On October 24, 2019, Illia Polosukhin joined us to answer our questions about NEAR Protocol, an end-user & dev-focused blockchain. Estimated to launch February 2020 with smart contract delegation, 0% slashing for downtime, 100% for equivocation & invalid state transitions. 3-day unbonding, 5% rewards, off-chain governance.
The NEAR team began with the aim of solving blockchain scalability, but quickly realized that solving scalability alone would be like building a huge stadium in the desert--lots of seats, but with nobody to sit in them. NEAR Protocol is designed to be one of the first operational, secure, fully sharded blockchains, and the team’s main focus is on developers and end-users. The design should make developing and using blockchain seamless, so that devs can build and deploy useful apps without knowing the mechanics of how the protocol scales.
The team is focused on the end-user experience. For example, NEAR’s protocol enables an app to pay the users’ transaction fees (similar to meta transactions), which opens the potential for new business models and user interactions. Non-crypto users will not need to immediately understand or manage private keys, but how? With an in-browser, “progressive security” onboarding, and the ability to increase security as required.
Stardust is migrating to NEAR from Loom, due to issues with scalability and usability, both of which being barriers to onboarding users to Stardust's platform. Flux.market has had similar issues with Ethereum, particularly with usability, since prediction markets interfaces are difficult to use. Joey Krug apparently mentioned that Augur could have been built 3x or 4x faster had they had NEAR’s tools, and Illia thinks there’s a lot more blockchain usage to be had before requiring off-chain processing.
The team is focused on the dev experience, with support for TypeScript (for open web apps) and Rust (for financial apps). NEAR enables app deployment in minutes to hours with an online IDE, and the team provides high-end group chat support for early teams building on NEAR. Finally, they're focused on good documentation and tooling, such as a block explorer and wallet.
In the short term, NEAR's design space should look similar to that of Ethereum. The team has seen some DeFi applications, lots of gaming, and a few DAOs ( 3 or 4) already being developed. Sasha told me that NEAR has decentralized transcoding services, prediction markets, gaming marketplaces, decentralized insurance, and crypto payouts for game tournaments.
For the long term, Illia is excited for building services and applications that string together to form the ‘Open Web.’ For example, he likes social platforms like Instagram, which is currently a permissioned and rule-changing platform. In future, new businesses may be enabled to unlock and bootstrap novel creations with an open network of social graphs and content. Sasha discusses that vision here.
As with most blockchains, ownership of the network will essentially be ownership of the NEAR token. We do not yet know the proportions of the network that will be owned by stakeholders such as private investors.
NEAR will be staked to provide security, and used to pay transactions and state fees. The team expects the token to be used as a store of value and as collateral for various (DeFi) financial applications.
New NEAR tokens will be minted with the production of every block. The rate of inflation is designed to vary based on a rewards rate target of 5%. What does that mean?
Stakers should expect 5% rewards annually, as a combination of block rewards (inflation) and transaction fees. So when usage (and transaction fee volume) goes up, rate of inflation goes down, and vice versa. This is meant to solve some of the problems that Bitcoin has with scheduled inflation reduction, which Illia believes at some point will incentivize miners to steal transactions from other blocks, based on research that he cited in their economic paper.
So far, sadly we have not seen any network, including PoW, that actually improved decentralization over time (after initial spike at the start). There has been some research on centralizing powers of PoW (Ethereum's Gini coefficient went up in last years) and especially PoS.
The team is, however, aiming to have an initial distribution that includes many participants globally. Staking rewards will be taxed by the NEAR Foundation (more about that later), and redistributed as grants to encourage new participantion.
What about token-holder and staking centralization?
We are using a lot of market dynamics around validator participation to prevent "rich get richer disproportionally" issues that most PoS have.
Token-holders that want to stake will rely on smart contracts to delegate their stake to validators, since NEAR doesn’t have in-protocol delegation. Smart contract delegation has the potential for both innovation and challenges. Validators will likely be the ones that create these delegation smart contracts, and Illia is expecting to see examples of delegation smart contract by the end of Stake Wars, NEAR’s incentivized testnet. While I find the innovation potential particularly exciting, I think smart contract security will be an important part of safeguarding delegator funds. I expect that we’ll see delegation smart contract standards emerge fairly quickly.
Delegation smart contract innovation could enable different risk profiles, whereby some delegators assume more exposure to the slashing risk and a bigger share of the rewards. Some validators may choose to offer their delegators some liquidity. The point here is that the mechanisms and parameters I’m about to describe apply to validators, but may be different in practice for delegators.
Time periods are described in 12-hour increments called epochs. Staked tokens don’t begin earning rewards immediately. It will take between one and two epochs (12 - 24 hours) to begin receiving rewards for newly-staked tokens, depending upon when the staking transaction is sent and how much time is left in the current epoch.
Unbonding staked tokens will take between two and three epochs (24 - 36 hours), also depending upon how much time is left in the current epoch, and rewards will stop accumulating during the final 24 hours (ie. 2 epochs).
NEAR Protocol won't slash validators' stake for being offline. Instead, the offline validator will earn fewer rewards in relation to their uptime, and eventually be removed from the validator set after reaching a 10% uptime threshold (ie. 90%+ downtime).
Equivocation (ie. double-signing a block at the same height) and invalid state transitions (ie. producing an invalid chunk) will result in a 100% stake slash. This is important to note. It means that a validator that breaks one of these protocol rules will lose everything they have staked, including all delegated tokens.
This looks harsh, and maybe it is. Here’s the reasoning. By not slashing validators for being offline, the protocol shouldn’t need specialized, highly-available infrastructure with a private key copied into multiple data centres, and thereby greatly reducing the risk of double-signing. The team believes that a private key should never move or be copied. That's why NEAR will support many keys per account, including a separate key to control staking.
NEAR’s validators will be randomly selected and assigned into 1) block producers and 2) hidden validators each epoch. To join the active set of validators, a validator-candidate must send a staking transaction with enough stake to win a seat. If your validator gets selected at the end of Epoch 1, you won’t be validating until Epoch 3.
Illia’s example: if you stake at 9:00 am, you need to first wait until 12:00 pm for the selection process, and you’ll start validating at midnight.
There are 100 validator seats per shard, so the number of validator seats will multiply by the number of shards that the network requires for processing and storage. You can read more in-depth here. The validator seat price (amount of stake required to become a validator) is determined by an auction mechanism that enables the market to determine the optimal security:reward ratio.
Your stake determines your number of validator seats. If you have 10x the stake required for one seat, you’ll need to run more nodes to process more shards in parallel. This incentivizes 1) professional staking services to run good infrastructure and 2) many nodes (proportional to the number of validator seats). The team has designed NEAR to target support from a few large, professional participants that control a large number of seats, followed by a long tail of many, many nodes that each control one seat.
A validator with a strong reputation across multiple networks is always an attractive candidate for me. Beyond that, you'll want a validator that maximizes uptime without risking the possibility of breaking protocol rules by double-signing or committing an invalid state transition. Most interestingly, innovative delegation smart contracts that offer delegators benefits such as different risk profiles, liquidity, or some other value-add could be attractive.
NEAR will start with what Illia called a “somewhat limited foundation.” Following governance experiments, foundation council membership should open up to diverse stakeholder representation, including validators, developers, users, and token-holders.
There are a number of the goals and principles that the NEAR team is following. Illia presented these at the latest DevCon, and we're anticipating a follow-up article on this topic from the team.
For upgrades, the team wants to begin with a model similar to that of Ethereum, while having one entity responsible for research and reference implementation (removable if found to be malicious or non-adequate). Improvement proposals (NEPs) and contribution would still be an open process.
The plan is for new network software to be backward-compatible, so validators can operate the new software prior to upgrade, and the network will only upgrade once a majority is running the new software. In a ‘reference maintainer’ model, the reference maintaining company would propose changes and encourage validators to upgrade.
Staking rewards will be taxed by the NEAR Foundation to fund ecosystem development and token redistribution. The team is aiming to have governance proposals distributing these funds eventually, but they do not think on-chain governance has matured enough yet. Their plan is to test governance ideas with increasing amounts of value and impact after mainnet.
The team regards two potential standards that they think the industry will converge upon in next couple years: 'app chains' (eg. Cosmos) vs. 'dapps' (eg. Ethereum) on shards, with NEAR clearly being designed for the latter.
The NEAR team is actively working on interoperability--an Ethereum to NEAR bridge that enables full smart contract interaction.
Felix (chorus.one) asked Illia about his thoughts on NEAR's security in the case that derivatives of staked NEAR tokens are used for a DeFi protocol such as Maker.
Great question, yes lots of financial security risk especially if large adversarial players (like Goldman Sachs) come in. The sad point though is that this is coming--I know 3-4 different parties who are working on this right now. Some of the exchanges are already offering it as well. So I prefer to have this on-chain and traceable (transparent) to make sure everyone is clear what is the security versus exposure of derivatives on this security.
I think the most interesting use-case is using stake as collateral for a stablecoin, because it creates a very interesting new value chains for validators.
Currently the NEAR team has some known issues that they are ironing out before mainnet launch. The goal of Stake Wars, NEAR's incentivized testnet, is to get lots of nodes and validators to participate, and to test everything thoroughly.
Who should participate in Stake Wars?
When mainnet? Sasha estimates NEAR Protocol will launch in February 2020.
Thanks to our Staking Hub community for the thoughtful and impactful questions that inspired high-quality answers. Since you’ve read this far, you might as well join us over in the Staking Hub Telegram channel
Hopefully you found this useful. Feedback is always welcome! I’m on Twitter.