My thanks for this artwork goes to Nicole Forrai

The Small/Big Dualism of Networked, Decentralized Systems

Modern decentralized systems exhibit a little discussed property. I call it the Small/Big Dualism. It is a direct result of the internet’s existence. As I’m going to explain in this piece it’s of major importance when thinking about decentralized systems. While the underlying developments are not new, I haven’t come across the particular framing I present here which is useful when analyzing and developing decentralized systems in our age.

Before the arrival of internet-based information technology, whenever a system was decentralized — as in: the system’s entities were distributed and largely autonomous — all the entities used to be rather uncoordinated. Thus, they couldn’t profit from certain benefits that unarguably stem from being big (economies of scale, power, public awareness etc.).

This has changed. The underlying effect is well known, at least among internet thinkers. Ben Thompson for instance hints at it when he speaks of his Aggregation Theory concept (emphasis added by me):

“Previous incumbents, such as newspapers, book publishers, networks, taxi companies, and hoteliers, all of whom integrated backwards, lose value in favor of aggregators who aggregate modularized suppliers — which they don’t pay for — to consumers/users with whom they have an exclusive relationship at scale.”

This is the effect created by platforms.

If we take a more abstract, analytical look at it, all platforms from Uber to AirBnB, from Facebook to Google aggregate and thereby organize and coordinate autonomous entities. Every single driver on Uber, the smallest entity on its supply side, is an autonomous economic subject. But at the same time, by providing the relevant information — in the case of ride hailing that’s data like current location, vehicle type, availability, personal driver data etc. — the platform or aggregator is enabled to coordinate all the entities.

Note that such a system is predicated on information flowing between the small/entity (driver) and the big/aggregator (Uber). What is limited¹ in Uber’s case but important for other systems of similar type is the flow of information between all the entities.


Networks and Aggregation Change the Game

So, instead of decentralized-yet-very-fragmented systems we are now seeing the emergence of decentralized-but-networked systems.

After this was drawn by my good friend Holger Pfahl (thanks, mate!) I changed the wording to Small/Big but didn’t want to take up even more of his precious time. Thus the discrepancy in the wording.

In there lies the Small/Big Dualism: While all single entities of the system are largely autonomous (i.e. small), they are at the same time part of and coordinated by another entity which functions as the overarching aggregator (i.e. big)². Neither the small/entity nor the big/aggregator is thinkable without the other — that’s why it is a dualism.

This dualism, I argue, has a considerable effect on the way we ought to think about decentralized systems.

Decentralized systems had several advantages even before the internet era:

  1. Great risk mitigation as there are no too-big-to-fail entities. That is, individual errors — which are unavoidable — matter less since they don’t endanger the entire system
  2. The autonomy of the entities allows for more experimentation (due to the small scale of the experiments) and thus, as a rule of thumb, more innovation emerges
  3. Distributed power not only hedges against fatal mistakes (see 1.) but also decreases the impact of its potential misuse. (Note: Decentralization doesn’t per se hinder misuse of power. It only limits its outcome because the scope of power is smaller. Thus, it might make misuse less attractive. But even if it doesn’t: the damage is relatively small)

However, those benefits came at a cost:

  1. Not taking advantage of economies of scale and thus inferior efficiency (but also: not suffering from diseconomies of scale)
  2. Spread of innovation between entities was slow due to missing or slow information-flows
  3. Overarching decisions on the system level were either hard or impossible

The Decentralized-But-Networked Advantage

The Small/Big dualism, however, reduces the cost of decentralization. First off, it allows the overall system to benefit from certain (though not all) advantages of scale. Independent drivers get access to a market that didn’t exist before. Independent developers can access a code database as if they worked at a giant company (likely even better). Local shop owners can sell to a global audience. Group buying, though not a new idea, is made a lot more convenient.

Further, once an (ideally frictionless) information flow is established within the system, the overall system becomes more adaptable. The spreading of innovation, thus, becomes more likely and can happen quicker. Also, the ‘networked’ property allows for new ways to make decisions at the system level. The latter is not yet truly understood (let alone solved) but there is a fair amount of experimentation going on. It ranges from software-based systems like LiquidFeedback (which makes use of proxy voting) or Loomio (focused on consensus building) to concepts like the advice process or similar methodologies used in decentralized organizations.

The result in simple terms: Decentralized-but-networked systems are small and big at the same time.

Thus, systems which exhibit the small/big dualism are very different in nature than decentralized systems of old (=decentralized-yet-very-fragmented). The former profit from the same inherent advantages while also being able to reap benefits formerly only attainable for centralized systems.

That’s huge. It opens the door for decentralization in new areas. If I’m not mistaken, we’re just at the beginning of a transformative phase which will result in many more decentralized systems across domains.


¹ I say limited but one could also argue its entirely lacking. I chose the former because you could argue that Uber does this ‘under the hood’. For instance, when they connect a driver and a passenger this decision implies knowing the location of all drivers and is thus at least an indirect form of shared information.

² Note that in all of my examples the aggregator is a separate entity, usually a company. That’s no necessity though. It’s also possible that the aggregator is collectively owned by the autonomous entities or even itself distributed as in the case of blockchain protocols.


This piece is an excerpt of a book I’m currently writing on decentralized systems. It’s an early draft. I intend to share some passages along the writing process. In the meantime, any feedback is much appreciated (I’m also grateful for reading recommendations)! If you have questions regarding the project please don’t hesitate to contact me.


About me

I work, think, write and speak about digital business, technology and decentralized systems. If you’d like to connect, follow me here on Medium, or check out my website to find out more. I’m always glad to talk & interested in inspiring discussions. My analog residence is Munich, Germany.