#ChainReaction: Notes on Centralized, Decentralized, and Distributed Systems

This blog post is the first in a series of blog posts I will write as part of my effort to take an inventory of what I am learning about supply chains, digital tokens, and distributed ledger technologies.

I expect these blog posts to be frustrating for most people to read because I suspect they will come-across as disorganized, and confused. That is a reflection of the complexity of the topics I am trying to learn.

If you feel I have got something completely wrong, please do not hesitate to let me know. As Marcus Aurelius puts it;

If anyone can refute me—show me I’m making a mistake or looking at things from the wrong perspective—I’ll gladly change. It’s the truth I’m after, and the truth never harmed anyone.

First, some context; I am a seed-stage VC who has been studying supply chain for sometime. I believe that the greatest technological shifts of the next 3 or 4 decades will happen at the intersection of supply chain, industrial processes, data and analytical decision-making. I believe this shift will transform the way global supply chains function in many different industries.

If you follow technology and business news then you know what some of the trends are that will lead to the kind of shifts I believe we are about to witness. They are; increasing efficiencies in industrial automation, exponentially faster, more powerful, and cheaper computing technology, the proliferation of electronic sensors capable of capturing large amounts of data in almost any industrial or non-industrial setting one can imagine, software that is capable of analysing huge troves of data in order to aid people in making decisions about complex processes and systems, and ubiquitous computing. The list goes on. A more recent addition to any list of ground-breaking technological developments is Bitcoin and its related technologies, including the Bitcoin blockchain, as well as other cryptocurrencies and their accompanying blockchains or distributed ledger technologies.

There is currently a lot of ongoing enthusiasm, and perhaps, even hype, about Bitcoin, the Bitcoin blockchain, other cryptocurrencies or digital tokens, and their accompanying blockchains or distributed ledger technologies. Mainly, the excitement is around the belief that this group of technologies has the potential to “disrupt” any number of existing business or social structures. Personally, I agree with the following statement by Marco Iansiti and Karim R. Lakhani;

True blockchain-led transformation of business and government, we believe, is still many years away. That’s because blockchain is not a “disruptive” technology, which can attack a traditional business model with a lower-cost solution and overtake incumbent firms quickly. Blockchain is a foundational technology: It has the potential to create new foundations for our economic and social systems. But while the impact will be enormous, it will take decades for blockchain to seep into our economic and social infrastructure. The process of adoption will be gradual and steady, not sudden, as waves of technological and institutional change gain momentum. ((Iansiti, Marco, and Karim R. Lakhani. “The Truth About Blockchain.” Harvard Business Review. February 17, 2017. Accessed February 04, 2018. https://hbr.org/2017/01/the-truth-about-blockchain.))

If you agree with the preceding statement, then you should also agree that, perhaps, before one dives into the intricacies of digital tokens and distributed ledger technologies it is useful to study centralized and decentralized systems in a broad, general sense. Therefore, though I will ultimately migrate to discussing centralized systems, decentralized systems, and distributed systems in relation to information technology systems, at the outset I am thinking more broadly in terms of social structures that exist in economic, political, and cultural organizations.

At the end of this process, I hope to have developed a good frame of reference for understanding why and how digital tokens and distributed ledger technologies will combine with other prevailing advancements in technology to cause the transformation in global supply chains that I believe is upon us. I hope this helps me see what is coming next – in a manner of speaking, and that the knowledge I will develop in the process helps me make better investment decisions.

If you have read any articles that discuss Bitcoin and its accompanying technologies, then you will recognize the recurring themes of centralization versus decentralization. So perhaps the place to start is in understanding when centralized structures should be desired and maintained versus when decentralized structures should be desired and maintained.

The following discussion is motivated by, and borrows heavily from, “Centralization and Decentralization: The Compunications Connection” by Stephen H. Lawrence. ((Lawrence, Stephen H. “Centralization and Decentralization: The Compunications Connection.” Accessed February 4, 2018. http://www.pirp.harvard.edu/pubs_pdf/lawrenc/lawrenc-i83-2.pdf. I am basically paraphrasing pages 6 – 26.)) In that paper there’s a quote from “The Computerization of Society”, a report prepared for the French Government by Simon Nora and Alain Minc;

It allows the decentralization or even the autonomy of basic units. Better still, it facilitates this decentralization by providing peripheral or isolated units with data from which heretofore only huge, centralized entities could benefit. Its task is to simplify administrative structures by increasing their effectiveness and improving their relations with those under their jurisdiction. It also allows the local municipalities more freedom. It reinforces the competitiveness of the small and mid- size business vis-a-vis the large enterprises.

Centralized Systems

A centralized system is a system in which a master-node makes decisions or performs systemwide functions on behalf of all the other nodes within the system – subordinate-nodes. Subordinate-nodes only follow instructions issued by the master-node. It should be obvious that centralized systems depend on a reciprocal relationship of trust between the master-node and every subordinate-node. Centralized systems are also described as command-and-control systems.

Advantages of Centralized Systems

1. Returns to Scale: Centralized systems generally benefit from increasing returns to scale, meaning that the system generates outputs at a rate that is proportionately greater than the rate at which it consumes inputs. More specifically, the value of a centralized system’s outputs should be proportionately more than the value of the inputs consumed by the system. This happens because resource-intensive decisions and functions can be performed by the master-node only, without burdening the entire system with performing those same functions. As a result, as the system grows, the per-capita system costs can decrease substantially. Increasing returns to scale are generally closely associated with increasing efficiency.
2. Optimization: It is easier to optimize the outputs of a centralized system given a set of inputs because the effort that goes into optimizing the system’s output need only be expended by the master-node and not by every node within the system. As a result, in a centralized system optimization contributes to the system’s overall efficiency.
3. Standardization or Uniformity: The hierarchical structure of centralized systems makes it easier to maintain standardization or uniformity within the system. Such standards are determined at the level of the master-node, and then they are implemented and enforced at each subordinate node according to rules established and maintained by the master-node. Standardization and uniformity ensures that the entire system operates as one unit, rather than as a collection of disparate, non-uniform, non-standardized entities. In certain instances, standardization and uniformity may be especially useful qualities if the system is to serve its intended purpose.
4. Criticality or Importance: A centralized system is preferred when there is a disproportionately high cost associated with the commission of errors or mistakes at the level of a subordinate node. In other words, centralized systems are prefered when the weight of responsibility for avoiding mistakes is high, and the costs of this responsibility are borne by the master-node.
5. Coordination & Interdependence: Centralized systems perform better when one must account for economic externalities. An economic externality is a positive or negative consequence that is borne by an entity which did not participate in taking the actions that led to that outcome. In other words, it is easier for the master-node in a centralized system to also account for systemwide externalities before choosing an action that is implemented by all the subordinate-nodes in the system.

Disadvantages of Centralized Systems

1. Information Overload: Centralized systems can experience breakdowns in systemwide performance if the master-node experiences an information overload.
2. Compulsion: Centralized systems are associated with bureaucracy and lack of freedom – from the perspective of subordinate-nodes. For example, centralized systems do not freely admit new nodes to the system unless such nodes are first approved by the master-node.
3. Lack of Flexibility: Centralized systems are characterized by an inability to respond with agility and flexibility in the face of changing conditions. This can make centralized systems more fragile in the face of threats to the entire system.

Decentralized Systems

By contrast, a decentralized system is one in which there is no single master-node issuing systemwide instructions that subordinate-nodes must follow. Rather, in a decentralized system every node is responsible for its own decision-making and, is capable of taking whatever actions its independent decisions require it to take relative to agreed systemwide goals. It should be obvious that the trust-relationship in a decentralized system differs from that in a centralized system in an important way.

A decentralized system is one which requires multiple parties to make their own independent decisions.

– Rohit Khare

Advantages of Decentralized Systems

1. Impartial Standards: Decentralized systems are better suited when the emphasis is on effectiveness rather than efficiency. As a result decentralized systems tend to exhibit standards that stress the results that each node in the system produces and how those results contribute to overall system wide goals rather than how each node accomplishes the desired results.
2. Initiative/Innovation: Since each node in a decentralized system is free to independently experiment with an eye towards maximizing system wide outputs, there tends to be a higher degree of innovation within decentralized systems. Once a superior method of accomplishing systemwide goals has been identified by one node within the system, other nodes will quickly copy that method if it increases their wellbeing. All else equal, this will lead to a higher level of system wide output.
3. Responsiveness: In decentralized systems, individual nodes are more responsive to local conditions. This is because each node in the system is free to determine local priorities on an ad-hoc basis given information available to that node even if this information is not available to other nodes within the system. It is not difficult to see how this quality of decentralized systems contrasts with the standardization/uniformity quality that is present within centralized systems.
4. Simplified Decision-making: Decentralized systems exhibit a simplified decision-making relative to centralized systems. This is because for a given situation, decisions can be made by only the relevant subset of nodes within the system while  non-relevant nodes conserve system resources. In such a situation, simplified and localized decision-making is an advantage of non-relevant nodes are not adversely affected by the decisions that have been made, and the resulting actions that have been taken, by relevant nodes.
5. Minimize Information Resource Requirements: A decentralized system could be designed such that each node only processes information relevant for its role within the system. This way, systemwide resource requirements can be minimized since each node conserves resources by focusing only on information and activities relevant to its specific functions and does not concern itself with matters outside that sphere of relevance.

Disadvantages of Decentralized Systems

1. Duplication of Effort: Decentralized systems can be designed such that each node within the system attempts to solve similar problems as other nodes in the same system – leading to duplicated effort. It is easy to see how this can lead to more waste than one would observe in a similar, but centralized system.
2. Suboptimization: In decentralized systems, a single node or a subgroup of nodes, might decide to pursue activities that increase their own well being at the expense of the well being of the entire system. Trade-offs have to be made within a decentralised system to ensure that suboptimization is minimized by keeping incentives between all the nodes within the system aligned with one other, and with the entire system as a whole.
3. Less Amenable to Standardized Change: Since each node is responsible for making its own decisions and taking actions independent of a master node, standardization takes a much longer time to diffuse through, and become adopted by the nodes within a decentralized system. As a result decentralized systems characterised by a lack of uniformity, whereas centralized systems are characterized by systemwide uniformity.

In a quest to find examples of decentralization in action within organizations that I am somewhat familiar with, I went looking for a book that discusses the topic. I found that in The Starfish And The Spider: The Unstoppable Power of Leaderless Organizations, a book by Ori Brafman and Rod A. Beckstrom, where they  introduce us to the major principles of decentralization; ((Brafman, Ori, and Rod A. Beckstrom. The starfish and the spider: the unstoppable power of leaderless organizations. Portfolio, 2006.))

• When attacked, a decentralized organization tends to become even more open and decentralized.
• It is easy to mistake a decentralized organization for a centralized organization because we are far more accustomed to centralized organizations. It is also easy to vastly underestimate the power of decentralized organizations.
• A decentralized system does not have central intelligence; the intelligence is spread throughout the system. As a result the best information and knowledge is located at the edges of the organization, close to where things are actually happening.
• Decentralized, open systems can easily mutate.
• Decentralized organizations can seemingly appear out of nowhere because they can mutate so quickly, and because they are easily overlooked at the outset.
• As decentralization takes hold within an industry, overall profits decrease.
• The power of decentralization comes from the phenomenon that when people are put into a decentralized system they automatically want to contribute, and their contributions are usually remarkably of a high quality relative to what one might find in a centralized system.

So far I have not said much about distributed systems. Think of a distributed system as a hybrid between a fully centralized system and a fully decentralized system. Businesses that blend the best of both types of organizational architecture in their business model are not that uncommon, and when they do so successfully the results can be overwhelmingly successful . . . But, we can discuss that another time.

In my next post, I will more directly delve into cryptocurrencies and distributed ledger technologies. Till then, you may delve further into this topic by reading Chris Dixon’s “Why Decentralization Matters“.