Appropriating Saito Routing Work for Incentivized Broad or Discriminate Fast Message Delivery
Routing Work
The Saito blockchain uses a measure of work which solves niche economic exploits inhibiting scaling and egalitarian consensus in traditional blockchains; it has the added benefit that it directs the efforts in competition for rewards towards the task of relaying data.
Saito routing work is an incentive metric which creates the dynamic where data with fees attached is relayed between nodes competing to claim a reward from those fees, who race each other to be the first to include that data into the blockchain and earn the reward stemming from those fees.
The protocol enforces the rules of this competition by paying out nodes in order of signage for each transaction which ends up on the blockchain. Every transaction will include a chain of signatures from nodes who marked their stake before relaying that data in hopes it is included into a block. Whichever augmentation of a transaction ends up in the blockchain will reward, exponentially in order of signage, the nodes associated with the chain of signatures imparted onto it. All other routing paths will be rendered worthless once some version of that transaction ends up in the block, creating a competition to get data into blocks as the consensus metric rather than hashing or staking.
Routing Utility vs. Block Utility
The blockchain is an inherently chunky medium of storage - accumulating, combining and publishing data in set intervals determined by block-time. While routing work serves to incent blockchain infrastructure to optimize itself, the naive use of a blockchain with self-optimizing infrastructure does not necessarily offer open access to that infrastructure in specificity; utility is limited to interaction within blocks even though infrastructure nodes have priority access to that data.
In seeking the use of these routing networks for optimized and discriminate transmission of data, it is helpful to first consider how one would motivate the node operators to optimally spread data indiscriminately, which happens to be an important security condition for block data in Bitcoin. Bitcoin miners already use, at best semi-centralized routing overlays1 - the security of these schemes is up for debate. If Bitcoin mining nodes vulnerable to selfish mining attacks began to relay their valid blocks as Saito transactions, competing routing nodes would spread the data in competition with each other; if any Bitcoin miners happen to be operating those routing nodes, they would be first in line for valuable Bitcoin block data which would allow them to spend less time mining on orphaned blocks.
The beauty in this is that those who value this data do not need to wait for it to be published into a block - by virtue of taking part in Saito infrastructure they gain privileged access to certain data as soon as it races towards their node, while keeping the security guarantee that nodes playing the zero-sum-game of information propagation will compete to do just that, counteracting parties which may have strategic reason to do otherwise.
Routing to Specific Targets
While launching one’s data as far and wide as possible is sure to land it in the intended recipients mailbox, it is less efficient and more expensive than using only the necessary routing nodes to get data only to where it needs to go. To achieve this specificity within the Saito incentive structure, a lock must be put on the data for which only its intended recipient has the key - a transaction which is only valid when digitally signed by its target, but which can accrue valid routing signatures along the way. Routing nodes with the ability to transmit towards this target race to do so such that their augmentation of the data is signed and published to the chain, unlocking their reward. Nodes which cannot access the recipient simply ignore it.
This allows the use of the underlying Saito infrastructure to extract utility far more specific and responsive than data merely waiting to enter a universal block. The discriminatory nature of the transaction narrows the set of network nodes capable of fulfillment, scaling the cost to something more comparable to centralized but direct service.
Incentives for Discriminate Message Delivery
Since a targeted recipient extracts utility from simply receiving the message and are a required signing party to pay routing nodes, they may as well in any isolated incident take that data and neglect to sign the transaction it was attached to, leaving its routers unpaid. As a router itself, the recipient will earn some money from forwarding it, though exponentially less than each node before it. Even worse than that, the recipient may take the data but sign the same transaction from a different set of routing nodes, perhaps ones they own themselves.
The simplest solutions involve introducing some trust: a recipient who denies payment to the most efficient relays eventually loses out on their service as unpaid nodes cease routing to that destination - as far at the routers know, they could be the slower relay, or the recipient could be scamming them; there is no difference from their point of view and their rational conclusion will be to eventually stop routing transactions which are not eventually profitable to them.
The primary issue is how new send/receive pairs can establish a reputation. A scheme which pairs network addresses to blockchain wallets is capable of bootstrapping these relationships: Receiving network addresses prove access to receiving wallets, and routers will know which half-signed messages to route to those recipients; Network addresses with wallets that routinely sign network funding transactions garner reputation making them more likely to be served later.
Users of this relay network which build their reputation by sending through their own nodes spend half of their money paying for blockchain security, making spoofing costly - the primary signal being sent by such behavior is that those users are willing to pay someone else, which is exactly what the reputation system needs to quantify anyways. On the other hand, users which receive a fastest relay from one set of nodes but choose to pay another set indicate to each relay path in the network their future odds of getting paid by serving that user and can adjust their prices accordingly. If a user attempts to get a discount by only paying a relay half of the time, that relay will, rationally, require twice as much payment.
Users who want the possible cheapest access to the fastest possible relays need only to pay them. If the relay network shifts in such a way that faster relay paths open up, those routing nodes aware of their potential to beat out the best path in the old configuration gain an increased incentive to try their hand anew at routing towards some specific target.
Optimization
Optimization has to do with the data payload. The need for block producers to process data into the chain is an potentially unnecessary cost associated with spreading that data. If the utility of that data is specific to one or a few parties rather than the whole network, the processing of it by block producers is energy expended after the message has reached its target and fulfilled its utility. This multiplies the cost of data.
Instead, routers can remove the data from the transaction completely and send a transaction with a fee to cover the routing costs for the full message, but only containing a hash of the message - the actual message is passed alongside the transaction between nodes until it reaches the receiver who may verify its integrity using the hash in the transaction. Only a small transaction of constant size must be included in the blockchain to distribute payment. The routers send the message data alongside the transaction, and the receiver can verify the hash matches what is included in the transaction; once the message reaches its receiver, it needs not burden any further blockchain network infrastructure.
Caveats
While the appropriation of Saito’s routing-work-motivated network does offer dynamic, open, incentive optimized relay network, it is not necessarily cheaper than the use of that same network infrastructure off-chain. This is because half of the block rewards in Saito will end up in the hands of miners or stakers - compensating routing nodes sufficiently means making an equal payment on top which pays for blockchain security. The benefit of paying the premium is that any changes in fastest routing path can be found automatically as the relay network retains attributes of the openness and universality of the greater blockchain. It may very well make sense for temporary private relationships to form around routers and patrons who discover each other through use of the relay network economics as optimal business partners.
The perhaps larger caveat is that, at the time of writing, this technique may not even possible on the Saito Blockchain. The delayed signature scheme requires a transaction to be fully signed after it has received routing signatures, which normally ‘sit atop’ a fully signed and valid transaction. It should certainly be possible to modify Saito to enable this behavior, and I am not personally certain it is impossible with today’s (Dec 7, 2022) protocol specification. The team, at the time of writing, seems intent on the development of more expressive scripting, though after a bootstrapping phase.
Conclusion
While Saito is mainly marketed as an economically viable and self-optimizing blockchain, it is the case that through user-level incentives and a basic transaction signing scheme that the network optimized for block production can provide utility as an incentivized relay network, for both open & efficient indiscriminate and discriminate messaging.
https://blog.marlin.org/are-centralized-relay-networks-a-suitable-l0-scaling-solution