What drives decentralisation?
Some rough thoughts on network decentralisation and the forces driving it
The degree of decentralisation in a system is dependent on:
The minimum resources needed to be a viable participant in the system
The ease of coordinating participants in the system
If a system requires economics of scale, due to high fixed costs and large upfront investments, to become a viable participant, it will be more centralised as fewer participants have the resources to join the network. If a system is difficult the coordinate, it will lead to a centralised institution(s) that can bear the costs associated with coordination.
The impact of minimum resources
Compare and contrast the road vs rail network. The road network is incredibly decentralised compared to the rail network.
Trains and railways require more resources to build, so a station and rail line need more traffic to be a viable participant in the network. Stations were only built where viability was reached with a large, local population, and lines were only built between two destinations that would have enough passengers.
Compare this to roads. All you originally needed was two feet or a horse, or now a car; anyone could be a viable participant. The network decentralised to a much greater degree as everyone could participate and roads were cheap to build. The minimum resources required to join the network are much lower and the network expands to bring you in. On the other hand, the resources required to lay a track and own a train are significantly higher; which is why we don’t have a personal train station with our private train on our front lawn.
This doesn’t just apply to hardware and physical infrastructure. For example, Substack and other tools are making publishing even easier and further reducing the minimum resources required to be a viable participant in the publishing network; you no longer need to set up your own website and figure out everything to do with publishing content on the internet.
Coordinating technologies
We have seen a lot of decentralisation over the last 20 years as the internet and software have reduced the challenges of coordinating network participants.
Anyone with a car can connect to Uber’s coordinating system
Anyone with a spare room can connect to AirBnB’s coordinating marketplace
Anyone who can write can publish their content to the internet
Anyone can participate in a crowdfunding campaign through AngelList, Republic and other technologies that reduce the costs of coordinating multiple investors
The coordinating layer needs to provide trust
In the financial system, the minimal viable participant is $0.01. Right now, anyone can participate and the technology exists to do so (we can verbally agree to lend money or pay for things). Trust hinders this. If I don’t believe someone will pay me back or I cannot enforce our agreement, I am not willing to participate in a decentralised, P2P financial network. I want my transactions to be mediated through a trusted institution, such as a bank, and backed by the weight of the justice department.
As this trust gets built into the coordinating layer, smaller participants are able to join and we no longer require large institutions to mediate on our behalf. I could lend money to anyone, but may not get it back or be able to enforce the contract. Paypal and other P2P networks provided this trust. I could hitch a ride with anyone, but I open myself up to the risks associated with that. Uber provided this trust.
Furrow failure - know which force is the limiting factor
I’ve spoken extensively about the reasons for failure at Furrow. However, I think they can be summed up as this: what I thought was a coordinating problem was actually a minimum viable resources problem.
I thought the reason farmers weren’t able to sell direct to consumers was predominantly due to the challenges in coordinating such a network. What I discovered was it is actually due to the minimum resources needed to participate. To ship produce, you need things like a warehouse, trucking etc. These all have upfront costs that require a certain volume of goods to be viable. This isn’t viable for farmers and so the network centralises as they rely on wholesalers and distributors to aggregate into a viable scale.
The system wasn’t ready to decentralise further as the minimum resources needed to participate were still the same. Our existing, more centralised supply chains are set up around this knowledge and meet the minimum volume of goods to be viable. Until a tool or technology comes along that reduces the minimum resources needed, we won’t see more decentralisation in our food system.
Decentralisation in the electric grid
Let’s look at these forces in a real-life example that is actively decentralising.
In the electric grid, the resources needed to become a viable participant are decreasing rapidly. To generate electricity, you no longer need to build a utility-scale power plant. Rooftop solar and residential batteries mean far more people can become participants in the network. As costs fall further, even more people will have the resources to become a participant in the network; which will enhance the flywheel as learning and cost curves continue to decrease.
However, the grid is struggling to coordinate these participants, which is why P2P and prosumer relationships on the grid haven’t fully taken off yet. Until utilities are able to successfully and safely coordinate these participants, we will continue to see less decentralisation in the grid than is possible.
Value accrual
The value in the network accrues to two layers; the network coordinator and the tool/technology providers.
The coordinating layer benefits from network effects. The more participants it can coordinate, the better job it can do. This is why the coordinating layer tends towards becoming a monopoly, or even a publicly owned network - think about roads, railways, even some of the electric grid.
On the other hand, a new tool or technology that enables easier participation in the network, such as Substack or solar manufacturers, will see an initial boom as participants flood to use their technology and participate in the network. However, this technology usually gets commoditised - how many car and solar manufacturers are there, how many blogging tools are there, how many phone makers are there - and in the long run, the technology gets competed into a smaller share of the market.
This isn’t to say there isn’t money to be made in building tools that reduce resource requirements for participation. Although they may not become monopolies, the market is usually larger. Owning a smaller percentage of a larger market isn’t necessarily a bad thing.
Where will decentralisation occur next?
What tools/technologies are emerging that reduce the minimum viable participant?
SMRs, rooftop solar, batteries
Manufacturing - Hadrian and other companies reducing the minimum viable factory
VTOLs / small aircraft reducing resource requirements for airports
Pipedream and delivery
Autonomous vehicles
Autonomous warehouses?
Charter cities?
Most hardware innovations are about reducing the resources required to become a network participant. The recent tech pivot to hardware, when viewed through this lens, is the latest in a long trend of decentralisation - now decentralisation through coordinating software is stalling, investors are looking to the other side of the equation to drive further decentralisation.
Where is a lack of coordinating technology preventing existing resources from transacting/participating in/creating a network?
Distributed compute and storage; see Gensyn
Other underutilised assets
P2P finance / bitcoin
Devolved government / DAOs
Blockchain and distributed trust/validation may be key here