Topic: What do you think about Hashgraph ? (Read 122 times)

The technology used by hashgraph differs in the following ways:

Hashgraph provides a total order on transactions, which means that you can use it as a multi-master database;
Hashgraph is Byzantine Fault Tolerant;
Hashgraph does not require Proof-of-Work;
Hashgraph reaches 100% certainty on the order of transactions; and
Hasgraph handles partitions which is important for how well it will work at scale.
A hashgraph is a history of how people have gossiped. So it tells you exactly what each person knows, and when they knew it which allows you to do virtual voting.

Hashgraph is more secure than blockchains are, more scalable than iota’s tangle, and is more efficient and less wasteful than both.

Hashgraph’s mechanism is actually significantly different from either of those. It’s more like the “practical byzantine fault tolerance” system used by NEO, except that it doesn’t require selecting a ‘leader’ which can become a central point of failure if targeted by DDOSers or other network attacks.

Unlike the blockchain and to a lesser degree iota’s tangle it also isn’t susceptible to attacks by quantum computers. The system it uses is based on ‘virtual voting’ which does not require solving any fancy puzzles or anything else which a quantum computer could affect. Since it uses proof-of-stake you’d have to own more than one-third of all the ‘stake’ on the network in order to break it.

Also unlike both the blockchain and iota’s tangle, hashgraph can and does reliably reach 100% consensus. Once it reaches 100% consensus on the state of a given set of transactions it’s literally impossible to overturn the decision. Even with infinite computing power or thousands of junk accounts there’s no way to successfully mount a double spending type attack without owning more than one-third of all the ‘value’ (currency or trust, depending on how it’s being used) on the network.

Even if you did own more than one-third of all the value on the network the other nodes wouldn’t accept conflicting transactions since they literally cannot occur in the model, although you could probably break the network since the other nodes would be unable to reach an agreement. Even allowing that the nodes are still pretty resilient and unless there’s a very even tie in the propagation rate for the conflicting transactions I think they should still be able to pick one or the other, whichever happens to get sent along more quickly. It’s just that freaking secure.

I would suggest skimming the white papers for all three. Even if you don’t understand all the complicated proofs you can still get the basic idea of how they work.


Will Hashgraph be the Keyword in 2018? Pls discuss