DAG Blockchain: Hashgraph
First, the blockchain was simply the blockchain. But recently we got the Directed Acyclic Graph. And now we have the HashGraph. The piece is about the latter.
It is interesting to see how in the world of Cryptocurrencies new phenomena can quickly end up as a sort of “fashion chick”. A few examples:
- Somewhere in the period 2014-2017 the phenomenon of ICO’s arose. First, you had a few. But in the meantime, they are sprouting like mushrooms! Some are nice, but most are even more absurd than the others.
- Forking has of course been possible since the first existence of Bitcoin. But in 2017 a large and relevant fork took place, which resulted in a kind of fork-wave. The counter for the number of Bitcoin forks is already 11 at the time of writing. And these too are becoming increasingly absurd.
- And then came the DAG, the “Directed Acyclic Graph”. As a variant of the blockchain. Meanwhile, you also read about the “HashGraph” and blockchain innovations starts to look fashion sensitive.
We are going to see if the HashGraph is absurd or not.
We know the “good ol’ blockchain” as a chain of cryptographically strung blocks with transactions (or appointments, or even software code) in it. Stringing is done by repeating the hash (fingerprint) of the previous block at the top of the next block and hashing it again. Blockchains are robust by distributing many copies across ‘nodes’ using peer-to-peer technology.
Blockchain networks come to a consensus through an often used mechanism: “proof of work”. In a race, the search for the solution of a cryptographic sum is sought. The winner determines the content of the next block. Unfortunately, PoW costs a lot of energy. But to this day this is the best-proven protection method.
So far so good.
In September 2016, Byteball and a little later IOTA, the Directed Acyclic Graph, came into the picture. A system in which no mining is needed for consensus, because every recorded transaction is only an unambiguous transaction if it has verified and confirmed two other transactions. There are no blocks but separate transactions. The more transactions, the more confirmations, the more reliable. There is still a little issue with a centralized component that still needs to be removed.
Colleague Sander Grootendorst has written an extensive piece about it.
Now the HashGraph. This too is a technique to achieve what blockchain and DAG want to achieve. Namely the reliable recording of transactions, appointments, and even software on a distributed network of computers. So without the intervention of a third party or person. (Also called distributed ledger technology (DLT)).
Like to gossip?
How does the HashGraph work? With the “Gossip about gossip” consensus algorithm. (Gossip about Gossip). It allows each node to expand blocks as often and frequent as possible. Without mining.
In the HashGraph, every node communicates with random other nodes and tells them what the other person does not yet know. In this way: “Hey, Charlie, I just did transaction PQR. And I just spoke to Peter, has done transaction XYZ! ” Each node thus adds a transaction to the network and confirms one that already buzzes in the network. This approach ensures that every transaction is distributed over the network in no time and is therefore reliably verified. The network converges to one truth.
And it makes it so that this happens in a fair order. More or less “first come, first served”. This is in contrast to the blockchain, where priority is determined on the basis of commission height.
With the HashGraph created like this, you know all the transactions and their progress through the network in time. You can therefore calculate (check) that other nodes have done their calculations correctly. It is therefore no longer necessary to vote on what the reality is. The consensus is inherently achieved.
The famous witness
Just as with IOTA and Byteball, there is a witness. But in HashGraph it is not a centralizing force. In HashGraph, the “famous witnesses” arises very quickly after making a transaction. They provide extremely fast confirmation. You could say that they are a kind of accelerator, but without it would work too. (A bit like the masternodes in Dash that support instant-send). In HashGraph we are talking about 250,000 to 300,000 transactions per second.
I can see you very well!
In HashGraph there is a difference between seeing another node and seeing well. A node is well seen when other nodes have already gathered enough knowledge to estimate together what another node would vote for. If they already know that, there is no need for confirmation.
Are there any drawbacks to this system? Of course. Of course there are,
- To start with, there is still no knowledge about how this system will behave on a large scale. And that is an important aspect. Time will have to prove it.
- And secondly, the idea is patented. And not open source. This is strongly against the nature of the crypto community. If it turns out to be a successful model, it would not surprise me when an anonymous software collective arises that develops an open source variant and releases it for everyone in the world to use.
- Also, there is not really a community developing around HashGraph. Something that is naturally inherent in choosing closed source but not desirable.
It seems that the HashGraph is a development with potential. Too bad it’s patented. What happens when applying the HashGraph on a large scale, or even public scale alone, we do not know yet. It remains a matter of patience.
Finally: Here is another explanation video about HashGraph. And here an interview with Leemon Baird the inventor. And here two (known) Bitcoin-core fans, who are not bothered by any lack of knowledge, undermine the HashGraph immediately.