Economic Model Design of Optimism-Based Service Network for Decentralized Services
In this paper, we describe an economic model design for Optimism(OP)-Based service networks based on decentralized services. By examining some early blockchain economic model designs, we found a lack of good economic model designs on OP-based instant service networks in the market. This OP-based instant service application network will be the foundation of the next era of application blockchain.
1.Meaning of Token Economics
Mises  named the typical branch of knowledge in economics as “human behavior”, so the essence of token economics is that by setting up a set of rules, each actor in the network, by reacting to the rules, ultimately fulfills the vision of the project when it was created, and we can think of this process as a Markov equilibrium.
Since Bitcoin, we have seen countless projects created, and as we read through their token economics designs, we can deeply feel that each unique token economics is designed to solve an implied problem. For example.
- To address the endless inflation of centralized central banks, the aggregate constant model of Bitcoin.
- EIP-1559 was introduced in order to reduce the Gas consumption of users and to make it more economical to use.
- In the COSMOS network, the inflation ratio floats between 7% and 20%, corresponding to the relationship between the network-wide token stake rate and network security.
- Tezos’s baker is required to have a margin for each signature, which corresponds to the minimum percentage margin to prevent sybil attack. And in COSMOS network in order to prevent the attack of splitting nodes, the right-squared proportional penalty of proportional slashing is proposed.
- The Phragmén election algorithm in Polkadot, on the other hand, tries to solve the problem of electing a given set of people from a larger range of candidates and giving a defined power to the minority group supporters.
- The Graph determines the ratio of the utility of capital and labor in an economy by using the Cobb-Douglas function.
- The mechanism design of Wildfire in Arweave, which motivates the nodes with wide network transmission and fast node response to requests.
- The design of (3, 3) in Olympus, where only everyone participates in Stake to ensure the maximum benefit for all.
Depending on the type of project, and what the project hopes to achieve, there are a number of rules designed.
2.OP-based Application Network Model Architecture
Using aSearch as an example
Before building the network model of aSearch, we need to understand the value of the decentralized search protocol. Based on the protocol value, we can build a protocol model that meets the needs and the corresponding economic incentives to motivate each ecological participant to build it together.
Distinguish from the burning of the Library of Alexandria in the old days.
“They say that Caliph Omar, when consulted about what had to be done with the library of Alexandria, answered as follows: ‘If the books of this library contain matters opposed to the Koran, they are bad and must be burned. If they contain only the doctrine of the Koran, burn them anyway, for they are superfluous.’ Our learned men have cited this reasoning as the height of absurdity. However, suppose Gregory the Great was there instead of Omar and the Gospel instead of the Koran. The library would still have been burned, and that might well have been the finest moment in the life of this illustrious pontiff.”
The new era of information incineration has been carried out in another silent way, namely, message control.The message exists objectively on the web, but by managing all the channels such as: media, search engines. You will not be able to see the full extent of events. The monopolistic quality of search engines leads to its extremely poor resistance to censorship, and the decentralized small search engines cannot survive because the search business profitability depends on the advertising business, and currently people rely heavily on Google’s not being able to do evil for the orthodoxy of knowledge information.
Based on the above theory, a decentralized search protocol must satisfy the following requirements.
- The codification of the search index should be decentralized.
- The reading and delivery of search results is decentralized and unmodifiable within the protocol.
- The revenue settlement of the search should be directly related to the quality of the search service and not to the advertising business.
The architecture of the overall search network will consist of a data search layer and a data delivery layer, with a network settlement gateway based on user behavior.
The data search layer role is assumed by Miner, and the data transfer layer role is assumed by Router, the Fisherman and Arbitrator responsible for maintaining fair network revenue.
Miner: Miner is the node in the network that takes on the search function. Miner will run the client locally and provide the service by compiling the index.
Router: The Router is the role in the network that plays the role of search messaging. The router needs to use a higher server configuration to collect and distribute the search service. Router serves the role of market-based competition for the entire network.
Fisherman: Fisherman is the main role of maintaining the honesty of the network by sending multiple types of requests to get the data returned by Miner/Router and to determine whether the Miner/Router has committed fraud.
Arbitrator: Arbitrator is responsible for secondary judging of fraudulent nodes submitted by fishman.
It is worth noting that through Optimistic network design, the nodes will focus on searching business services rather than performing consensus chain synchronization, and the chain design will be difficult to carry high TPS decentralized services.
It is worth noting that by Optimistic network design, the nodes will focus on searching business services rather than performing consensus chain synchronization, and it will be difficult to carry high TPS decentralized services if it is designed as a blockchain.
3.OP-Based Network Economic Model Design
3.1 Model Design Basic Rules
Let’s start with some economic model design rules.
1.Incentivize network development according to demand, and do not expand the network in a disorderly manner without demand.
2. The model should feed back perturbations in the network. When token prices and nodes services change drastically, the model should always restore the overall network to a smooth state.
3. Maintain fairness in the network, giving different returns for different risks of capital, and giving different network parricipants a return commensurate with their technical capabilities.
Currently, there are about 1.5 PB of website data in the whole Internet, which is a very large number for a single query which . The content stored on Arweave is still very small compared to the entire network, and the demand for decentralized search services will depend greatly on the amount of valuable content in the decentralized network. In terms of web search demand, we see search demand as a function that gets larger depending on the amount of data stored, and likewise, the integrity of the search infrastructure promotes a boom in content.
Healthy protocol development should fluctuate upward, i.e., the value of the protocol fluctuates up and down around the network demand curve.
Early blockchain projects used an aggregate fixed token economy model, which was a good solution to the problem of low network value, but there was no way to solve the problem of the emergence of trading market perturbations. Let’s assume a scenario where there is no significant real demand for use in the network when the project is pomped by unsuspecting speculators, and the users who actually use the network will bear high fees as a result, leading to harm to the real providers of network value.2020 The attack on Chainlink in September 2020, where speculators made additional profits by Pump Gas. 
An inflation-capable network model would be well positioned to provide a solution where current holders of project tokens would have the ability to earn excess returns when speculative money enters the network, thereby reducing the harm of speculative trading behavior on the network’s current players.
In the economic model of aSearch network, excluding the Genesis tokens, aSearch network total token changes are divided into three parts: Inflationary reward (R_I), boost reward (R_B), and Burning.
Network augmentation is the revenue component of the aSearch network builder, and the amount is related to the Staking ratio.
3.2 Modeling Dynamic Equilibrium
Now that we have the overall idea of the token economy, we need to find some parameters to go better to help us translate the concept into a model. First we need to find the difference between the roles, and by acting differently towards the roles we will find the right parameters. We divide the network into users and builders. The difference between users and builders is whether the tokens are free flowing or not. The percentage of Staking in the network will be a good indicator to determine the percentage of contributors involved in the network.
The exact value is difficult to determine, the COSMOS network uses 2/3 as the ideal parameter, while Polkadot uses 50%，I used the golden ratio 61.8% to determine this, but of course this may require more research to determine the number in detail.This value is mostly used in blockchain networks to indicate the relationship between network security and liquidity, and in OP-type networks, although there is no need to mining, the stake rate still affects the penalty received by nodes for fraud. The following figure shows the relationship between node stake ratio and network inflation：
- When the stake rate < 61.8%, network inflation increases with stake rate
- When the stake rate = 61.8%, the network has the highest inflation rate
- When the stake rate > 61.8%, network inflation rate decreases with increasing stake rate
Next we define the annualized rate of reward of Staking at this stake rate, which represents our prescribed network growth rate of 10% per year.
In the above chart, the horizontal coordinate is the stake rate, the blue line indicates the trend of inflation with stake, and the green indicates the trend of annualized return with stake.
stake ratio x = total number of stake tokens / total number of tokens issued
Annual Inflation Rate R= (Year-end Token Issuance — Beginning Token Issuance)/ Beginning Token Issuance
Expected stake rate X_ideal
Ideal annualized rate of return of 10%
Ideal inflation rate Ri
Inflation rate R0 when the stake rate is 0
Decay rate d = 0.05
Inflation rate R1
Inflation rate R2
Inflation rate R
When the network is on the left side of the ideal stake rate, the increase in the network stake rate will increases the network revenue and therefore encourages users to stake token. When the network state is on the right side of the ideal stake rate, it indicates that users are willing to put tokens into the network with lower yield, which is usually caused by the high reward of the network services, and such high reward represent a break in the supply-demand equilibrium, so the model encourages new nodes to join the network to balance the network reward.
3.3 Accelerated Return to Dynamic Equilibrium
When there is a market glut, i.e. when the service far exceeds the demand, the network needs to return to its ideal state. And the only option for a single inflation network is for the token price to go down, which usually takes a long time, so all users in the network will lose confidence. At this point the network inflation reward is still gained by the node and their only option is to sell these assets, which is inappropriate for the other players in the network. Typically, it is the potential users that the network needs to incentivize, so it is especially important to build a trend of token appreciation in the network.
The use of a token burning model will better introduce a game mechanism than a single inflation model, where tokens are burned so that the reward on liquid assets is higher than that of assets locked in the network. The assets in the network will gradually be unlocked to liquid assets, reducing the network inflation rate. At this point, the token burn will be larger than the token increase, thus reducing the services scale and returning to the ideal state.
Tax_burn(the following will be represented by T_b) represents the burning tax rate, where users’ token payments will be burned for the tax rate portion and subsequently redistributed. The tax rate will be low in the early stages, providing an incentive for the network to grow, and when there is a surplus of services, increasing the tax rate through governance will increase the speed at which the network returns to dynamic equilibrium.
A mild inflation is a good mode of stimulus for any economy, but the nature of inflation is to dilute the value of the entire network and encourage the entry of new contributors.This is a secondary transmission, i.e., participation in the project contribution-for-incentive-reward rule in the presence of complete information flow.This reward is worth for the contributor’s investment in new labor in exchange for the reward until the capital return on labor returns to the social average.
Mild inflation is a secondary transmission, i.e., in the case of complete information flow, the network rewards all participants equally. This reward is not responsible for the outcome of the network services increase (profitable participants will not become new service providers directly), so we introduce an additional acceleration mechanism, named Boost Reward.
The condition for receiving the Boost reward is that when the network is at a stake rate higher than the ideal stake rate, the node needs to complete the task of query service increase to receive the extra boost reward, which will be given to the nodes entirely. This design can avoids complicity in gaining profit by limiting network growth. The details of the Boost distribution will be noted later.
The following function is the boost function, representing the inflation enjoyed by nodes at a certain stake value across the network.It is reflected in the chart below as a red line.
The boot equation is as follows：
The red line represents the BOOST yield curve, which represents the inflation of the network against the node stake volume at a certain staking rate.
When the stake rate is too high, the whole network will be in an inflationary state, but please note that as we said above, at this point the Boost reward will be fully rewarded to Miner and Router, and the user yield will decrease as the stake rate increases. It is a good choice for the user to sell token at a high price instead of staking it, which will bring the yield back up, and it is the best choice for the builder to build a new node to win boost reward at this time.
3.4 Market-based Design of Roles
A market-based incentive model will encourage service providers to provide better service, but there is a point to note here. Incentives need to directly incentivize single-factor behavior and not be coupled to each other. Take Arweave as an example: the miner is incentivized to increase the configuration of the storage nodes and cannot use the factor of the number of data calls. The miner’s hardware upgrade can provide more data storage, which will increase the number of data calls, which also leads the miner to store the data that is called more.
The marketability of services is a complex process and the incentive design cannot be in conflict with the final incentive outcome.
Let’s go back to our protocol to think about role specific incentives.
Miner, the miner is the core worker of the network. the marketization of Miner should avoid the marketization of the amount of query data, which will lead to the tendency of Miner to store data that is frequently called. The difference in revenue between different Miners will be correlated with three parameters: the Miner’s network environment, computing speed, and the amount of stored data.
Router, the main part of the network in the search protocol, is required to provide algorithm development, high performance task distribution, and a different price offer for Miner. Thus Router is a market-run node that can accept market-based operation strategies. And the user delegation is also carried by Router.
Fisherman,The fishermen network behavior is resource-consuming, so the fisherman is with extremely low access mechanism, and its gain is also dependent on the degree of network fraud .When the network is extremely honest, the gain of the fisherman will be very small, and when the number of dishonest nodes increases, the gain of the fisherman will increase with it.
Arbitrator, its benefits are related to the fisherman, but is an important part of the network security, it will have a higher barrier to entry and heavier penalties than the fisherman if it does fraud.
Therefore Router can embody the efficiency of the network, and the delegated behavior represents the work of capital exercising efficiency for guidance. Arbitrator can embody the justice of the network, and represents the work of capital exercising for network security. Therefore, these two role can accept user delegation.
How to avoid the formation of monopolies at the nodes of marketization?
In addressing monopoly, the usual approach is to limit its economic returns.
1) In terms of revenue distribution, the model of Bonding Curve is chosen to allocate revenue to users on a first-come, first-served basis, with a high weight of revenue for early stakers and a lower weight of revenue for later stakers. Therefore, users choose new nodes with high allocation weights although the labor coefficient is low, which also reflects the directional guidance role provided by capital in network construction.
2) In terms of network penalties, the nodes with higher power suffer more penalties, so users need to carefully consider their choices to select their trusted nodes.
3.5 Specific Distribution of Rewards
Well, the general framework of the token economy model about the network is done. Next comes the question of how to distribute it. In terms of distribution, we follow the principle of letting the role that really pays to enjoy the benefits, not all token holders. Bit with a clear distribution of the revenue, we have to analyze how much the workers have put in their labor. As already mentioned above, there are multiple forms of gains regarding running this network: Network Inflation Token Distribution、Boost Reward Distribution、Quiry Fee.
In the following, we will introduce these benefits in the context of revenue distribution
1.Network Inflation Token Distribution
Network inflation tokens are the result of increased network value , so this value-added needs to be given to the network according to the network contribution .The Delegators of Miner, Router, Fisherman, and Arbitrator themselves do not have a market-based difference in terms of value-added, so their revenue distribution will be based on the number of their respective Stake tokens. However, the Delegater’s behavior is actually a selection of the market, so the Delegator who chooses the better Router will get more revenue than the others.
The Delegator’s reward is related to two parameters, the stake volume and labor volume of this router node, and we can assume a cobb-douglas to handle such relationship, as follows equation：
Ri_i is the delegator’s reward.Qi/Q represents the ratio of labor to total labor at the this node.Si/S represents the ratio of stakes to total stakes at the this node.
Each node distributes the total Delegators’ reward according to the cobb-douglas, and the distribution within each delegator is done using the Bonding Curve.
2.Boost Reward Distribution
Boost is for the direct contribution of nodes’ efforts to the network growth. Miners and routers in the network distribute boosting rewards on the basis of miner:router = A:1. The setting of A is determined by the best Miner : Router ratio tested in the test network.
The boost rewards Miner can receive are related to several factors: node quality (response speed, etc.), size of data stored, and number of stakes. There will be a mechanism to score each Miner and finally the score will be used to allocate the Boost revenue.
Router’s Boost reward will also depend on the availability of the nodes as well, using the amount of search requests handled by Router in the last epoch and the number of staking to determine.
The decision of the query fee is determined by the Router, which provides an exact quote to facilitate the user to judge the Router’s service more intuitively, while the Miner has the right to decide whether to connect to the Router based on the price provided by the Router.
A portion of the inquiry fee will be used for tax rate burning and the rest will be distributed directly to Miner and Router, router will take the appropriate amount of money from each service according to the revenue sharing ratio signed with Miner.
The OP-based service network economy model we introduced achieves dynamic synergy between the rhythm of token issuance and network growth, and this game-filled token economy design makes players more willing to participate in network construction. Designing the protocol with negative feedback mechanism will effectively improve the robustness of the protocol.
 Ludwig von Mises — Wikipedia(https://en.wikipedia.org/wiki/Ludwig_von_Mises)
 Janson S. Phragmén’s and Thiele’s election methods[R]. Technical report, 2016.