Joseph Poon and Thaddeus Dryja published the Lightning Network (LN) whitepaper in January 2016. “The Bitcoin blockchain holds great promise for distributed ledgers,” the document states. “But the blockchain as a payment platform, by itself, cannot cover the world’s commerce anytime in the near future.”
Poon and Dryja use a well known comparison to highlight a possible problem in Bitcoin's future, the network's ability to handle volume. The Visa network achieved 47,000 transactions per second (tps) during the 2013 holidays, and currently averages hundreds of millions per day. Bitcoin supports less than 7 transactions per second with a 1 megabyte block limit.
“If we use an average of 300 bytes per bitcoin transaction and assumed unlimited block sizes,” the authors reason, “an equivalent capacity to peak Visa transaction volume of 47,000/tps would be nearly 8 gigabytes per Bitcoin block, every ten minutes on average.”
“Achieving Visa-like capacity on the Bitcoin network isn’t feasible today.”
- Joseph Poon and Thaddeus Dryja, The Lightning Network
The LN whitepaper proposed a solution based on a simple idea. It’s preferable to only have the bare minimum of information on the blockchain. Deferring transactions and net settlement to a later date could enable Bitcoin users to conduct many transactions without “bloating up the blockchain or creating trust in a centralized counterparty.”
The paper proposes using payment channels to offload transaction data. In simple terms, two parties open a payment channel, transact back and forth as much as they like, and then broadcast the result to the blockchain in one net settlement.
The are several development teams working to deliver LN solutions, some of which have agreed to a set of Lightning Network Specifications called Basis of Lightning Technologies (BOLTs). Much like bitcoin wallets, each LN solution will be a separate piece of software using the same network.
While none of the development teams have a complete, working example of LN client software, a developer’s alpha version called the Lightning Network Daemon was released on Jan 10th, by the original architects of the network, Dryja and Poon.
While we have yet to see what the application looks like, Poon feels that we’re going to find out pretty soon. “I think some of the UX will be worked out in the coming months, but we expect the payment flow to be simpler than bitcoin currently, and look a lot closer to paypal,” Poon told BraveNewCoin. “The seller generates an invoice, the buyer pays it, they get confirmation, and you're done.”
“To use Lightning you have to use software or merchants compatible with it. I expect backwards compatibility, though, so you can make on-chain transactions as well.”
The network will consist of nodes, just like the Bitcoin network, where hardware requirements are minimal. Poon suggests that they could run along alongside Bitcoin nodes, perhaps on a $5 Raspberry Pi, “for maximum security.”
The developer states that there will be several incentives for hosting a LN node: “For payment processors and exchanges, the advantages in giving instant payments gives significant business benefits, so I hope adoption will be fast, but that's up to the market to decide.”
Blockstream’s Rusty Russel told BraveNewCoin that he sees LN nodes establishing one to three channels initially, if they have any funds. If not, creating a channel might be the first thing they do when they receive on-chain funds.
There will naturally be few nodes keeping payment channels open initially, but those that do will be able to collect fees. Each LN node will charge a base fee plus a percentage, which is paid if users choose to route payments through that node. “Lightning logically charges by amount, unlike bitcoin which charges by transaction weight,” explained Russel. “I expect it to be somewhere between 0.1% and 1% of the total, plus a few satoshi base cost.”
The fee structure has raised concerns within the Bitcoin industry. Redirecting fees away from Bitcoin miners may weaken the Bitcoin network, some feel. Poon argues that the LN will compete with low value transactions on platforms that already use off-chain transaction solutions, such as Coinbase and Blockchain wallets. The LN “increases the probability people have direct interaction with the blockchain long-term.”
Both Russel and Poon point to an initial limit for a LN transaction, 0.042 BTC, making it suitable for micropayments and small purchases like lunch or a coffee. However, the limit will be raised as the integrity of the network is proven.
“For low-value the option isn't LN vs. on-chain, the option is LN vs. hosted wallets.”
The goal is decentralization, where every LN compatible wallet is a node. Once that happens, dedicated nodes may be replaced by smartphones, which is appealing to Russel from a radical decentralization point of view. “It's not crazy that cell phones become reliable peers in the network,” Russel said. “Coverage is near-ubiquitous, IPv6 offers simple direct connectivity, and bandwidth is fairly cheap.”
The recent alpha release of the Lightning Network Daemon included added support for Onion-Routed Micropayments, which encrypt payment details from end to end. The Onion Router (TOR) is a web browser that lets users surf the web in near-complete privacy, by hopping across a global network of other TOR users while masking IP addresses. The Onion-Routing code is from a LN development team at Bitfury.
Bitfury, Blockstream and Poon’s team are three of at least five teams today working on their own version of the LN. Although each of these teams are trying to fulfill their own vision of how it will work, their software is all open source, so they can mix and match the parts they want from each others' projects.
However, current implementations rely on a Bitcoin soft fork called Segwit. A rejection of the patch would force LN developers to rethink their respective approaches. In the meantime, LN is likely to be deployed to Litecoin first as the Litecoin Segwit soft fork looks likely to be accepted first.