Blockchain 2.0, or the Directed Acyclic Graph

We continue to have breaking news in the blockchain world. First of all, since our last post,  Facebook has made official its new cryptocurrency, Libra. Secondly, and possibly as a consequence of Libra’s introduction, Bitcoin has at the time of writing shot past USD 12,000, when in December 2018 the rate was hovering at around USD 3,400. Distributed ledger technologies (DLT), of which Blockchain is an instance, are generating significant interest, and are developing very quickly. For those who still haven’t wrapped their head around Blockchain,  they will be either dismayed (or relieved?) to know that almost as soon as it has appeared, there is some question as to whether it is in fact already obsolete. How can this be? This needs further explanation, lest the moderately perplexing topic of DLT become hopelessly incomprehensible!

An integral feature of DLT is the idea of consensus. Broadly we can describe the motivating problem behind DLT as creating a verification mechanism in a network with no centralized verification. In the case of cryptocurrencies like Bitcoin, for example, this translates as the problem of creating a secure ledger of transactions (i.e. a currency) in the absence of a trusted central organization that would check them all. Generally speaking, and omitting technical details, transactions need to be verified by each node (member) in the network, before they can be added to the overall ledger. This process is what is defined as “consensus mechanism”.

In the case of Blockchain, the consensus mechanism is simple enough: once the network validates the set of transactions, it adds a “block” to the “chain”. There are therefore 2 steps: the network validates the transaction, and then the new block is broadcast to the rest of the network, which substitutes the old blockchain with the updated version.

 

This process, however, has many shortcomings. Chief amongst them is the question of scaling:as the network grows, the verification process becomes quite slow. For Bitcoin it is around 10 minutes, which makes payment impractical – imagine having to wait 10 minutes to pay for  a cup of coffee. Secondly, the length of the record/ledger of transactions becomes a liability, as each node in the network needs to keep a complete record of all individual transactions, i.e the entire blockchain.

 

YOU MIGHT BE INTERESTED IN: Blockchain and IOT, how do they add up?

 

For the above stated weaknesses, alternative solutions to blockchain have begun to appear in order to make DLT consensus less costly. Most notable is the Directed Acyclic Graph (DAG).

The DAG is a type of mathematical chart which in this case represents cascading stream of transactions happening at random spaces of time, and potentially also simultaneously. This structure can be visualized as a tangle of yarn. Instead of validating a long sequence of transactions , with the DAG the addition of a new transaction only involves validating several previous transactions. As a result, there is a much more limited scope of verification. The process is as a result is much lighter.

The second problem with Blockchain invoked above was the storage of the entire history of transactions on each node. Storage on the DAG can be configured differently, but it would not involve every node keeping a record of all transactions. There would specially designed nodes that keep the entire transaction history for records purposes, but this involves much less duplication than Blockchain. Another solution is keeping a “snapshot” of each user’s balance at a certain point in time, which is again much less costly than keeping the entire transaction history.

What can we infer from this? DLT is very much in its infancy. The problem facing the industry is that there is a massive proliferation of protocols, which is the official word for the fundamental DLT structure, which includes the consensus mechanism but also other things. These technologies are not typically thoroughly tested when they come to market, and many fizzle out before reaching any sort of critical mass in terms of network size. It is possible that a few years need to pass for the dust to settle and a few protocols to establish themselves as leaders. Most companies are still only experimenting with DLT, and with a variety of protocols, as well as the uncertainty surrounding their speed and security, there can only be further caution around their adoption, as enterprise use cases require proven and reliable technologies that are to be used several years or decades into the future.