Staking Design for Resource Networks
Universal challenges and solutions. (Speaker: from NuCypher)
Thinking about decent access control: for key or data management, messaging.
The basic dichotomy we think about is: do you want access controlled by third parties, or by cryptography (threshold proxy re encryption) and decentralization? [w]
what’ the problem with TLS? Say Alice encrypts things, decrypts on decent storage w/ untrusted machines…
proxy re-encryption: as before, the data owner encrypts data w/ public key and stores it elsewhere. Sends ciphertext; Bob wants to decrypt, but can’t. Her device generates a reencryption key (pairwise w Bob) a proxy machine reencrypts the message for Bob, the result is a ciphertext for Bob. [the reencryption key is unique to A-B and doesn’t allow reconstruction of A’s key]
Decentralizing this: ‘threshold’ proxy re-encryption: the reencrypt key is split to many proxies, Bob reconstructs it on his end. “To allow hit to get the data he deserves” [-- shirt slogan?]
I’ll use the word ‘Worker’ now for the proxies; also use the terms ‘staker’, ‘node’, ‘Ursula’ in our docs.
Worker duties include access control; secure management; availability.
Protocol building blocks
token distribution process: min stake size
basis for earning rewards: supply growth rate; reward adustment; reward coefficient
payment channel, fee structure, sampling:
grounds for punishment(!): calculation
[elaborate whole-system diagram w feedback loops]
Problems + Solutions
Problems: Disconnect b/t work and reward; monitoring worker uptime; overhead of on-chain payouts; enforcing reliability
Potential solutions: Worklock distribution model: incentivize stakers to also do work on confirmation. Correctness via ZKPs.
Competitive advantage, since tools are open source, are: worker network.
Decryption/reencryption market price is currently $0.03/10k requests.
Selling points: uncensorable, trustless, edge cases such as ‘disappearing Alice’ are not straightforward to price. Critical aspect of service appeal are determined by network effects.