At one point in the not-so-distant past “decentralization” was a common buzzword traveling in the company of blockchain. However, many of the more recently established blockchain projects differ from the past in the various sorts of centralization added to their architecture. In the following I will give a couple examples and then a couple guesses at the forces driving this trend.
Examples
EOS:
The cryptocurrency EOS and blockchain protocol EOS.IO have garnered a lot of attention for their large and successful ICO, their technical innovations, and the (possibly undesirable) centralization of power those innovations seem to produce. This system grants the ability to confirm transactions to a fairly fixed, limited number of “block producers” and empowers arbitrators to alter or delete transactions in the event of disputes. This article describes a dispute around a confirmed transaction, supposedly the result of a phishing attack, that was negated by an EOS arbitrator. For our purposes, the quote from CTO Daniel Larimer says it all: “Decentralization isn’t what we’re after.”
Amazon’s Two Blockchain Products:
Amazon recently announced, to much fanfare, a pair of services for building blockchain networks on AWS. The first is Amazon Managed Blockchain which allows the construction of blockchain platforms using Hyperledger Fabric or Ethereum, although the latter and more decentralized of these is tellingly not yet ready. The second is the Amazon Quantum Ledger Database which is approximately a conventional centralized database with some simulated immutability properties that is intended to provide an interface point to Amazon’s other blockchain product. Notable here is the appearance of a centralized database product with limited emulation of blockchain properties and the implicit suggestion that many customers would prefer to interact with this technology and a blockchain technology directly.
Drivers Good and Bad
Both of the following are mostly about large corporate customers, themselves centralized institutions, and their efforts to accommodate a disruptive technology.
Misguided Buzzword Lust: ”Blockchain” is a buzzword and nearly a cultural phenomenon of considerable power and limited correlation with blockchain the technology. There is certainly an appetite out there to work with a blockchain product so that one can say that one is working with a blockchain product… but perhaps some of those who hunger have poor understanding of what they are really buying. There is certainly a market out there for something that can be called a blockchain in the boardroom but otherwise functions in a more familiar way as a centralized database.
Some-not-at-all Requirements: Blockchains are really several pieces of functionality bolted together, particular a consensus process and a game theory process occur across multiple actors to produce immutability. Some of the savvier customers have realized they have need for one of these and not both, or that they need them together in different configuration. Many firms struggle with consensus problems under other names and have embraced the articulation of this idea on its own but have limited need of immutability and its associated costs. Still others are looking for something like a cryptographically hardened file synchronization system - immutability is in fact the goal, but the circumstances where it might be threatened are different.
The expectation around technical progress in blockchain, in general, should be that blockchain is increasingly broken down into its constituent parts and refactored, sometimes for sales reasons to meet customers where they are and other times to meet genuinely different needs.