to stay competitive.

As running a full node is costly, lowering commission rate means validators need to find some other profitable business to cover the costs of infrastructure maintenance and human capital. This will drive amateur validators out of the system. Over time, only big centralized exchanges, custodians and other crypto giants could afford to stay in the system. The entry bar will become high for new entrants and the system will end up with big crypto giants.

The concern raised in the proposal does not hold water as it appears.

First, only 13 out of 100 validators charge 0% commission and the voting power they own is less than 17% (see the figure below). It is far from alarming.

Second, do validators get nothing from providing staking service for free? By charging 0% commission, validators attract more delegated tokens. With more tokens backing them up, validators have a lower chance of being squeezed out by new entrants. Thus, more delegated tokens secure their validator identity. Validators earn reputation and other non-material benefit for being one of the 100 validators in the Cosmos network. Plus, validators gain more voting power as delegators inherit validators’ votes if they do not cast their own votes.

Third, even if the system mandates some minimum commission rate, say 1% instead of 0%, validators could still do kickbacks/rebates to delegators one way or another. What belongs to the market should be left to the invisible hand of the market. Too much unnecessary intervention, even with good will, will speed up the perdition of the project.

Fourth, imposing non-zero commission harms delegators’ interest. This might lower the competitiveness of Cosmos with respect to other cross-chain PoS projects, especially Polkadot. The annual real return for staking atom tokens is discounted by the token dilution rate. Namely,

Real Staking Yield = Nominal Staking Yield - Token Dilution Rate.

The current nominal staking yield is around 11% and the dilution rate ranges between 7–20%. That is to say, the real return of staking atom tokens is wearing thin if certain minimum commission rate is imposed.

Fifth, is the commission rate the only consideration when delegators choose validators? When delegators stake their tokens with validators, they share not only revenue but also risk with the validators. When validators breach the protocol, a fraction of the staked tokens will be slashed. Thus, the delegators will strike a balance between return and risk. Established validators could charge a certain commission for its reputation and still attract sufficient delegators, especially institutional investors with a large amount of atom tokens. They would diversify their positions to multiple established validators.

Last but not least, validators could stake their own tokens in the system to receive rewards that can cover their operating costs. A high percentage of self-delegated tokens also implies the validator has more skin in the game and shows the market its commitment to the Cosmos.

Not surprisingly, the proposal has won the vote in the end. The majority of the voters (or rather the majority of the validators) agree that charging 0% commission is harmful to the Cosmos network (or rather to the validators’ revenue). Isn’t it?

The author of this article, Yulin Liu (yulinzurich@gmail.com), currently leads research on token economics and governance system at DFINITY. He also serves as Affiliated Economics Professor at Huazhong University of Science and Technology. Yulin specialises in monetary theory, bank supervision, cryptocurrency, token economics and blockchain governance system.

He holds a Master of Science in Quantum Computation and a Ph.D. in Economics from ETH Zurich. Yulin was a visiting scholar at the European System of Central Banks and has been invited for talks at major central banks and conferences worldwide.



Yulin Liu
The Startup

research on crypto-economics and blockchain governance