Testnet Wave 2 Recap – The Tokenomics Suinami
During Testnet Wave 2, we focused on testing and validating Sui’s tokenomic model. We discussed some of the results in a previous recap article looking at network performance. Now let’s look at tokenomic performance.
Wave 2 offered Sui validators the opportunity to participate in a simulated environment resembling Mainnet’s economic conditions, providing us with an opportunity to analyze their response and behavior. The results surpassed our expectations, providing strong evidence that Sui’s Delegated Proof-of-Stake System and Gas Mechanism will work as intended.
Designing Sui Tokenomics
We designed Sui’s tokenomics to work seamlessly with Sui’s engineering design. Sui’s Delegated Proof-of-Stake provides the right mix of economic incentives to secure Sui’s scalable and parallelizable architecture. Sui’s Gas Mechanism keeps fees low as the network’s computing power grows horizontally in response to increased demand. Sui’s Storage Fund prices storage appropriately, so that the chain remains financially healthy when users leverage on-chain storage at scale.
Because of this design, Sui’s tokenomics deliver a great user experience, facilitate sustainable business models to network operators, and let SUI token holders enjoy the full benefits of a decentralized network.
When Rubber Hits the Road
Many of Sui’s core engineers love the show Formula 1: Drive to Survive due to its portrayal of world-class engineering talent, outsized ambitions, and community strength, reminding us of the commitment and values driving Sui. But as any Formula 1 team can attest, the pre-season months spent designing and building the car are for naught if the car does not deliver on race day. While theory and models are incredible tools, the moment of truth comes when the driver presses the pedal and the engine roars to life.
This same spirit carries through to blockchain development and testing. Testnet Wave 2 delivered a perfect opportunity to explore Sui’s tokenomics model and validate whether its theoretical design works as intended in a live environment.
Testing Incentives Without Financial Incentives
Testing tokenomics means tackling an important challenge: How to test economic incentives when the SUI token is not live? On the Sui network, the SUI token plays the role of aligning the views of network users, operators, and token holders. But why would these incentives kick-in if the token has zero financial value? Why would a validator care about how many rewards it makes if they aren’t worth anything? Why would an end user care about the gas price if gas fees are meaningless and there is a freely running faucet?
These challenges gave us an opportunity to re-think the testing of tokenomics in pre-Mainnet blockchain networks. We used this open canvas to create a testing environment in which validators were artificially incentivized to act as if the SUI token were live. Since this testing environment closely replicates the economic conditions that will be present once Mainnet is live, its outcomes should also closely match those of Mainnet. Hence, this experimental setting provides the first hard evidence of how Sui’s tokenomics work in a live setting.
The Validator Contest: Creating a Quasi-Economic Testnet
Testnet featured a contest played amongst validators and was based on two main assumptions:
- Simulated validator costs: Each validator received a CSV file with a time series of marginal costs, aiming to simulate each validator’s cost structure throughout Testnet. Validators were asked to act as if these numbers represented a validator’s marginal cost (in dollars) of executing 1 gas unit. The time series were distributed prior to the genesis event and each validator got a different set of numbers.
- Simulated SUI token price: Throughout Testnet, and prior to the beginning of each epoch, we gave the validator set a simulated value for the next epoch’s SUI token price in dollar terms. Validators were asked to act as if this value were real. All validators observed the same simulated SUI token price and the specific value changed from epoch to epoch.
These two assumptions played a critical role because they represented the missing elements needed for computing each validator’s rewards, costs, and thus profit margins in dollar terms. In other words, together with these simulated values, each validator was able to compute their simulated financials and gauge whether they were operating as a viable business model or not.
To see how this modeling works in more detail, let us quickly review Sui’s main accounting formulas. From a network perspective, total stake rewards come from two main sources:
StakeRewards = StakeSubsidies + (ReferenceGasPrice * GasUnits)
The second term on the right hand side corresponds to gas fees and this is where validators influence the network’s tokenomics. Specifically, validators directly affect gas fees by setting the reference gas price (RGP), and indirectly through the price’s effect on network demand. All things equal, a higher RGP leads to lower demand for on-chain activity and thus less executed gas units.
The following accounting approximates the bottom line, in dollars, for each validator:
Profits = (SUIPrice * RewardShare * StakeRewards) - (MarginalCost * GasUnits)
A validator’s share of stake rewards depends on several variables, such as the size of the storage fund, the validator’s percent of total stake, the validator’s share of owned versus delegated stake, the validator’s commission, and the output of the tallying rule (see Sui's Delegated Proof-of-Stake accounting). Importantly, the simulated values of the SUI token price and each validator’s cost structure are necessary to translate stake rewards and the execution costs into dollar terms.
Finally, the remaining open question was how to incentivize validators to take the simulated values described previously seriously. The assumptions let validators calculate their simulated profits – but, in practice, these numbers were made up, so why bother? To this end, we created a scoring metric to distribute points based on the following criteria:
- 2 points per epoch if the validator operated sustainably. Specifically, at each epoch close, the finalized data from StakeSubsidies, ReferenceGasPrice, GasUnits, and RewardShare became available. Subsequently, the formulas described previously were computed using this data and the assumptions on SUIPrice and MarginalCost. Each validator got 2 points if, and only if, they achieved Profits>0. Importantly, Profits were evaluated using the price the validator submitted to the Gas Price Survey (not the actual RGP). This was to encourage validators to estimate the most appropriate gas price for their operations.
- 1 point per epoch if the validator submitted a price quote below the RGP. Specifically, in the Gas Price Survey, each validator submits a quote for the ReferenceGasPrice prior to the start of each epoch. Sui’s Gas Mechanism collects these quotes and sets the epoch’s ReferenceGasPrice at the 2/3's percentile weighted by stake. Each validator got 1 point if and only if they were in the bottom 2/3's of submissions.
The profit criteria replicated a validator’s concerns with running a viable business. This criteria ensured that each validator had artificial incentives to accumulate more stake, gain higher stake rewards, and set a sustainable RGP.
The price criteria replicated the fact that, in a real setting, validators face the threat of external competition. If Sui’s validators set exorbitantly high gas prices and make large profit margins, then new validators would enter the validator set and seek to absorb some of these margins. Indeed, the most performant validators would likely aim to lower gas prices in order to weed out the non-performant validators. This criteria captured the threat of external competition and created downward pressure on the RGP.
Hence, the scoring criteria, together with the simulated values for the SUI token price and each validator’s marginal cost, created an environment replicating the major economic forces present at Mainnet. To further incentivize validators to take the contest seriously, the Sui Foundation promised that it would use successful participation in the contest as an input into its SUI delegation decisions.
Sui’s Reference Gas Price Delivers Low and Stable Fees for End Users
The Validator Contest showed that the Reference Gas Price works as intended and delivers low and stable fees for end users. Throughout Testnet Wave 2, validators reacted to movements in the simulated SUI token price and drove the RGP in the opposite direction. On net, changes in the SUI token price were offset with changes in the RGP and the fiat value of using the Sui network remained roughly constant.
The first epochs in Testnet Wave 2 were used to help onboard validators onto Sui's tokenomics and educate them on the staking and delegation mechanics. After this trial and error period, the RGP began working as a well-oiled machine by the start of the second week of Testnet.
To better understand the RGP in action, let’s do a mini case study around a period in which the simulated SUI token price fluctuated violently. Specifically, between epoch 16 and 18 the simulated SUI token price increased abruptly by 53% and subsequently crashed violently by 70% by epoch 21. During those same windows, the RGP moved in the opposite direction, first dropping by 12% and then rising by 122%.
On net, the price of gas in fiat terms only rose by 34% when the SUI token rose by 53% and then only fell by 33% when the SUI token dropped by 70%. In other words, the RGP successfully offset about half of the volatility in the SUI token price – and these were very severe movements. In the longer run, as the Sui network matures and longer time windows are observed, it is likely the RGP will provide an even stronger cushion to significant movements in the SUI token price. What is really remarkable about this case study is that the RGP achieved its goal while being set jointly by a distributed set of 41 independent validators.
Finally, an interesting feature of the RGP is that it is designed to handle variance across the validator set and ignore outliers. Since the RGP is defined as the 2/3's percentile of the Gas Price Survey, the majority of validators – indeed a quorum of validators – is willing to support gas prices below the RGP, while the remaining 1/3's of validators will not. This design ensures that both extremely low and high values do not affect the RGP and that a quorum of validators is always content with the RGP.
Indeed, while the maximum submission to the Gas Price Survey was often hundreds or even thousands of times higher than the minimum submission, the RGP mostly moved within a relatively tight window.
Creating a Resilient and Decentralized Network
Sui’s network security relies heavily on alignment between SUI token holders and network operators. Sui achieves this through a Delegated Proof-of-Stake mechanism whereby SUI token holders delegate voting power to Sui validators and the latter charge a small commission in exchange for this service.
Since Sui distributes stake rewards depending on performance, SUI token holders have a financial interest in staking with the most performant validators. Likewise, Sui validators are incentivized to perform well or else risk the double-whammy of losing stake rewards today due to bad performance and stake rewards in the future due to having less delegated stake.
Together with permissionless participation, what really matters for Sui decentralization is that SUI token holders always have enough candidates to delegate their stake to – and, specifically, that there are at least a handful aligned with each member of the Sui community. Two SUI token holders need not agree on who their best candidate is, but a SUI token holder can rest well if assured that they will have many good options to choose from. The threat of stakers switching delegation encourages validators to align with staker interests and operate efficiently. And, if no validator reflects the views of a segment of Sui community members, permissionless entry lets them join forces and enter as a new validator.
The most basic requirement for a decentralized future is for Sui operations to be able to handle sharp changes in the stake distribution. If SUI token holders decide to shift the stake distribution in order to better align validators with the Sui community, then it better be the case that the network can remain performant during these shifts. Testnet provided important evidence that this is true.
Throughout Testnet, Sui remained performant in the presence of a healthy stake distribution. On average, the minimum quorum required to process transactions consisted of 23 validators, indicating that Sui can successfully operate as a distributed system with a large number of operators.
More importantly, the distribution across Testnet changed in radically different manners, providing evidence that Sui is resilient to drastic changes. On one extreme, Testnet was bootstrapped with a uniform distribution in which the minimum quorum took its maximal value of 28 (i.e. 2/3 * 41). On the other extreme, there was one moment where the stake distribution was so skewed that the minimum quorum could be achieved with only 7 validators. In practice, though, to prevent the centralization risks that come with such a skewed distribution, voting power is capped at 10% and when a validator has stake share above 10%, the remaining voting power is spread across the other validators.
The fact that Testnet remained performant in the presence of these drastic changes provides comfort that SUI token holders will be able to support decentralization without affecting network performance. A win-win that hits both operational excellence and successful economic governance.
Frenemies Game: Testing the Network While Having Fun
The Frenemies Game was introduced in the latter half of Wave 2 with the goal of driving massive amounts of users to stake SUI. This helped test whether the network was able to handle such large amounts of delegations and withdrawals, and to see if the stake distribution changed in potentially unexpected ways.
One of the most interesting side effects of Frenemies is that it proved to be the great equalizer. By randomly assigning users to different validators, it became much harder for a validator to amass a huge amount of staked SUI, and thus tended to flatten the stake distribution. The average minimum quorum size increased from 20 to 27 validators, making Frenemies a big supporter of decentralization.
To our surprise, Frenemies surpassed even our greatest expectations! On-chain activity ultimately increased by over 700% by the time the Frenemies Game was shut down. And during that period, gas prices remained roughly constant in fiat terms.
Gas prices remained stable, by design, since Sui’s horizontal scalability enabled validators to parallelize transactions when demand increased. When activity increases, profit margins remain constant even if the gas price does not change since both operational costs and stake rewards increase linearly. In other words, Sui validators can operate sustainable businesses regardless of the amount of on-chain activity and without having to increase gas prices.
It’s very exciting to see the first indications of a network that can increasingly scale its operations in response to increases in demand for on-chain activity while remaining cost efficient. Ultimately, this delivers a good user experience to end users. No matter what is going on in the rest of the Sui ecosystem, the Sui community can rest assured that Sui will remain a viable financial option for them.
The Tokenomics Suinami is Just Getting Started!
These early results are very exciting. The Sui community is already enjoying the value of Sui’s object-centric model, parallelizable architecture, and accessible programmability. The Sui community will now also start reaping the benefits of an on-chain tokenomic model designed to provide a financial layer that is equally as performant as its engineering design.
We are extremely grateful to everyone that participated in Testnet Wave 2, and especially to the 41 validators that provided the first evidence of Sui’s tokenomics model. In particular, a huge shoutout goes to Sui’s gold medalist validators Stakin, DSRV, Passive Trust, Chain Node, and Staking Cabin who managed to get the most points in the Validator Contest. In addition, a special recognition goes to the silver medalist validators Crypto Coders, Everstake, Cosmostation, NodeInfra, and Daniel Moon, who together with the above made the top 10 leaderboard.
The engine is hot, the tires are soft, and the skid marks are all over. But we made it. Pre-season is behind us and the testing was a resounding success. Sui’s tokenomics are ready for prime time. We are super excited for what will come during the next iteration of Testnet and the highly expected Mainnet launch. We look forward to continue working with the Sui community and, especially, the Sui validators to take Sui’s tokenomics to the next level.
Disclaimer: Please note that all references to the SUI token price and its dynamics are fictitious and solely intended to provide context for the Validator Contest. All numbers discussed in the evolution of the Validator Contest are exclusively for testing purposes and should not be construed as having any relation to the future market price of the SUI token.