A new distributed ledger technology known as the hashgraph has been creating buzz in blockchain circles. But what is a hashgraph? In a few words, it's a technology that can offer many of the same solutions as blockchain, but which uses different mechanisms to transmit information and confirm transactions within the network.
Hashgraph is essentially a graph in the mathematical sense of the word, held together by simple, regular hashes with no high-level cryptographic implementations, like zero-knowledge proofs, for example.
A hashgraph network achieves consensus (meaning that nodes come to agree on the state of the network, including records of who owns what assets within it) by using what Leemon Baird, who originally conceived the hashgraph, calls "gossip about gossip." Before examining this concept, it's important to look at the role of gossip itself in the network.
Hashgraph nodes are constantly "gossiping," or sharing with each other every piece of information that they know about the current state of the network as well as its history. Rather than having the whole network agree at once that a certain transaction took place, as happens in blockchain environments, a node involved in a hashgraph transaction randomly selects another node to inform about the transaction (and everything else it knows), then another, then another, ad infinitum.
All nodes receiving this information also share it with randomly selected nodes over and over, each time adding to their reports any new information that they received since they last broadcast the sum of their knowledge. This constant sharing prevents malicious actors from withholding information on transactions, an act that can form the basis of an attack on a DLT network.
Gossip about gossip builds on this concept by including hashed values in nodes' reports to other nodes which indicate the time at which the reporting node became aware of a piece of information. This allows the network to ascertain the order in which nodes received each piece of information, significantly enhancing network security by making it almost impossibly difficult to fake a transaction.
Rather than each node having to vote on what it believes to be the true state of the network, the network can simply examine these hashes to determine which node knew what information at what time. This allows the network to infer consensus through "voteless voting" – the process of conducting a virtual poll by assuming that a node's vote would not contradict the information available to it regarding the state of the network. According to Baird, this allows the network to achieve consensus "for free. You will not have to talk to anyone and you'll have perfect consensus with math guarantees that you're right." In other words, a hashgraph's consensus mechanism is extremely hands-off relative to the PoW and PoS schemes being implemented and discussed in blockchain circles.
During a recent press event for Hedera, a hashgraph firm that lists Baird as co-founder and CTO, Baird explained that nodes always know which segment of the network they're communicating with, making certain types of attacks on the network significantly less feasible if not impossible. According to Hedera CEO Mance Harmon, this architecture also precludes the possibility of forking the network. This is another difference between hashgraphs and public blockchains.