Main navigation

Scaling Blockchain by Going off the Block: Constellation’s Partitioned DAG

Tue, 05/22/2018 - 06:40
article image
Katya Michaels
Finding solutions to encourage widespread Blockchain adoption may require going beyond the block
Scaling Blockchain by Going off the Block: Constellation’s Partitioned DAG
Cover image via U.Today
Read U.TODAY on
Google News

While the crypto community continues to struggle with the implications and applications of Blockchain technology, cryptocurrencies and token sales are becoming more mainstream, creating the additional challenge of scalability. Ironically, the gradual increase of adoption reveals the obstacles to universal adoption.

Some teams are looking for a solution for Blockchain usability by moving off the block to a different kind of network. Two members of the Constellation team – CEO Brendan Playford and COO Benjamin Jorgensen – spoke to CryptoComes about their way of addressing scalability issues and envisioning the interoperable Internet of Blockchains.

Blockchain DIY

Katya Michaels: Before you set out to create your own network, you were building a project on Ethereum. What were some of the challenges that inspired you to work on a new solution?

Brendan Playford: I've been in the Blockchain space about four and a half years. I had the fortune of being a very early miner and managed to use mining to get myself out of the place that I grew up in the UK, where there weren’t very many opportunities for people like myself. I saw the way that Blockchain allows individuals to get economic mobility – I’m absolute proof of that, being in San Francisco now.

In 2016, I was listening to NPR around August and all I was hearing was coverage of fake news with Trump. It was relevant then and it's relevant now – monetizing and weaponizing ad arbitrage on Facebook with fake news purposefully produced by a network of writers. The appetite for fake news was so aggressive in 2016, you could publish whatever you wanted and the volume would decimate any legitimate news.

That was the genesis: to build a platform that would allow factual information to be incentivized and recorded in a way where it became self-regulating and self-sustaining. We quickly realized that the high volume and high throughput we needed was totally impossible to do on Ethereum.

In order to publish the 100 - 200 articles a day, have micropayments going out globally to individual contributors and notarize the content on the Blockchain – there was no solution available that could do that.


Bring your own bandwidth

BP: The nature of Blockchain is synchronous. You have the state replicate across every node, but those nodes are not widely distributed between individuals. Very few people run a full node of Bitcoin or Ethereum. That has led to a somewhat centralized server system based around proof of work or proof of stake where a small selection of the network own it and maintain the state. Although it is very secure, it’s not as decentralized as it was envisaged.

I saw the future not as a synchronous Blockchain, but an asynchronous network that functioned like a graph, with node to node connections. It’s very hard to read a sentence if every word has to be shared between every participant before you can go on to the next one. That's synchronous Blockchain. In a graph, you have a conversation which gossips out to the network. To avoid the information getting distorted with Byzantine actors, you have to create some kind of consensus or architecture to maintain its state.

We set out to build a network that was a horizontally scalable Hylochain. Distributed data systems have been around for years, this technology is not that new. We could create a network with the characteristics of Bittorrent or Tor, that could scale as more participants joined and brought resources – a laptop or a mobile phone –  adding to the throughput of the network.

We have this notion of a mesh net of interconnected devices creating a new Internet. These devices provide bandwidth to the network, and the availability of those resources is incentivized in two ways. One – through the mining period of ten years, and two – by creating a two-sided marketplace where the resources on the network are available for computing services.

We would like to see people coming to the network to use these resources and pay the people providing the devices. That would unlock a huge amount of economic value that's unused right now and create more upward mobility in areas that are less economically developed.


A different kind of chain: partitioned DAG

KM: The Blockchain’s blocks are what make the technology secure, transparent and immutable. Are the DAGs blockless systems?

BP: If we look at IOTA’s Tangle, I would say yes, that is very much blockless. Hashgraph as well, although it has a notion of direction and a linear transaction history. With our chain, we dove into some really novel new research that addresses scaling. One piece comes from TU Delft university in the Netherlands. It’s called extended trust chain – like a blocked DAG (directed acyclic graph). Effectively, you have a partition, a cluster of 1,200 nodes that all participate in consensus and have a certain scaling characteristic.

There is a checkpoint block that happens in the DAG which creates immutability. The checkpoint block does a locality-sensitive hash on the previous transactions. In that partition, we roll up all the previous transactions from consensus and they get blocked.

In our DAG, we have partitions of 1200 nodes, with the capability of having 10 partitions. Above that, we have a galaxy node that's built up enough reputation over time to be given the responsibility of validating larger blocks of transactions, and also send transactions out to cluster.

Proof of Meme

KM: You have said that networks based on proof of work or proof of stake consensus are like a plutocracy – more power is held by those with more resources. Constellation’s consensus is “proof of meme,” reputation-based, but reputation takes time to build. So the network has the potential to become an oligarchy, dominated by the few people who have been there the longest. It’s conceived as a meritocracy, but could become an oligarchic meritocracy.

BP: You are absolutely right. To address this, we have devised a clustering algorithm. Over time we're going to get a curve, almost like a histogram of reputations from zero to let's say a hundred. In the hundred block you have participants who came in at the beginning, in the zero block you have those who came in most recently.

How do we get those new participants up to the higher reputation while maintaining security of the network? We want to give them a fair opportunity to progress.

Imagine taking that histogram and clustering participants into cohorts – first year entry, year 1 to year 9, year 9 to year 20… We create a weighted algorithm that takes a portion of nodes from each cluster and puts those into consensus. You're always going to be taking some from the zero cluster and as they perform consensus properly, they will move up.

We actually have a test net out, modeling the clustering algorithm to see what it would look like in five years and make adjustments to that. We are looking for ways to avoid that oligarchy as much as possible. We are using the REGRET reputation model, but we may find that there's a better measurement we can use in our machine learning algorithm.



KM: Some critics say that Blockchain is a solution looking for a problem. Despite many proposed use cases, they are not being realized. What will it take to bring real usability to Blockchain? Or are we in a post-Blockchain stage already, moving to more scalable technology?


I think a lot of negative press comes from the tendency to overpromise and underdeliver. As thought leaders and pioneers in the space, all of us need to be conscious of that and make sure that when we do say these things openly, that there is some substance behind them – whether with academic groundwork or actually delivering a viable product.

Hundreds of ICOs and dapps launched in the last year where the promises will not happen. They just won’t deliver. That exacerbates the problem.

Are we post-Blockchain? I don't think we are. I think we're about to enter into the Internet of Blockchains era. You will have Bitcoin for store of value. You will have Ethereum for certain uses, and Zcash for certain uses. Don’t underestimate the Bitcoin core guys. There is a roadmap that they have and it is highly likely we will not see all the solutions at once.

You want to bring about change gradually and slowly, and you want to bring about adoption in a sustainable way. What we’d like to see with our architecture is an underlying fabric that connects and mixes these chains. It's not one Blockchain to rule them all.

There will be individual solutions like Constellation that applications can interface with. Over the next 5, 10 years Bitcoin will still be around, Ethereum will still be around. There’ll be other technologies that will enable the interconnection of these chains.

Enterprise adoption

Benjamin Jorgensen: One of the main hurdles for adopting Blockchain is the cost benefit analysis for major enterprise companies to shift out of their legacy platforms and come onto Blockchains.

We look to the Fortune 500 companies to guide us, but if you consider the history of venture in Silicon Valley, you see that innovation is always done at the grassroots level.

Blockchain does solve problems. We've identified that it gets rid of middlemen. It allows for a distributed ledger so that people can own their own data, going back to what we initially set out to do. Major commercial banking isn't going to shift a significant portion of their business under the Blockchain because of the cost benefit analysis, hiring and firing new people, getting rid of services, the time it takes to ramp up. I think we're going to see a new era of businesses come to the forefront and actually recreate this new world.

KM: Perhaps it's a space for businesses that are going to be natively Blockchain.

BJ: Absolutely. We know that by implementing Blockchain you can significantly improve margins. The initiative is probably going to come from new businesses that are able to create something that's leaner, faster and stronger, and scale rapidly to catch up to those Fortune 500 companies. We've seen this before in the venture space with traditional startups.


Internet of Blockchains

KM: Going back to having different Blockchains for different uses – obviously that depends on good solutions for Blockchain interoperability. Are we getting close to that in terms of technology?


Every Blockchain is just data held in a space where it’s notarized as factual and correct. If we can create some way of formally defining and verifying each chain in this ecosystem mathematically, every developer who builds on one chain or across chains will have a framework.

Think about banking – we have SWIFT for a reason. It is the standard that allows transactions between banks. We have to have a SWIFT for Blockchain. When you start to see those frameworks come out and be supported by native languages like Java with plugins for other compilable languages, that’s when we'll see an explosion.

BJ: You have to look at the implications of this - why does it matter to have Blockchain interoperability? Where Constellation’s transaction speeds are really coming to the forefront is around the IOT space. Connecting software devices, automation, AI, micropayments. They're going to have to be able to communicate in a seamless manner. Maybe it’s not a problem, but it's an opportunity.

Smart contract usability

KM: A lot of people have misconceptions about smart contracts – how they work, what they can and can’t do. A smart contract is limited to the Blockchain. In order to access relevant outside data, there have to be solutions for bringing that data onto the Blockchain. How close are we to real smart contract usability?

BJ: When you think of a smart contract, you think it's legally binding, when it's really just a document that says: these two terms have been met, let's exchange the value that comes in. Tennessee courts just reaffirmed that a certain document that's on a smart contract with certain verbiage actually can be upheld in courts. We're just getting to that point where it's going to have a legally binding impact.

BP: This goes back to appropriate use cases. What is a contract? It’s just written verbal logic. What is code? Code is logic. So we figured out this way of notarizing a bit of logic on a chain and then replicating that state across nodes.

Ethereum relies on oracles to provide this data, but there's no way that thousands of data points would be able to transact and go onto Ethereum with the current throughput without being somehow aggregated and centralized.

A chain like Constellation could collect that data together, creating a data marketplace that becomes the oracle for Ethereum in a cross-chain sense. Ethereum could call Constellation through an ACI (application chain interface) for that data, which will be known to be provable and factual, instead of relying on a sensor and a centralized server. If you're connecting to Constellation, that will give a decentralized source of truth.

How far are we away from a place where you could rely on sensor data to provide an outcome of a smart contract? I would say five years. There’s a curve with this kind of technology, Metcalfe's law.

We’ll start to see the first commercial applications, the beginning of a mesh network where there is a source of truth that connects everything, all underpinned and underwritten by reputation. That's what we're building.


article image
About the author

Katya Michaels is a writer and editor living in California. She is passionate about excellent writing and dedicated journalism, but ambivalent about the Oxford comma. When not crusading for the rescue of long-form content, she watches sunsets, scuba dives, plays Chopin Nocturnes and teaches her daughter to express herself without the use of emoticons.