You can’t have your cake and eat it too, the old saying goes. In the cryptocurrency space, the big question is whether you can have decentralization and scale it too.
The struggle between these two goals has become more noticeable as a rush of new enthusiasts have entered the space, lured by the technology being floated as a fix to many of the world’s problems (well, that and greed).
Underscoring the urgency of the problem, a number of researchers are studying whether there’s a balance to be struck between decentralization – the key non-financial motivation for cryptocurrency and blockchain excitement – and these networks’ ability to scale to an ever-growing number of users.
Cornell University researcher Adem Efe Gencer explains the predicament well, telling :
“The key impetus behind the interest in blockchains is decentralization of power among entities with minimal trust relationships. But, there’s a fundamental tension between scale and decentralization. While we know how to get scale, scaling solutions may require compromising the decentralized nature of blockchains.”
And this dilemma has become more serious as users face deep transaction backlogs and rising transaction fees on the most popular blockchains dealing with the weight of increased adoption.
In response, many wonder whether the blockchain deserves all the attention it’s currently receiving – or if it’s yet another overhyped technology sector where expectations outpace actual usefulness.
At this point, there are solutions under development, such as bitcoin’s Lightning Network, but there are always trade-offs to be made.
Gencer’s new paper and a handful of others recently published explore just how decentralized the most popular blockchains are, and how resistant they will be in the future to being taken taken over by a controlling force as new technology for scaling is added.
Mirroring the debates in the bitcoin community about the best way to overcome the problem, the researchers each stressed different possible solutions in their findings: on-chain, off-chain and at the consensus protocol level.
Just how decentralized?
In their paper, Gencer and his co-authors aim to “scientifically quantify” the decentralization of the networks bitcoin and ethereum.
To do this, the paper brings together a range of metrics, including how many miners (the stakeholders that order and verify transactions on the network with the use of computer hardware) a network has, how many nodes, how much bandwidth it takes to send blocks across the network and so forth.
Looking at this big picture can help researchers to potentially address the scalability problem. Gencer told :
“While we know how to get scale, scaling solutions may require compromising the decentralized nature of blockchains.”
With that in mind, the research sheds some light on the “viability of different scaling proposals.”
For example, the block size can be increased up to 1.7 times to improve on-chain scaling without harming decentralization in some ways, the authors found.
It’s worth noting Gencer and his colleagues’ findings have been debated fiercely on social media. Because decentralization is such a complex property to measure in a system like bitcoin, many contend other elements of decentralization would need to be taken into account to get the full picture.
The scaling ‘triangle’
Another quantitatively focused paper exploring the issue was published on arXiv last week by Greg Slepak, the founder of decentralized technology non-profit OkTurtles.
The paper describes the “basic theory” behind the scalability problem by way of a triangle (another researcher, Ocean Protocol founder Trent McConaghy, independently came up with this same idea).
Each point of the triangle represents one of the three properties that blockchains need – scale, the ability to support many users; decentralization, a state of affairs in which no one entity controls the system; and consensus, the agreement between each node on the validity of transactions.
Just as one line of a triangle can touch only two of the three points, so too, are blockchain developers only able to focus and succeed in two of the three properties necessary for decentralized blockchains.
For instance, blockchain developers can have a system that scales and reaches consensus, but it will be at the expense of full decentralization.
Still, Slepak’s research was more optimistic than Gencer’s, since the paper claims there’s evidence that off-chain networks, such as Lightning Network, can “get around” this tricky trio of trade-offs.
To do so, the paper contends, off-chain solutions like bitcoin’s Lightning Network can “relax” the definition of consensus. Rather than requiring nodes to hold every transaction that ever happens within the system, nodes only need to keep track of some transactions requested by users in ‘rare moments.'”
Arguing that his research describes, in more formal terms, the trade-offs that have been at the heart of bitcoin’s years-long scaling debate, Slepak told :
“This seems pretty important to me given the fuss around [bitcoin’s] block-size debate.”
The base layer
Meanwhile, University College London postdoctoral researcher Shehar Bano has also been researching the scaling/decentralization conundrum, but as it relates to each particular blockchain’s consensus protocol.
In a paper titled “Systemization of Knowledge,” Bano aggregates a number of different consensus protocols, including proof-of-work (which bitcoin uses) and proof-of-stake, and compares them, in order to figure out where development is wanting.
“Researchers can then look at it and think, ‘Oh, I can contribute here,’ or ‘Oh, here’s a gap,'” Bano said.
Despite the sprawling scope of the paper, decentralization was one of the main trade-offs Bano mentioned in conversation with .
“Because what shows up in some blockchains is that, while the system is decentralized, somehow due to the governance structures or the nature of the protocol, centralization happens again,” she said. “We can see this with bitcoin mining,” where just three pools account for roughly 56 percent of the hashing power right now.
Solving the problem will hinge on developing better consensus protocols, Bano argued, concluding:
“The other features are really the icing on top of the cake. Consensus protocols are the cake.”