As interest in blockchain reaches a fever pitch, organizations are aggressively building their first projects and seriously thinking about how they will begin testing and using blockchain more broadly. While the blockchain concept first appeared in Bitcoin and other cryptocurrencies, as an immutable record of transactions, maintained collaboratively by all parties that hold and trade the currency, I personally believe it has profound, disruptive implications for a range of settings, from finance, banking, IoT, healthcare, supply chains, manufacturing, technology, government, and the legal system, and more.
On the heels of today’s announcement that IBM is offering a new framework and cloud services to help organizations understand how to create secure cloud environments for blockchain, I want to share some insights from a cryptographer’s perspective about the true innovation behind the blockchain.
If we break it down to its most simple elements, a blockchain is:
A replicated and distributed record of all exchanges and transactions
Strong cryptography to ensure the record is immutable and to ensure privacy of transactions
A shared consensus mechanism that decides on transactions to execute and distribute the power to verify and validate operations over all parties
Business logic for transactions, ranging from recording who owns which asset to executing self-enforcing, complex smart contracts
Consider a service suitable for being executed on a blockchain: an online auction platform; an exchange for a financial asset among banks; a digital land registry. The service maintains a central repository of the assets and executes transactions issued by the participants. All participants accept the market rules and asset holdings decided by the service, such as the price of a deal or the owner of a property.
A platform like this must represent the interests of all participants and customers equally. For establishing a deal between two parties, who both wish to maximize their own benefit, the service should be neutral. Both parties must trust the exchange to correctly implement the rules. Otherwise, if one party would perceive the exchange as unfair, that party would not be interested in using the exchange.
Considering the four features of blockchain, the most important one is the distributed power of establishing consensus, such that every participant considers the resulting outcome to be fair and neutral.
Traditional centralized IT systems cannot provide such fairness because they always contain a single point of failure, in the sense that the organization that controls a database holds the ultimate power over its content.
In practice, dedicated organizations have been formed to operate such exchanges among the participants, such as stock exchanges, national real-time gross settlement systems, or land registries operated by a government on behalf of its citizens. On the Internet, for example, the authority over domain names, a critical element for its operation, has been delegated to a global nonprofit organization.
Therefore the crucial feature of blockchain protocols lies in what is known as distributing trust: directly, to the participants and immediately, over the Internet. Suitable mechanisms are known as consensus protocols and have already been included in textbooks, such as the one I co-authored. Although variants of them that tolerate faults and network uncertainty have been in operation for years, they need to be extended for blockchains, so that they tolerate selfish behavior and attacks by some participants. The novel element is distinct ownership and shared control over the participants. Multiple consensus protocols have been proposed as contributions to the Linux Foundation’s HyperledgerProject, which develops a distributed ledger platform, including one based on Byzantine-fault tolerance from IBM.
Interested in learning more? Join me and my peers on 25 July for a workshop in Chicago.
The annual conference on Computer Vision and Pattern Recognition (CVPR 2019) takes place June 16–20 in Long Beach, CA. IBM Research AI will present technical papers describing our latest results in our quest to give AI systems sight. Our research explores a variety of areas within computer vision and multimedia, with a focus on multimodal […]
Since its introduction in the 1970s, the backpropagation algorithm (backprop) has been the workhorse for training neural networks has and contributed to impressive successes in deep learning for a wide range of applications. Backprop plays an important role in enabling neural networks to track the errors they make, learn from those mistakes and improve over […]