TL;DR Summary of the AI Litepaper
The AI Litepaper explores the potential to deploy Large Language Models (LLMs) on POKT Network
Pocket Network v0 has developed a balanced and incentivized system between three key elements: applications, nodes, and the network layer. A report from Messari highlighted the technical architecture of Pocket as a Web3 infrastructure protocol. This architecture has enabled exponential growth in data relays on the network. Make sure to see Part 1, Part 3, and Part 4 of this series for more of Messari's insights.
In Pocket’s v0 structure, the network acts as the intermediary between the demand (developers/applications) and supply (nodes) sides. Developers want to connect their applications to blockchain data in a reliable way. Meanwhile, Service Nodes receive and fulfill those requests to relay data from supported blockchains back to the applications.On the demand side (demand for data relays, that is), applications benefit from access to a decentralized, cost-effective, efficient, and reliable network solution that evolves towards near-zero cost basis over time.On the supply side, Service Nodes currently need to buy and stake at least 15,000 POKT to serve applications with data relays. The optimal strategy in v0 is for nodes to fulfill as many requests as possible by spreading their POKT across multiple nodes (rather than piling as much POKT as possible into one node). That being said, if node operators do decide to increase the amount of POKT staked in a certain node, and in turn qualify themselves among the top 1,000 nodes in terms of staked POKT amount, they can take on a Validator role and earn additional block production rewards.Pocket’s network layer creates the rules and incentives systems that are in large-part based on the native token - POKT - which benefits both the supply and demand sides. This network layer conducts “service node coordination, reward and punishment mechanisms, and dispute settlements,” among other functions.
With Pocket, application developers don’t face upfront costs to access RPC endpoints in the Pocket Portal, but will be able to subscribe to a stake-to-use model based on POKT. In other words, developers can stake certain amounts of POKT for the right to have certain amounts of relays serviced by nodes in the network.Moreover, Pocket offers a free tier, where applications can enjoy up to 1 million daily relays, offering a similar capacity to centralized and paid alternatives.By subscribing to the stake-to-use model, applications can benefit from more power based on their total POKT holdings, while the new Pocket v1 (more on this in a future edition of Deep Pocket) will offer more onboarding benefits for applications.
Pocket uses a “cherry-picking” model to ensure it selects nodes that are “synced to the chain’s current state and operating on low latency,” and matches these nodes with applications to fulfill their data requests.When applications request data from blockchains via the Pocket Portal, the network “dispatches that relaying request” and “initiates a session between them and the service nodes.”In one session, Pocket matches applications with 24 Service Nodes, which serve the application for roughly one hour before the matching process starts again. Service Nodes complete the data relays, while Validators confirm the “proof of relays,” leading to valid transactions on the blockchain.
With Pocket v1, the validation model will be overhauled, leaving nodes to grow their focus on the quality of service while improving the overall efficiency of the ecosystem. Again, we’ll have more on v1 in a future edition of Deep Pocket.
Pocket’s incentive system distributes rewards for valid relays. New POKT tokens are minted in proportion to the number of successful relays that nodes complete.The current monetary policy allocates 89% of rewards to Service Nodes, 10% to the Pocket DAO, and 1% to Validator Nodes.Pocket started its journey under a Bootstrapping Phase, focusing on deploying solid infrastructure and guaranteeing a reliable service for applications and node runners. The protocol focused on attracting new applications to the ecosystem (by offering a thriving network of active nodes) while also offering an attractive reward model for the node runners servicing relays and providing blockchain data for applications. During this stage, early node runners received particularly high rewards in POKT, given that huge relay growth largely outweighed rising token supply pressure.As Pocket enters a new stage, its inflation model (the rate in which POKT is minted) is undergoing changes to keep a focus on long-term sustainability. As more applications and node runners entered the network under the Growth Phase (which followed the Bootstrapping Phase), Pocket’s inflationary economic model required some adjustments to ensure a long-term sustainable path. Important proposals were introduced to tackle the expected inflation resulting from the growth goals of the protocol.
In the PUP-11 proposal, a plan was outlined for an annual inflation rate - Weighted Annual Gross Max Inflation (WAGMI). By setting an annual inflation goal, every ecosystem party is aware of the inflation rate, growth path, and outlook for the project and its native token.With PUP-13, a follow-up proposal to PUP-11, the mint rate was further defined to guarantee an eventual 50% annual inflation rate, based on the total POKT supply at the time of the proposal passing. The proposal not only defines the PUP-11 WAGMI rate at 50% (achieved via monthly step-downs in the mint rate), but establishes a clear path of parameter adjustments to reach that target and leaves options open for more changes in the future if the community reaches a consensus.This change to the inflation rate will allow participants in the ecosystem to have sustainable benefits in the long term, while reducing the rate of POKT rewards per relay and controlling inflation much more effectively in the short to medium term.Overall, the native link between newly minted POKT, its potential return, and its use-case as both a demand-side requirement for service and the reward vehicle for nodes, leads Pocket’s infrastructure to grow alongside the evolution of its token.
Beyond the incentive system, Pocket’s friendly onboarding strategy for new chains has contributed greatly to the network’s growth.Over time, Pocket has integrated with Layer 1 networks like Harmony, Solana, and Fuse, while also supporting the Layer 2 emergence via partnerships with projects like Polygon, Gnosis - xDai, Boba Network, and others.Pocket’s exponential growth in daily relays comes from this effective onboarding strategy and an increasingly global network of nearly 50,000 nodes that provide reliability and comfort for developers. As a result, Pocket has recently achieved 1 billion daily relays, growing at a breakneck pace from ~100 million average daily relays in November 2021. Still, compared to centralized alternatives like Infura and others, serving tens of billions of calls per day, Pocket has a lot of room to grow (and a lot of market share to capture) in the long-term.Furthermore, Pocket’s upcoming v1 will propel the network to a new level of maturity, growth, and quality of service, in order to become the go-to global node service for blockchain data relays.Learn more about Pocket’s tremendous growth from the full Messari research piece.