Decentralization is a popular technology trend at the moment. It is linked to edge computing, blockchain and Web3. And it’s happening at a time when many services are very centralized, and in fact are becoming still even more centralized when they are moved to clouds. “Centralized vs. decentralized” is not a new discussion – it is actually something that has gone in different directions for decades. But even if we can compare it to a pendulum, things never come back to the old models.
In the early days of computers, there were only a few mainframe computers. A company may have had only one mainframe from vendors like IBM, Digital Equipment Corporation (DEC), or Hewlett-Packard. Employees who needed access to this computer had dummy terminals that were text-only display and keyboard inputs. Everything revolved around this mainframe computer: data, intelligence and processing power.
Then in the 1980’s, we started to get personal computers and also home computers, like Commodore 64, Apple II, and MSX standard computers. IBM introduced PCs in 1981 and brought to the market Microsoft’s PC-DOS for IBM. We can say that this was the beginning of the decentralization of computing and data storage. PCs came to the office and changed the whole market and computer landscape. Mainframe vendor giants like DEC that could not adapt to the PC-driven market disappeared.
Centralized vs. decentralized web
The Netscape Navigator browser was launched in 1994. While it was an important milestone for the development of web services, Tim Berners-Lee had already proposed the world wide web a few years earlier in CERN, and they were the first ones to create the first implementations of the web. One could claim that the world wide web created a more decentralized structure for data storage and processing. Still, we can also claim that the web contributed to making the structure more centralized – many services were run as web services (as opposed to local programs on a PC). This is also a great example illustrating that the notions of “centralization” versus “decentralization” and related trends are not so straightforward.
Anyway, web services have led to more centralized services. Some internet giants have started to dominate markets and cloud services have got most companies to run many of their services in their cloud and store their data there. Enterprises use a lot of SaaS products. It’s ironic that we now have more and more powerful personal computers and laptops, yet we use them almost like dummy terminals in the 1970s, where most of the intelligence of services, data, and processing occur in the cloud.
The emergence of app markets
Apple created the real mobile app market when it introduced the first iPhone in 2007. Compared to many web services, mobile apps are more decentralized, as they are native software that run in the terminals, i.e. in phones. But of course, they typically also interact a lot with cloud-based services.
I was working with software and network services in the 1990’s, and I remember that I was able to participate in Ericsson’s strategy work when I worked for one of their customers and partners. One important theme and discussion in the strategy work was to create scenarios looking at how much intelligence, data, and processing will be in network and servers versus in terminals. The objective was to evaluate whether future services would be centralized or decentralized.
An additional dimension for the discussion is whether a service is decentralized or distributed. This has been a hot topic since at least 2016 and 2017 when blockchain really started to emerge. True blockchain believers wanted to emphasize that any real blockchain solution must be distributed, not simply decentralized. Now it looks like decentralization is the dominating term.
I wanted to offer this historical summary to illustrate that decentralization is not a new idea, and it is not always so clear what is really decentralized or centralized. This is good to keep in mind while we observe current trends towards greater decentralization. We’ve also learned that it is totally feasible to decentralize services; a centralized model – e.g. SaaS and clouds – is not the only one that works.
There are now many reasons why we are moving again to more decentralized services:
- Blockchain type models enable new business models
- Users need better control over their data and applications, as well as better value from their personal data
- AI and IoT enable more personal and local services where latency, security, and availability are critical
- Technology developments such as more local processing power and federated learning empower decentralization
- People, regulators, and politicians have become more skeptical of centralized solutions offered by a few internet giants.
Although decentralization is now a trend, there are also many ‘old school’ IT people who are skeptical of it and want to list reasons why it is not feasible. But if we look at technology development history and user needs, it is naïve to think that we would somehow stay with centralized solutions and thus arrive at the end of development. Change is constant.
This raises the question of the modalities of decentralization we could achieve. As we see from history, many new steps have had components of both decentralization and centralization. For example, it is very likely that user-held data models will be based in personal clouds – i.e. each user has their own cloud instants, rather than keep data in a terminal or local server. The details of how edge computing will develop are still unclear. Typically the sustainable new solutions are those based on real customer needs and use cases, not only on technology standards. Decentralization must follow that example.