Blockchain: MultiChain's Gideon Greenspan talks about the future of peer-to-peer databases

Coin Sciences' MultiChain blockchain platform is collaborating with Wolfram Research, which created the powerful Mathematica system.

Initially, Wolfram will offer users direct access to a zero-configuration private blockchain accessible in the Wolfram Cloud, and will also allows users to integrate their own blockchains running local nodes. The MultiChain integration is slated for release with a Wolfram language update later this year, said a statement.

Today's announcement was put in motion some time ago, said Dr Gideon Greenspan, CEO, Coin Sciences. Wolfram had been looking into blockchain technology and decided to provide it as a feature so Mathematica users could create shared datasets. "From our perspective this is a really nice integration with a great piece of software from a well respected company," he said.

Greenspan has always viewed blockchains as a general purpose technology for creating peer-to-peer shared databases. Despite the focus they have received in the finance sector, blockchains are equally useful for decentralised data aggregation and collaboration in many other fields, such as science, engineering, economics and government.

"Blockchains are ultimately a database technology, and I have been saying this for a long time. A lot of people think of them as fintech but I don't agree. While blockchains can obviously be used for financial purposes, they're ultimately a new type of shared database, which doesn't require a central intermediary. Blockchains can therefore be used pretty much anywhere that databases can, and there's nothing specifically financial about that."

Like many other companies, Coin Sciences started developing MultiChain with a specific financial use case in mind, which was using blockchains to settle transfers and exchanges of assets between financial institutions.

"An interesting thing has happened." said Greenspan, "Companies like R3CEV and Digital Asset who are focused on that use case have concluded that blockchains aren't the right tool for them, or are at least only a small part of their overall solution. They now talk in more general terms of blockchain-inspired architectures and distributed ledger technology.

"The other way to go, which we've done as a company, is to say: let's not restrict ourselves to a particular use case, and instead focus on the technology, which is a bona fide new type of database. Let's see where it can actually be successfully applied, in finance but elsewhere as well. So these are two legitimate directions leading to two very different places."

In a recent blog post Greenspan said that, while there has been much focus on things like smart contracts running in virtual machines, the more interesting abstractions for blockchain platforms are yet to be imagined. In this regard he asked what will be the blockchain equivalent of foreign keys in relational databases or map-reduce in big data stores.

"If you look at any particular type of database –relational databases, in-memory databases, or noSQL databases – these products and platforms developed their own languages and abstractions, which map well to how they technically work and their most common use cases. This leads to simplicity and productivity for the developers who are building on top of these platforms."

"Take the example of a relational database, where you have multiple tables of information which have relationships between them. As the relational paradigm was developed, the notion of a foreign key became popular, which lets you state that every entry in one table must have a corresponding entry in another. For example, a university's database shouldn't store an exam result for a person who doesn't appear in the list of students. This is a very natural way of thinking about relational databases, where tables relate to each other, and we want to define some rules to govern those relationships.

"Blockchains are a new type of database which haven't been productised before, so the same type of process needs to happen again. Over time we are learning how our product should look and what types of functionality it should provide."

"We've got some ideas, some of which we have released, some of which we've written about, and others we're not yet talking about openly. To be frank the process will probably take five to ten years as we learn more about how people are using blockchains and what they want to do with them. Our long term goal is for MultiChain to support the widest possible range of applications."

There are groups looking to create standards for blockchains, perhaps driven mostly by the potential of financial applications, but of course it's early days. Greenspan pointed to SQL as an example; while it's not a single perfectly unified standard, it's very similar between different platforms and so developers know what code to write and what results to expect.

"Of course blockchains don't have this kind of standard yet and you could say that's a problem," said Greenspan. "But it's also inevitable at this stage of the technology's development. We don't yet know what functionality blockchains are meant to provide, and it's impossible to define a practical standard until we do.

"If you look at the early development of other technologies, there were always multiple independent vendors going off in different directions. Some platforms win and some platforms lose; some paradigms win and some paradigms lose.

"Once the picture starts to become clear, that's when there should a standardisation effort, so multiple platforms end up working in similar ways from the perspective of application developers.

"We are seeing many consortia and groups of companies working together on specific platforms, but to me that's not the most valuable type of standardisation. What's more interesting is creating rules that allow different products to work together, to connect together, exchange transactions with each other and be programmed and managed in a similar way."