TL;DR Summary of the AI Litepaper
The AI Litepaper explores the potential to deploy Large Language Models (LLMs) on POKT Network
At present, Pocket Network Inc (PNI), the creator of the Pocket protocol and initial ecosystem around the protocol, does not itself run any Pocket nodes for profit. The company does have 30 actively staked nodes, but they are maintained for bootstrapping new chains. Given the current macro environment and our continual need to grow the company, it has become of the utmost importance that we focus on revenue generation. While we have many ideas and ongoing projects that we will share in the months to come, the most common-sense one to implement is to become a larger first-party consumer of our own technology.
The primary way that PNI raises capital is by selling tokens from our treasury, which are tokens that were allocated to the company in the Genesis block of our Mainnet launch in 2020. Alternatively, the company can author a proposal to the Pocket DAO requesting for a disbursement of funds. This is something we have yet to do, and would like to avoid doing for the foreseeable future, directing those funds to instead be used for contributions borne from the community.For those who may not know, the DAO obtains tokens by receiving 10% of the token rewards generated from all relays on the network, which, at the time of writing (Aug 11, 2022), amounts to ~150k tokens/day. Given the token’s present price (~$0.13), and the current reserves of the DAO (~106m POKT), the maximum amount of money that the company could raise today were the DAO to be fully liquidated would be ~$14m.As the current exchanges that the token can be found on are highly illiquid, and given that there is interest from investors, we are able to raise funds at a valuation above our current token market price. However, this is not a viable long-term strategy. To continue building within our current runway, and to avoid any possibility of having to use DAO funds to extend that runway, we are planning to run a pool of nodes, using the rewards to fund the company.
The strategy here is multifold, as it aims to solve for the following:
This first point is clear; we run nodes, we generate tokens, and we eventually sell those tokens to continue funding the growth of the company without relying so heavily on the DAO. This is akin to the strategy our node runners employ to fund their operations. There is no plan to sell any tokens in the near future, as we are well capitalized for the time being. The plan is to have these tokens available to us so that when the time comes, we are liquid enough to raise funds in a deliberate and thoughtful manner through our market makers. This is similar to secondary share offerings of public companies.
It is our plan to have PNI nodes make up at least 10% of the validators on the network.The security of a Proof-of-Stake consensus algorithm that is Byzantine Fault Tolerant is driven by two dimensions:
By securing 10% of the network, PNI increases the total value at stake while still limiting itself from being a major actor that could theoretically attack the network. As other node runners stake more POKT to earn more validator rewards, PNI can increase its own stake to maintain a 10% share. This creates a gamified feedback cycle that continues to increase the security of the network.You can stay up to date on the change and distribution of the validator distribution on Twitter @POKTValidators.
Running a large pool of validators exposes PNI to the same risks and rewards as other node runners and requires sophisticated tooling around monitoring and alerting. This will provide detailed insights into the resources required to run a validator node, which we will publicly share with the community and use to make decisions on where to focus internal resources to further improve the network.Presently, PNI is dependent on contributions from community members such as POKTscan or C0D3R to gain analytical insight into the top performing chains, average rewards, servicer geo-distribution and much more. While we intend to keep supporting and promoting external contributions, having the analytical data ourselves will enable us to draw additional insights and expose that data on the network.
We can stand up multiple nodes in multiple regions, and run a full set of performance testing and analysis without negatively impacting the community network. We can then use our internal nodes, to determine the peak throughput our nodes have on the network. With these metrics we can identify the full network's capabilities as we know the peak TPS (transactions per second) a node can perform (broken down by regions), and we know how many nodes are running in the network. This can give us quantifiable data on how over/under utilized the network is, and the head room we have for processing requests. We can use the results of these tests to spin up more nodes in regions that are needed, when utilization determines. All of this aids in a more efficient and performant ecosystem.
Our existing 30 nodes are split across two of the highest trafficked regions on the network: Singapore and Germany. These nodes are used to bootstrap new chains as we onboard them internally. With this initiative, the plan is to stand up node infrastructure across many more geographical regions so that we can onboard more new chains with high quality-of-service on day-one. This gives node runners across the community time to adopt support for each new chain, increasing its redundancy, reliability and latency. Doing it in this manner puts much of upfront financial operational expenditure on launching a new chain on PNI, as not every new chain that is allowlisted has enough relays at launch to provide justification for node runners to use one of their 15 node slots on the chain.
Finally, we’ll be able to share the core team’s best practices, so that node runners of all sizes are given the same baseline to start from. This is also important, as we periodically discover new infrastructure setups that may have material changes on network rewards. Giving all infrastructure providers the same information evens out the playing field.
In the coming months, we’ll begin ramping up the number of nodes we run. We are being transparent about it at the onset of the initiative so that there are no surprises. We believe in the success of a fully decentralized network, but to get there, we believe that we need to dog-food our own technology at scale so that we can ensure the security and growth of the ecosystem.