by Evan Sixtin
Four cryptocurrency researchers from UIUC, Cornell, Microsoft, and Newcastle University released a paper on February 19 describing a solution for scaling blockchains using state channels which in theory would be faster than the Lightning Network model. They are calling it "Sprites."
Presently, Bitcoin is approaching maximum capacity and might soon be unable to support the amount of transactions that users are requesting. One strategy to solve this scaling issue supports increasing the block size or making block size unrestricted. While this solution is supported by many, it has been criticized by others, stating possible negative effects on consensus and security.
Another type of proposal for improving the scaling of cryptocurrencies, called "Lightning Network" (LN), describes a network of off-chain payment channels which act like credit lines. In this case, on-chain currency would be used as collateral to secure these off-chain credit lines. This method would greatly decrease interaction with the blockchain directly, decreasing the amount of on-chain activity, and allow for a significant increase in user transactions per second since most of the work would be done in state channels.
Here is a description of Lightning, as described in an interesting Medium article written by “Cryptoconomy” on the benefits of LN:
"The Lightning Network operates by building secure payment channels between users, services, nodes, and any other participants on the network. With these channels, funded with a specified amount of bitcoins, the two participants can make a vast number of instant transactions that essentially occur "off-chain" with the provable, on-chain guarantee that the bitcoins are redeemable. In addition, the channel can be left open without having to trust the service, user, or node that you are connected to. No one has the power to take or move your coins without your explicit consent."
The article goes on to explain how LN has the ability to replace custodial control of funds by cryptocurrency exchanges with Lightning nodes, increasing security fundamentally. In addition, no custodial control of user funds means there is no liability and no need for regulation of such exchanges.
However, LN also has drawbacks. It works great with rapidly resolving payments, this is when payments complete quickly, and only require off-chain point-to-point messages. But when a dispute occurs, a timelock is executed and results in delays. These delays are magnified by the length of the payment path; the more hops, the longer the wait. Funds held in escrow for long periods result in collateral costs.
This is where Sprites come in. Sprites present an improved construction of payment channels that directly improve the worst-case collateral costs compared to Lightning by using a global contract mechanism. From "Sprites: Payment Channels that Go Faster than Lightning":
"We create a global contract, called the PreimageManager (PM), which records assertions of the form ‘the preimage x of hash h = H(x) was published before time TExpiry.’ We then augment the payment channel construction with a ‘conditional payment’ feature, where each party in the path reserves conditioned on the presence of a record in this global contract."
Because of the limited programming language used in Bitcoin scripting, Sprites will not work directly with the Bitcoin blockchain at this time, due to the inability for a transaction to depend on a global event recorded in the blockchain. Sprites will currently work with the Ethereum blockchain, and any other chain with a scripting language which would allow for the global contract mechanism that Sprites relies on to make its model work, but the research paper mentions only Bitcoin and Ethereum chains.
"In the decentralized ideal, users would establish channels with peers in their social network (i.e., forming a scale-free network structure). However, worryingly, high collateral costs associated with long payment paths may create an economic pressure towards a more centralized structure, with most individuals forming channels with only a small number of well connected bank-like hubs. By reducing the collateral cost, our work takes a significant step towards realizing the vision of decentralized credit network."