Features  •  Technology

How Consensus Algorithms Solve Issues with Bitcoin's Proof of Work

(@danielcawrey) | Published on September 11, 2014 at 11:12 BST

It used to be that bitcoin mining was a community affair, but the process of confirming bitcoin transactions now requires an immense amount of hashing power specific to the SHA-256 mining algorithm.

As a result, the amount of electricity required to process and verify transactions has reached an immense scale.

In the early days, when the first users like the late Hal Finney began experimenting with bitcoin, users could just open the early bitcoin client and let their CPU whir away, creating bitcoins for fractions of a penny. By contrast, mining now requires tens of thousands of dollars in equipment, electricity and optional hosting expenses.

As a result, those who have significant financial resources have come to dominate the bitcoin mining space. Mining today is embodied by the emergence of enterprise-style, datacenter-hosted mining operations.

Jackson Palmer, who created the scrypt algorithm-based, proof-of-work altcoin dogecoin, is not overly positive on the prospect of mining overall, particularly in regards to the energy consumption.

He told CoinDesk:

“I'm quite openly against [proof of work] mining as the future of digital currencies based on the amount of energy it wastes, and damage it does to the environment – without giving anything back apart from making the fiat rich richer.”

One possible solution that has already been deployed in the digital currency ecosystem shifts away from proof of work, but still embodies the principles of a non-centralized systems.

The big technology race 

A major issue that plagues mining today – but may not receive much attention – is that the application-specific integrated circuits (ASICs) for SHA-256 or scrypt mining are only capable of one process.

ASICs can mine digital currency and confirm transactions, but once they are rendered obsolete by next-generation mining gear, the equipment quickly loses its value.

While most components in mining gear are recyclable, the turnaround from product creation to product demise in this industry is a short.

Total hasing power of the Bitcoin network continues to rise. Source: Blockchain
Total hashing power of the Bitcoin network continues to rise. Source: Blockchain

In order to mine competitively, there is a constant race to build and obtain more powerful equipment.

There’s nothing wrong with a little competition, but the race to build better miners has seen the rise of enormous operational costs. As a result, many miners are locked into long-term return-on-investment (ROI) schemes that are susceptible to any significant changes to bitcoin's exchange value.

A decline in the price of bitcoin could diminish the incentive for many miners to participate. While this wouldn't lead to any immediate disruptions, the long-term health of the decentralized network could be put at risk as only the well-financed mining outfits could continue working.

Despite the increase in hashing, mining revenue has remained flat. Source: Blockchain
Despite the increase in hashing, mining revenue has remained flat. Source: Blockchain

Given these threats, is there an alternative to proof-of-work systems that hash out cryptographic problems and, all the while, use up excessive amounts of money and natural resources?

While there have been a number of proposals to use proof of work for good use, such as for solving prime numbers or promoting personal health, these may not be the ideal solution either.

Solution in the pipeline?

A possible solution to issues related to proof of work may come from an academic paper published by Miguel Castro and Barbara Liskov from MIT in 1999 presenting the concept of Practical Byzantine Fault Tolerance (PBFT).

PBFT is a system initially devised for low-latency storage systems – something that could be applicable in digital asset-based platforms that don't require a large amount of throughput, but do demand many transactions.

Daniel Feichtinger, co-founder of a startup called Hyperledger that utilizes PBFT, explains how its distributed system works:

“Each node publishes a public key. Any message coming through the node is signed by the node to verify its format. Once enough responses that are identical are reached, then you can agree that is a valid transaction.”

That agreement constitutes what is known as 'consensus'. Just as bitcoin uses a decentralized system in order to confirm transactions without the need for a trusted third party, PBFT relies on the sheer number of nodes in order to confirm trust. As a result, hashing power is not required in this process.

“When you interact with multiple parties, you need some sort of consensus mechanism to ensure everyone has got the right records,” said Dan O’Prey, the second co-founder at Hyperledger.

Consensus already at work

Consensus-based transaction mechanisms have been on the market for some time. Ripple Labs utilizes this approach as part of its underlying protocol, based on the concept of PBFT.

How Ripple Labs portrays connectivity in its system. Source: Ripple Consensus Whitepaper
Ripple Labs portraying strong (bottom schematic) node connectivity in its system. Source: Ripple Consensus white paper

Ripple Labs recently released a white paper on its consensus algorithm that points out the Byzantine Generals problem of communicating over an unreliable link, something PBFT is intended to solve.

The white paper states:

“It is provable that no solution to the Byzantine Generals problem (which already assumes synchronicity, and known participants) can tolerate more than (n−1)/3 byzantine faults, or 33% of the network acting maliciously.”

The likelihood of a group conspiring to take down the Ripple goes down based on number of UNLs. Source: Ripple Whitepaper
The likelihood of a group conspiring to take down the Ripple network decreases based on number of nodes on the list (UNL). Source: Ripple white paper

As a fork of Ripple, the Stripe-backed digital currency Stellar, founded by Jed McCaleb, also encompasses these same ideals. The idea, as shown in the graphics above, is that for every node added to a network's Universal Nodes List, the stronger the system becomes.

Yet there is a key weakness to this kind of structure – the source of incentive – and it is a fair question to ask. Unlike mining, what incentive exists for individuals or entities to host nodes?

Both Ripple and Stellar, along with Hyperledger, believe that there will be enough stakeholders to support PBFT-based transaction systems.

Hyperledger's approach

The Hyperledger project allows developers to create their own digital assets with a distributed ledger powered by nodes built on the principle of PBFT.

The system could be used to digitally back a real asset (such as a house), create new coins, or form a fault-tolerant system of consensus. Hyperledger co-founder Feichtinger said, using the latter application, his company's platform could even be used for advanced IT systems.

He explained:

“If you were Amazon running a distributed file system for AWS [Amazon Web Services] in your datacenter, [you] want to ensure an error is not going to corrupt AWS [and] you don’t need incentive. You run the nodes that you want, [using] a consensus system.”

Hyperledger prefers a distributed model of rather than Bitcoin's decentralized one. Source: Hyperledger
Hyperledger prefers a distributed model rather than bitcoin's decentralized one. Source: Hyperledger

The idea for Hyperledger’s use of PBFT goes beyond asset-based systems. It takes the idea of an algorithm for consensus and uses it to distribute all sorts of technical solutions – not just the low latency, high-speed file storage solution it was originally built to provide.

This might be a good method of testing the power of nodes that do not use incentive to develop their strength. What will happen without such rewards? Systems like Hyperledger aim to find out.

“If you use Byzantine Fault Tolerance, ideally [corruption] problems are contained. The other nodes can realize [a] node is misbehaving, and not respond to its messages,” Feichtinger said.

Consensus as a service

There could be value in providing consensus-as-a-service in the way that cloud providers sell software-as-a-service or infrastructure-as-a-service. In the case of Ripple and Stellar, consensus becomes a medium of value exchange.

Feichtinger explained:

“Much like Google run their own DNS servers – it’s in their interest to have healthy DNS. It means they get to serve up their ads quicker. I think DNS is good example for this low-level protocol. Increasing the health of the system."

Given the fact that consensus-based transaction systems are still in the experimental phase (much like the whole of digital currency), expect to see additional solutions like these to appear on the market. Whether they prove to be a viable replacement to proof of work systems will come down to one thing: if they are used or not.

Palmer, whose dogecoin creation relies on auxiliary proof of work (or 'merge mining') to secure its network, added there needs to be new and concrete solutions like Hyperledger and Stellar. These, along with Ripple, he believes, can help grow the digital asset ecosystem as a whole.

He concluded:

“Exciting times [are] ahead. It's great to see real innovation in the digital currency space rather than more Kool-Aid drinking.”

Encryption key image via Shutterstock

HyperledgerPractical Byzantine Fault ToleranceRipple LabsStellar

Load Comments