The fundamental features that make blockchain such an interesting tool are not specific to blockchain itself. They are features that belong to a family of technologies known as distributed ledger technologies (DLTs). It just so happens that blockchain is currently the dominant method, or data structure, used to implement distributed ledgers. Pick any metric - number of projects, developer resources engaged, investment dollars attracted - blockchain dominates.
However, as previously discussed, scaling blockchains for mass adoption is somewhat of a challenge. This is because blockchain relies on all participants agreeing on all new state changes globally and doing so one block of data at a time, in series. Direct Acyclic Graph (DAG) implements an alternative distributed data structure that attempts to get around that by enabling state changes to be executed in parallel.
Data structure and flow in a DAG (top) and a blockchain (bottom) Image credits
The question is whether this theoretical benefit will translate into real-world successful projects that enable mass-adoption leading to DAGs superseding blockchains.
Tech Wars Vs Evolution
DLT development is still at the embryonic stage - adoption is even further behind - so we would be foolish to believe blockchain is the be-all and end-all while discounting other DLTs.
We've seen it time and time again, where a pioneering technology was eventually superseded, sometimes before mass adoption even occurred. A great example in computer networking was where IBM's 'Token Ring' specification was superseded by Ethernet - which eventually became the global standard for local area networking (LAN). It was a similar tale for Sony's Betamax technology that was swamped by VHS in the race to bring video recording and playback to the home.
The key here is that it is not only the technology that counts. Keeping a technology proprietary can kill it no matter how good it is. In the same way, effective branding and marketing can achieve the critical mass required to kill-off better technology that didn't make it into the hearts and minds of the population. The tech industry has observed these lessons repeatedly over the past decades.
In parallel to the micro process of competing products killing each other off, we have the macro process whereby whole industries make discrete technological leaps over time. [3.5" floppy] to [CD-ROM] to [SSD] happened in 30 years, with each step enabling 1,000 times more data to be stored in roughly the same form-factor.
Both of these processes - product wars and technology leaps - are playing out in the world of DLT and it is happening at breakneck pace with exponentially increasing amounts of players joining the game. How to keep track of it all?
To give some structure to where DLTs are now and from where they have come, we can try to categorize the progress to date as follows:
DLT 1.0 - The implementation of Bitcoin in 2008
It is not possible to overestimate the magnitude of this breakthrough - it was the first system that enabled trustless peer-to-peer value transfer through a distributed public network while solving the double spend problem. At this point blockchain was not even a term. Satoshi Nakamoto’s white paper, did use the words “block” and “chain” but only as adjectives for the data structure. While Bitcoin was a huge leap, it was a single application network with no built in flexibility to unlock blockchain's wider applications.
DLT 2.0 - Blockchain as a technology type in c. 2012
The idea that the technology behind bitcoin could be useful in other applications led to pioneers coining the term 'blockchain' to describe the underlying technology. The evolution of this idea eventually led to the birth of Ethereum - Vitalik Buterin's brainchild - in 2015. Ethereum allowed bespoke distributed applications to be built open source on top of the underlying blockchain, through the use of smart contracts. This led to a second wave of development that moved blockchain away from being purely crypto-currencies to a solution effectively applicable to any system where records are kept.
DLT 3.0 - Next Gen DLTs for mass adoption (under development)
These platforms will begin to address the significant teething problems that blockchain has felt through stage 2. These have typically related to the scalability limitations associated with the fact that blockchain networks must relay every transaction executed to every network participant. The most successful implementations that emergy from DLT 3.0 may expect to gain significant traction as they will remove the current bottlenecks and enable mass adoption.
Who is Fighting it out for DLT 3.0?
It is a busy space with new projects seemingly cropping up every day. A clear set of criteria is required to help in separating the likely contenders from the likely failures. Three of the most important criteria when assessing the potential of a project are:
- Concept outlined in the white paper (technical solution, scalability, governance system, innovation, foresight, realism)
- Resources available from ICO or private investment ($$$)
- Quality of developers (reputation among peers and their track record of innovation and success)
Ethereum is certainly a contender - it has a proven network, the largest community of developers and massive resources based on the soaring value of Ether (increased by 5,500% in the past 18 months). With hundreds of applications being built on Ethereum, it now finds itself in a race against time to scale its throughput. It's road-map proposes to transform the platform completely - effectively re-launching it as Ethereum 2.0. Firstly its consensus mechanism will transition to Proof of Stake through the release of Casper. Secondly they plan to achieve orders of magnitude of scaling through Sharding, Plasma and state-channels. But their biggest enemy may be time - it will take years to achieve all of these upgrades in full.
And that is opening the door to competitor platforms like EOS, Polkadot and Cosmos, who are building platforms from fresh. They have the advantage of no legacy platform to build on and they may well outpace Ethereum's progress.
But not only will blockchains be competing at this stage. DAGs has been quietly emerging and gaining traction with some high profile projects that claim to provide more scalable, energy efficient and time efficient networks compared with the classical blockchain.
Some of the prominent DAG-based projects
IOTA’s ‘Tangle’ is probably the most well-known project built on a DAG - it's currency MIOTA is one of the largest crypto-currencies of all, by market capitalisation. The idea of tangle is to enable billions of internet-connected machines to perform autonomous micro-transactions with each other as part of the future Internet of Things (IOT). One thing that grabs the eye about IOTA is their impressive list of serious industry players as partners, like Siemens. And their global recruitment suggests they are planning to push roll-outs aggressively.
Hedera Hashgraph is possibly the closest direct DAG competitor to Ethereum. It bills itself as the trust layer of the internet and a platform that enables and empowers developers to build an entirely new class of distributed applications never before possible.
Byteball and Nano are also DAG-based platforms - but these are specifically designed for value transfer, so they are effectively aiming to achieve what Bitcoin originally set-out to do - create peer-to-peer digital cash. Nano is a more pure-play value transfer app that is built for simplicity and performance, whereas Byteball includes a lot more user-functionality using smart contracts.
Where does DAG beat Blockchain?
In essence, it can do more things at once. Current blockchains are limited to updating the 'database' one block at a time. The whole network must always be synchronised - this hampers speed and scalability as more participants join the network.
DAG on the other hand is asynchronous - verifications are only performed when a participant wants to make a transaction and it is only that participant who carries out that verification. Each participant only records transactions that relate to itself and only verifies with the immediate participants that it previously transacted with. Because of this, no miners are required resulting in lower costs, less energy consumption and reduced latency.
Of course there are trade-offs to this improved performance over blockchain. The most cited one being security. Because each participant only typically needs to verify a transaction with two other participants, in theory there should be more potential for successful malicious activity.
The other current downside is that a lot of the DAG implementations seem to struggle to implement satisfactory decentralisation.
- IOTA uses a centralized coordinator node to verify every transaction. This is used to mitigate the Tangle's susceptibility to a 34% attack on the network, which while the network is still small, would be very easy to achieve. IOTA claims that the coordinators will only be necessary until the number of network nodes reaches a certain threshold.
- Byteball relies on only 12 witnesses for the entire network. These are "highly reputable users with a real-world identity, who acknowledges each transaction seen".
- The Hedera Hashgraph algorithm is not open-source but patented and its governance is overseen by the Hedera Hashgraph Council - a group of up to 39 enterprises and organizations, so Hashgraph will be governed by a board of corporations in a system akin to that of Visa.
While no system is perfect, all of these examples appear to fly in the face of the core ideologies of a permissionless public blockchain.
On top of this some prominent minds in the industry have cast doubt on the benefits of DAG. Vitalik Buterin said "DAGs don't solve any fundamental scalability problems. They solve latency problems at best, and in general I think DAG tech is overhyped." Dan larimer, the founder of BitShares, Steem and EOS.io has also been critical of DAG implementations. Whether there is an element of partisanship in these critiques is open to debate, but in any case, they can only be applicable to the current level of technology and we must remember that future developments could render these critiques academic.
DAG offers an innovative alternative to blockchain that improves on some of the current performance challenges of blockchain including transaction latency and throughput scalability.
It looks likely that DAG will have significant applications in the micro-payments infrastructure of the IoT as well as peer-to-peer digital cash although as with any new technology it will need to overcome some growing pains. Its potential security drawbacks may impede its applicability to certain applications such as store-of-value or identity, however it remains to be seen whether these issues will be exposed or not in practice. And while current implementations of DAG rely heavily on some form of centralisation for verification or governance, there is no obvious reason why the technology cannot be implemented in a proper decentralized manner.
With DAG technology (as a DLT) being even more immature than blockchain, we are in effect looking at a group of foals and trying to predict the winner of the Grand National but it will certainly be interesting to follow the developments of the two technology types over the coming years. In all likelihood the end-game will be 'horses for courses'.