By Jeff Stollman
Principal scientist, Rocky Mountain Technical Marketing, Inc.
As I pointed out in a previous blog on decentralization, there is a great zeal in the blockchain community around the notion of decentralizing everything. It is viewed by many in the community as a guiding principal that is vital for any blockchain solution. But the term is used so differently in so many contexts, that it is hard to understand the vision of these enthusiasts. And, I suspect, that much of this enthusiasm is misplaced.
I will attempt to demystify the term by revealing my ?? tenets of decentralization.
Tenet #1: Decentralization is not an end in itself.
Decentralization is not an end; it is a means to an end. Even the most outspoken advocates of decentralization will likely agree that they decentralization is important because it serves at least one of the following purposes:
Transforms the blockchain a trusted intermediary
Serves as a security control to make corruption of the database more difficult
Makes governance of the process more democratic.
Corollary 1A: Blockchain solutions should not be rated based on decentralization; instead they should be rated on their ability to achieve their goals.
Contrary to many popular opinions these ends can be met using approaches other than decentralization. Decentralization is a control technique. But is is only one of many. If the above goals are met, does it matter if decentralization is used as the tool?
Corollary 1B: If the goals for which decentralization is employed can be met without it, decentralization may not be necessary.
Tenet #2: Decentralization is not a single characteristic.
In my blockchain glossary blog, I deconstruct the concept of decentralization into the following independent dimensions:
Architectural Decentralization: the redundancy of the architecture of a data system that provides it with the resiliency to continue operations if at least one node containing at least some of the data becomes inoperable or compromised.
Data Decentralization: the requirement that there be more than one nodes to represent the full content of a blockchain. If there are many nodes, but each node contains a complete copy of the data, the data is not decentralized.
Procedural Decentralization: the requirement that more than one independent validator is required to validate new transactions added to the system
Political Decentralization: the requirement that power to make and change rules for the system (governance) be distributed among the independent stakeholders of the system
There are likely other components that I just haven’t thought about yet. I welcome your suggestions on these others.
Corollary 2A: A blockchain can be decentralized along one dimension, but not another.
The different dimensions listed above are independent of one another. Being decentralized along one dimension does not imply decentralization along any of the others.
Tenet #3: Most blockchains might not be decentralized.
Once we understand that decentralization is not a single characteristic, and we further grasp Corollary 2A that being decentralized in one dimension can be independent of being decentralized along another dimension, concluding that any blockchain is “decentralized” becomes complicated. To qualify as decentralized, do all qualities have to be met? If not what are the rules for making such a determination? My personal belief is that talking about the broad term “decentralization” is a fool’s errand. We need to be specific about which dimension we are talking about in order to have a meaningful exchange of ideas. Each one should be considered a topic unto itself. Perhaps we need to remove the term decentralization from our vocabulary and find substitute terms for each of its various dimensions.
Table 1 provides a comparison of popular blockchains based on the four components of decentralization described in Tenet #2.
Table 1: Component decentralization for five popular blockchains.
| Blockchain | Architecturally Decentralized? | Decentralized Data | Procedurally Decentralized* | Politically Decentralized |
| Bitcoin | YES | NO | 3 | YES |
| Ethereum | YES | NO | 3 | YES |
| Bitcoin Cash | YES | NO | 2 | YES |
| EOS | YES | NO | 12 | NO |
| Litecoin | YES | NO | 3 | YES |
*Minimum number of colluding parties (typically mining pools) needed for a successful 51% attack, based on current mining distribution. The higher the number, the more decentralization exists.
From Table 1, it can be seen that the popular blockchains included in the table are all architecturally decentralized. But none of them decentralize their data. In the blockchains selected, each node has a complete copy of every transaction. This may be beneficial for transactions that are public records such as land titles. It can become problematic when transactions involve personal or confidential data. The pseudonymity of being able to hide ones identity behind a complex account number is commonly broken. The tracking down of Dread Pirate Roberts by the FBI to bring down the Silk Road is a good example. The US government’s recent blacklisting of account numbers of Iranians stealing and laundering money is another. [See “US Blacklists Bitcoin Addresses of Iranians Behind SamSam Ransomware”]
You will have to draw your own conclusion whether the ability of three parties to control a blockchain constitutes decentralization. In the case of all of the blockchains, except EOS, mining pools have grown so large that it only takes two or three of them to collude in order to carry out a successful 51% attack. Such attacks have already been successful on some smaller blockchains. [See Hertig, Alyssa: “Blockchain’s Once-Feared 51% Attack Is Now Becoming Regular” Coindesk.com 08 JUN, 2018]
While I show in the table that all but EOS are politically decentralized, this point can be argued. Most of the public blockchains allow anyone to participate in governance decisions. But in reality, only a handful of developers are typically making decisions for the “silent majority” of stakeholders.
Corollary 3A: It doesn’t matter if a blockchain is decentralized.
What matters is that a blockchain meets its stated objectives for its stakeholders of availability, integrity, fairness, and confidentiality (when applicable). If it uses decentralization as a tool to achieve one or more of its objectives, so be it. But if it achieves the same end using another approach, does it really matter if it is not decentralized?