Some fundamental questions of centralization vs decentralization have nothing to do with blockchains. Not even use cases. Centralized and decentralized processes have alternated for centuries based on trust, cost efficiency and scale considerations. For our business use cases, we equate trust with service levels. If an entity delivers to your expectation of service, you trust them. If they do not, you don’t. Whether your service level expectations include data protection vs data collection expectations, different cost/trust considerations will emerge.
When shared services emerged in businesses in the 1970s and 1980s it was a process of centralization in large businesses. The tradeoff was between localized decision-making and agility vs centralized process efficiency and scale. Shared services as a centralization and standardization effort were easily justified on cost savings (hard metrics) and the loss of local service levels (soft metrics) were to be mitigated through change management and retraining. The same tradeoffs were seen in various public sector areas from education through public safety. Banks in the middle ages emerged because localized moneychangers could not compete at the regional or global scale of a centralized entity like the Medici.
In times of business disruption, recession or crisis, decentralized processes tend to emerge for the same use cases. In times of hyperinflation local trading and barter (decentralized currency) apprears. When service level (trust) in a centralized business or institution falters, decentralized alternatives emerge in all areas from airlines through financial institutions.
Industries that have inherently decentralized infrastructure (education, internet, trade, communication) tend to do a better job routing around disruption of service levels (trust).
Blockchains have an inherently decentralized infrastructure. There are clear cost and speed efficiencies of centralized (permissioned) blockchains. Some are designed to fulfill narrowly defined service levels of their constituents and therefore will accomplish trust at low cost and high efficiency (Ripple may be such a case). If the use case is about security or process integrity, decentralization will likely win.
If a blockchain fails to deliver on its expected service levels it will no longer be trusted and more decentralized alternatives will emerge. On the other hand, as long as we live in a commercial society, cost/speed efficiencies will be natural forces to push towards consolidated or semi-consolidated business models. The push and pull of cost and trust will create several solutions for the same use case. Some will be lower cost and some will be higher trust.