This post provides a general overview of the economic model choice for the Biconomy Relay Network (BRN). See Introduction to the Biconomy Relay Network for an overview of the network’s intended purpose, and BRN Transaction Allocation Mechanism for an overview of network transaction coordination.
In the BRN each sovereign blockchain or rollup is considered an independent network. Nodes must bond, earn delegated stake, accrue rewards and fees, and experience penalties independently across chains.
The worst a node can do in the network is censor user transactions. The network needs only one non-censoring node to operate successfully, and therefore incentivizing a very high number of nodes to join the network is unnecessary. Due to this reason, we place greater emphasis on the lack of node work permission than node decentralization. However, node redundancy is desirable to prevent situations where existing nodes begin censoring and the network must wait for new nodes to join before transactions begin processing again.
Successful nodes exhibit liveness, superior transaction processing, and refrain from censoring transactions or breaking protocol rules. These factors determine a node’s earnings potential and ability to attract delegated stake. Our incentive framework is designed to heavily reward nodes that own a large share of the network. However, the penalties for liveness failures are swift and painful, and are designed to keep the network processing transactions–even with low node counts.
Delegators vote (using BICO delegations) for high earning nodes, which have demonstrated highly performant and reliable relay network services. As long as the node continues to perform (i.e., stays online executing transactions) delegators will continue to vote in their favor. This voting mechanism prioritizes rewarding nodes that have a history of reliably performing in the network, and ensures the majority of network transactions are placed in the care of such nodes.
A secondary benefit of delegated staking is to provide community members access to network revenue without them having to run technical infrastructure.
BICO is the work token of the network and ensures operational fairness of transaction management. A BICO bond is required for network participation, and ensures third party relay nodes are highly available and acting in adherence with network rules. Through the BICO token, network participants are able to earn network revenue proportional to their total stake (bond + delegated stake). All forms of transaction censorship are penalized by removing a small amount of BICO bond from the misbehaving node. Nodes that repeatedly misbehave are placed in jail and lose out on network earnings.
The primary objective of the Biconomy Relayer Network (BRN) is to provide optimal end user experiences. Requiring end users to pay gas fees (plus network premiums) in BICO tokens would be a sub optimal end user experience that would significantly impede our ability to compete with similar services that have no such requirement.
The key ingredient to providing magical end user experiences is highly performant and reliable infrastructure. In order to attract node operators to join and remain committed to the network, we must keep node user experience as a primary design objective. We will partner with node operators to develop a simple, yet thorough network specification. Nodes in the BRN facilitate transactions for end users by paying gas fees and ensuring transactions are included in the blockchain. If BICO were the payment token of the network (similar to LINK), node operators would be forced to continually swap BICO for blockchain gas tokens. This is sub optimal.
Nodes must bond 10,000 BICO to join a supported network.
Our primary concern is user experience, which translates to node liveness. As long as there is at least one node available and willing to execute transactions, end user needs will be served. Therefore, the network must be accessible to a wide spectrum of operators that can, without much friction, join the network and start executing transactions. Setting the bond requirement at an obtainable level is a key consideration.
But, what happens when a node misbehaves (i.e., censors)? We must have some monetary recourse for undesirable behavior. This means the bond must be set high enough that it hurts to misbehave—even for the wealthiest end of the infrastructure spectrum.
The amount of 10,000 BICO was selected to maintain a balance between affordability and network health. At $1/BICO a new node would need $10,000 to join the network. This is a considerable amount of money across many global economies, and a nontrivial sum for the most developed economies.
If a node voluntarily exits the network, a bond withdrawal delay period of some number of blocks is required in order to ensure the economic security of the network is maintained. We have not specified the exact number of blocks (it will vary across networks) a node must wait to be able to remove its bond.
- Unbonded: the node is not in the active set. They cannot relay transactions and do not earn premiums/rewards. They can receive delegations.
- Bonded: once the node commits sufficient bonded tokens they automatically join the active set and their status is updated to Bonded. They are executing transactions and receiving premiums/rewards. They can receive further delegations. They can be penalized for misbehavior. Delegators to this node who unbond their delegation must wait for the duration of the unbonding time.
- Unbonding: when a node leaves the active set, the unbonding of all their delegations begins. All delegations must then wait the unbonding time before their tokens are moved to their accounts.
Nodes require monetary incentive to show up and work. The primary source of earnings come from fees charged to end users/dApps. A premium, set by the network, will be charged on total gas cost (expressed as a gwei premium per gas unit).
|Chain||Native Token/USD||Avg Gas Used/tx||Avg Gwei Price||Premium||User Tx Cost Native||User Tx Cost USD||Premium USD|
The elaborated example includes gwei caps. For instance, 10 gwei on Ethereum. Meaning the user would be charged a maximum of 10 gwei in premium per gas unit.
Leaving the premium to be set by the protocol is a simple solution that doesn’t require additional mechanism implementations (i.e., those for finding market equilibrium fees). Further, it allows nodes to focus on competing on liveness, transaction management, and supporting a variety of use cases, instead of developing and managing pricing strategy. This lowers the barrier to entry, as price strategy is complex and sophisticated nodes will easily outcompete unsophisticated nodes. In addition, users (i.e., gas payers) do not have to develop and maintain pricing strategy.
There are tradeoffs of course. The utility of the network is user and use case dependent, and the cost at which a node can provide quality service varies across operators. In a fixed gas premium fee model it is left up to governance to set (and update) the network premiums to account for the variety of use cases and utility functions using off chain coordination. This job would be better left to a market auction.
Future versions of the network may attempt to incorporate market based fee mechanisms.
The BRN will pay perpetual network rewards as a means to ensure node engagement and service quality.
Over a one year period, the network should set a target node ROI (e.g., 10 percent). This amount can be higher than the market required risk-adjusted rate of return.
Using the target node ROI and some target number of network nodes (e.g., 10 nodes earning 10% in rewards per year) a base reward factor can be formed, which determines the amount of rewards the network should generate per second.
The base reward factor is expressed as an amount of rewards earned per BICO bond (i.e., 10,000 BICO). The rewards earned per BICO bond is adjusted up or down depending upon the square root of the number of nodes in the network. Again, at the target number of nodes, the network should be rewarding at the target return rate.
This model allows the total amount of newly issued BICO to fluctuate depending on the number of nodes in the network.
The base reward factor ultimately determines the reward rate for the network. Increasing this rate would incentivize more nodes to join the network.
In the below chart, we can see the max amount of BICO issued increases (in a diminishing way) as new nodes join. At 90 nodes, the network issues 30,000 BICO per year.
Rewards are paid per second, pro rata based on total node stake to all actively bonded nodes per chain.
Jailed nodes are not eligible for rewards.
Nodes can claim their accrued rewards at any time.
Note, after network bootstrap the rate of BICO supply increase can be set by governance and should be dependent upon the target number of nodes and expected node ROI.
There are ten nodes on one particular supported network. The nodes are as follows:
|Nodes||Bond||Delegated Stake||Node Network Ownership|
The network has set a base_reward_factor of 0.000001003 BICO per second. This translates to a base_reward_per_increment of 0.00003170979198 BICO per second. The result is a max issuance curve that looks like the following:
And a network one-year ROI curve that appears like the following:
The realized one-year ROI for individual nodes will vary based on their total share of the network, and the delegator reward share rate chosen by nodes. Here are the realized ROIs for each of the nodes and their delegators, given randomly selected delegator share rates:
|Nodes||Node Network Ownership||1YR Rewards||Reward Share Rate||Node ROI (before fees & expenses)||Delegator ROI (before fees)|
Note, delegator ROI from rewards alone is insignificant at this level. However, it is expected that nodes will share transaction fees, as well.
The above represents one network, and the BRN will have many such networks.
Also note, the reward rate will change as nodes enter or leave the network.
Paying a perpetual reward amount maintains some incentive for nodes to remain connected to the network in times of low network activity. BICO supply increase expands as nodes join the network, but is limited by the base reward factor. The DAO can easily adjust the base reward factor to control the token issuance rate.
Token issuance from network rewards adjusts to accomodate the number of nodes in the network. As nodes drop from the network, the overall token issuance rate decreases.
The reward rate is a concave function of node count, which means the reward rate does not scale linearly with the number of nodes joining the network. This is important as a linear function may provide incentive for nodes to attack other nodes in order to persuade them to exit the system, thus increasing the rewards for themselves. This attack is referred to as a discouragement attack.
Under this model, BICO supply will increase overtime. To counter this the community must find ways to introduce token lockups or burns.
Note, the network reward model was largely borrowed from PoS Ethereum. See here.
Undesirable behavior must be mitigated to ensure network health. Nodes will lose premiums and rewards for being offline (i.e., opportunity cost of missed future earnings). In the BRN, we cannot easily distinguish between relayer attacks on the network and undesirable behavior. This leads us to enforce penalties and not slashing. Slashing would be draconian, since an honest relay node that experiences a local power outage would have the same experience as a malicious node looking to damage the network or harm users.
Penalties, without slashing, should provide enough incentive for nodes to avoid misbehaving. Penalties in PoS Ethereum consist of a small amount of stake being taken away from the misbehaving node. This same mechanism is considered for BRN. However, a key component of BRN is delegation. It acts as a weighing scale, allocating BICO “votes” to the highest performing nodes. Substantial pain is felt from being placed in a penalty box (i.e., jailed), as delegators are free to move stake away from underperforming nodes not earning at their full potential. Lost delegated stake is not easily earned back and determines a node’s future fee income. Delegators are not directly penalized for node liveness failures.
Nodes will be penalized for undesirable behavior for each offense, and after a certain amount of lost bond the node will be jailed until they top up their bond amount. For example:
- For each liveness failure the node is penalized 2% of the required bond amount
- At 90% effective bond/required bond, the node is placed in jail
- After the
DowntimeJailDurationperiod has passed, a node must top up the bond to the required bond level and call
MsgUnjailfunction to escape jail. Note, the node earns no premiums/rewards while jailed, and thus delegators earn no fees.
- If the node fails to call
MsgUnjail, delegators would be forced to re-delegate their stake.
Introducing penalties disincentivizes undesirable behavior. By leaving slashing out of the penalty model, honest nodes are not over-penalized for honest liveness failures (some bond amount is lost for the honest node, but they remain online earning, which is most important to delegators). The possibility of jail time acts as a deterrent to repeat offenses, and protects network health by holding nodes accountable to delegators, as jailed nodes are not allocated transactions. Delegators are more likely to participate in the network, as there is no principal risk for them.
The BRN economic model is simple. Highly performing nodes gain a majority share in the network through delegator votes (i.e., delegated tokens). These nodes gain a disproportionate share of network fees and rewards, and should maintain their incumbent advantage. In order to ensure the network remains lively and reliable for end users, swift and severe (but not draconian) punishment is introduced for misbehaving nodes.