Private blockchains v centralised databases - what's the difference?

Ethereum recently wrote an interesting blog explaining the difference between public and private blockchains. However, I believe there is a much more fundamental question that needs to be answered – what does a distributed consensus ledger (DCL), such as a blockchain, enable that cannot be achieved with commonly used technology such as a centralised database, whether open to all (public) or under private control?

I have seen little that comes close to answering this question, but I am convinced it needs answering before we can be clear on which DCL use-cases will succeed. Claims that DCLs will disrupt Financial Services beyond recognition, saving tens of billions of dollars in costs while reducing risks and improving services, strike me as mere hypotheses until proven through hard evidence and analysis.

I don’t have an answer to this critical question, but I can take an initial step by observing differences between a centralised database with open access, and a DCL.

Some of the unique features of a fully public DCL I can identify are:

• Control of the ledger is distributed across all those that use it, and requires consensus among its users to make changes – a centralised database is under the control of a central authority
• Users of DCLs have full independent control over their assets/data on the ledger, those of a centralised database are dependent on the control given to them by the central authority
• The ledger exists in multiple copies which makes it potentially more resilient than a centralised database
• Data on DCLs is transparent to all – you may not be able to decipher all of it, but you can check that it never changes; on a centralised database, access to data on it can be restricted through rules set by the central authority
• On a DCL, a user’s data is uniquely identified by the owner’s private key, created cryptographically by the owner – on a centralised database, keys to identify data do not need to be created  or enforced cryptographically
• Permissionless participation is possible on a DCL – anyone can join and use the ledger; on a centralised database, only users allowed by the central authority can participate
• Similarly, permissionless innovation is possible on a DCL, anyone can innovate using it – on a centralised database, innovation is subject to central control
• On a DCL, historic transactions are unalterable and permanently accessible (unless changes are allowed by consensus) – on a centralised database, historic transactions can be changed by the central authority
• DCLs are borderless – they cannot be governed by national rules/laws including rules on data residency (although users of the DCL can still choose to adhere to national laws); centralised databases have centralised authorities who can be made to conform to national laws and regulations
• DCLs are inefficient users of computing resources – for example the cost of mining (running “proof of work” processes), the inefficiency of running multiple ledger copies and the unwieldy data structure of chained transactions – in contrast, centralised databases can be run very efficiently, and use relational database structures that are flexible to the needs of applications and reporting requirements.

I am sure there are other differences between DCLs and centralised databases, but the overriding one to me is that of control. A fully public DCL such as Bitcoin has no controlling authority, it is governed by consensus of its users, and its rules are enforced by cryptography. A centralised database has a controlling authority, which can enforce rules through software logic, but it has no specific dependency on cryptography.

Therefore, it strikes me the use-cases where DCL s will succeed are those where distributed, evenly balanced  control can make a difference, for example, where a central authority is not feasible – typically where cross-border transactions occur, such as in international payments; or where centralised controls  exist that create unnecessary inefficiencies, set up by, for example, intermediaries.

It is less obvious why DCLs are better suited than centralised databases to use-cases where central authority is possible and necessary e.g. land registry, car registration; and it is also less obvious to me why so-called private DCL s, or private blockchains, whether internal to an organisation or shared between a closed group of consenting organisations, are any different to a centralised database where control starts to look more centralised than distributed.

Non-cryptographic technology already solves for requirements such as reach, speed, access, authentication, authorisation, validation, risk, resilience and codified (“smart”) contracts, and has done so for years. However, it is the use of cryptography for control in DCLs that is the real game changer – by enabling control through consensus across a disparate group of entities; and by bypassing control barriers set up by middlemen, rent seekers and gatekeepers who create  inefficiencies, risks and unnecessary costs, sustained by their own self-interests and by the inertia of legacy business models.

There is clearly a lot of hype in the world of DCLs and blockchains, with 100s of millions of dollars being invested in FinTech companies to exploit them, with senior bankers leaving well paid jobs to join them and with banks competing to announce DCL initiatives.

We may be in a classic hype-cycle, where today’s high expectations are followed by a trough of disillusionment before we start seeing the true potential of DCLs. It will be very interesting to see what developments come from these FinTech companies and the bank innovation labs over the next 12 months - hopefully, we will soon start seeing breakthroughs leading to a clearer understanding on the DCL use-cases that will succeed and the true scale of benefits they will bring.