The Hashgraph distributed consensus algorithm was invented by Dr. Leemon Baird. Dr. Baird received his Ph.D. in Computer Science from Carnegie Mellon University. Together with his longtime friend and colleague Mance Harmon, Swirlds Inc. was organized in 2016. Baird and Harmon were both former faculty members at the US Air Force Academy program for Computer Science and Cybersecurity. During that same year, Swirlds published its White Paper.
The vision Baird and Harmon shared for Swirlds was to enable developers to create distributed applications with unlimited scope and scale. Leveraging the Hashgraph distributed consensus algorithm, developers can now build trusted applications that are always available, without the use of centralized servers. Applications built on the Hashgraph platform are fair, blazing-fast, and achieve consensus quickly, giving the user 100% certainty in the consensus order.
Hashgraph is blockchain on steroids. It uses two techniques to ensure fast, fair and secure consensus. The first protocol is “Gossip about Gossip.” A piece of gossip contains a snippet of information that is provided to the hashes the last two participants talked to. Each node has the ability to randomly share transactions, events, or any other kind of signed information with others. Then, the node that has received the message converts it into a new one and sends it to other random nodes, which repeat the action until all of them get the information. In this way, all nodes receive and remain aware of the constantly updated information.
According to Dr. Baird, “…
I can guess how you would vote, but you don’t need to vote, so you reach a consensus for free. It’s the fastest way known to humanity to send information.”
The second technique incorporated into Hashgraph is “Virtual Voting.” One can easily calculate all votes if they are divided into rounds. This is what virtual voting ensures, as it is conducted in several rounds. It is quite easy to find out which node will vote because if new information is provided, all nodes immediately know the initiator. This can be used as input to the voting algorithm to find out which transaction has reached consensus. Virtual voting makes it possible to define which transactions are valid and which are not. If a transaction has two-thirds of the node in the network as witnesses, it is valid. They have to confirm that they are connected to these nodes.
Hashgraph was created to be Asynchronous Byzantine Fault Tolerant (aBFT). The aBFT protocol makes Hashgraph the gold standard of security and ensures the Hashgraph network will continue to operate even if a third of the network nodes are corrupted and malicious. According to Dr. Baird “… it means that when you are trying to figure out the order of transactions, there comes a time when you know that you have reached consensus.
Ultimately, aBFT means three things:
1.) we are going to come to consensus,
2.) we will let you know when we have come to consensus, and
3.) we’re never wrong, you are mathematically guaranteed that everyone else is going to reach the exact same consensus.”
Distributed ledger technology has evolved rapidly during the past decade. There are two primary and competing forms of DLT. The first and original DLT technology is Blockchain. The second and more recent version of DLT is based upon Directed Acyclic Graph. The genesis of Blockchain dates back only 10 years to the launch of Bitcoin in 2009. Because it was the first and most widely known, Bitcoin has become synonymous with Blockchain.
During the years following Bitcoin’s launch, other forms of Blockchain have been introduced. Litecoin presented in 2011. Ripple was launched in 2013 and Ethereum was launched in 2015. In 2016 the Linux Hyperledger Foundation was established to develop the Hyperledger Fabric Blockchain. Blockchains, as the name suggests, are ledgers composed of a chain of blocks, reaching right back to the first (genesis) block and each block containing a number of transactions. New transactions are broadcast to the network by a submitting node and remain pending until selected for inclusion into a new block by a miner node. New blocks are produced at intervals by miner nodes, attached to the current head of the chain with a backward-linking hash of the previous block and broadcast to the network.
Blockchain technologies such as Bitcoin are slow by design and provide high probability, but not 100% certainty.
Directed Acyclic Graph (DAG) is the second and more advanced form of distributed ledger technology. DAG based distributed network solutions began to emerge in 2016. Hashgraph is the most advanced and widely known distributed solution based upon directed acyclic graph technology. Other DAG-based DLT solutions include IOTA, Byteball and Nano. DAG data structure is a mesh of events, each event composed of one or more transactions and linked to two prior events by hashes of those events, similarly to a blockchain. Prior in this instance means previously in time at the event-issuing node, which due to network latency and simultaneity concerns may not necessarily be the same ordering as seen by other nodes, which implies the need for an additional process to extract a consensus order. This mesh forms a Directed (backward in local time) Acyclic (no loops) Graph (a mathematical construct consisting of vertices (events) and edges (backward-pointing hashes).
In contrast to blockchains, DAGs progress in a more granular and parallel fashion an event at a time and although an event may contain a number of transactions, in some cases this may be just a single transaction or a metadata update. With DAGs there can be multiple concurrent events emitted by active nodes, not effectively limited to a group of miners with massive compute (hash-rate) power. The network as a whole eventually determines the order. Various DAG algorithms make different choices regarding how to reach this distributed consensus, how fast, how efficiently, and how deterministically that happens. They also make varying choices of the information about the events recorded into the DAG. Some broadcast events to the whole network, others are more point-to-point or use a group of trusted witnesses to validate the growing DAG. DLT solutions based upon DAG are more scalable and fairer while offering much higher throughput, lower latency, and higher certainty and security than blockchain-based DLT’s.
In 2018 PwC prepared an analysis that compared 8 leading forms of distributed ledger technology. DLT technologies based on both Blockchain and Directed Acyclic Graph were included in the evaluation. Hashgraph was included in the PwC evaluation along with Bitcoin, Ethereum, NEM, EOS, IOTA, Byteball and Nano. PwC evaluated performance, measured by throughput, latency, and scalability. They evaluated fairness measured by order of transactions, censorship, stability, and fees. Security was evaluated related to access control, confidentiality, cryptography, availability, denial of service vulnerability and smart contract vulnerability. Programmability and governance were also key elements of the PwC evaluation. The PwC analysis resulted in Hashgraph receiving the highest comparative rating in all 5 evaluation categories.
Hash Labs is proud to have been among the first to complete a license agreement with Swirlds for the Hashgraph technology. Hash Labs is developing and testing its own private permissioned DLT network, powered by Hashgraph. The private Hash Labs DLT network will support the Q4 launch of Coro, our global payment application.