Fundamental Principles of Network Economics

Fundamental Principles of Network Economics by Arie Trouw

With the advent of blockchain networks like Bitcoin, Ethereum and many others, the need to understand some of the fundamental challenges of network economics becomes much more important. In many cases, simple conclusions have been arrived at for these very complex systems and people have used those conclusions to make important technology and financial decisions.

There are several Fundamental Principles of Network Economics that drive the viability of these networks and are greatly under discussed, or in some cases possibly not discussed at all.

Primary Use Problem (PUP)

Conservation of Cost (CoC)

Network Relative Efficiency (NRE)

Deflationary Network Native Currency

Inflationary Network Native Currency


Primary Use Problem (PUP)

This is because the high-value use case can consume the entirety of the network’s resource at a price that is higher than the break-even point of the next best use case.

For example, you can create a chess dApp on Ethereum that has a market where people are willing to pay $1 per game, which translates to around 1 cent per transaction, but compared to various other use cases, the price of gas on Ethereum will never approach 1 cent per transaction, even if the supply of transactions is increased by an order of magnitude. DeFi, for example, will then just continue to find uses for the additional resources at a higher price. More specifically, DeFi has made the price of a Crypto Kitty contract call prohibitive and has killed nearly all other dApp usage.

The Primary Use Problem applies to any multi-purpose network that has an artificially restrained supply of resources. For example, in most blockchain networks, there is a maximum block size and preset block time. This limits the number of transactions that can be done per unit of time, regardless of demand. For the network providers, this is a good thing in that it makes the buyers compete for the available inventory, much like bidding on an advertising network such as Google Ads.

A major difference between a network such as Ethereum and Google Ads is that the processing of blocks on Ethereum is a generalized commodity, so the value of a transaction is equal to all users, but on Google Ads, the inventory is very fragmented, by location, demographics, and intent. The Primary Use Problem only exists on Google Ads when there is a sub-network of resources such as clicks for people searching for ‘bitcoin’.

The solution to the Primary Use Problem is to not artificially constrain resources on the network, but there are security and stability side effects of this. For example, if Ethereum were to increase frequency of blocks 1000 fold or increase the maximum size of blocks 1000 fold, attack vectors on the network would increase and the cost of running the network would increase.

Conservation of Cost (CoC)

The tcO is a net outflow of value from the network, including the host of hardware, power, and networking. An equivalent (tcU) inflow of value must exist for the value of the native currency to stay constant.

The tcO usually requires a native to non-native currency exchange (ETH to USD for example) and the tcU usually requires a non-native to native currency exchange (USD to ETH for example).

tcO >= tcU

Operating a network has fixed overhead and variable transactional costs. As a network scales, the transactional costs drive the cost for the use of the network.

For operators of a network to be willing to operate the network, the amount that they are paid must be more than the cost of running the network. Additionally, if they have alternative networks that they could be running, assuming limited resources, then they would choose the network that provides them the highest return for their effort.

A user of a network chooses a network to use for their purpose based on similar heuristics, including speed, security, and cost.

Currently the outflow of value from the Ethereum network in the form of operational costs is about $10M per day, which means that a new $10M a day needs to flow into the ETH market to keep the value of the currency constant. This could be justified based on the value of the functionality of an Ethereum contract and that users at large may be willing to pay $10M per day for this service.

For Bitcoin this is more serious. The outflow of cost for Bitcoin is higher than $10M per day for tcO, which means that the equivalent amount has to flow in via a combination of users paying for transaction fees and speculation.

Network Relative Efficiency (NRE)

N = nodes in network

DTc = Data Transfer cost
DPc = Data Processing cost
DSc = Data Storage cost

DTrf = Data Transfer Redundancy Factor (based on network size)
DPrf = Data Transfer Redundancy Factor (based on network size)
DSrf = Data Transfer Redundancy Factor (based on network size)

tcO = (DTc * DTrf) + (DSc * DSrf) + (DPc * DPrf)

Ethereum Estimates:

N = 10878

DTc = 1
DPc = 100 (processing costs 2 orders of magnitude more than transfer)
DSc = 10000 (storage costs 2 orders of magnitude more than processing)

DTrf = N * N
DPrf = 2 * N
DSrf = N

Centralized Cost Equivalent = 10101

Ethereum Cost = 118,330,884 + 2,175,600 + 108,780,000 = 229,286,484

Ethereum Cost for Decentralization = 22699x vs. centralized

By this estimation Ethereum is currently more than four orders of magnitude less efficient than an equivalent centralized network.

Based on Conservation of Cost, these costs must be passed on to the users of the network.

Deflationary Network Native Currency

Nearly every blockchain network to date is based on a deflationary currency. Even though ‘new’ native currency is being created every time a block is mined, the total eventual pool of that currency is known at the start of the network’s life and is based on a finite algorithm. Added to this, the native currencies also experience attrition over time due to lost wallet keys or bad transfers that send the currency to unreachable parts of the blockchain.

In the case of Bitcoin, some estimates are as high as 4% of supply per year for networks such as Bitcoin. The amount of BTC added to the accessible pool is approximately half that, at 2%, and decreasing as time goes along. This means that there is a natural 2% interest rate (assuming the above estimates) that is built into just holding BTC, assuming flat demand. If you assume growth in use/demand for the currency, that number goes way higher. For Bitcoin, this is not an issue as such, since the single use for Bitcoin is as a currency and at most, the cost of transactions will increase as the cost of BTC increases. This can be mitigated by increasing block size or other adjustments. Furthermore, the cost of transactions is constant in relative terms to BTC even though it may not be relative to external values such as USD.

In the case of Ethereum, the dynamics of the deflationary currency are very similar to Bitcoin, but since the value of use cases for dApps are generally disconnected from the value of ETH, the upward deflationary pressure of the native currency has a substantial negative impact on the viability of a given dApp. Despite this, the natural market of transactions should overcome this, and normalize to the cost set by the Primary Use Case.

In both these cases, the miners receive the block generation subsidy plus the transaction fees that are associated with that block. The goal is that as the subsidy per block decreases, the transaction fees will stabilize and become the market setting force and cover the cost of running the network.

Inflationary Network Native Currency

EOS is an example of this, but not enough time has passed to determine whether or not this approach works as well or better than a deflationary native currency.


In the case of Ethereum, the true cost of operation is very high vs. centralization (several orders of magnitude), and the primary use of Defi is currently pricing out nearly every other use of the network. It is possible that Ethereum will become a de facto single use network for Defi, which opts it out of the Primary Use Problem, since there will be only one use, much like Bitcoin, or a way is found to remove the artificial restraint of supply (Ethereum 2.0?). Even if the artificial supply cap is removed without sacrificing security and stability, then the next dApp killer will be the absolute cost of operation. Even with Proof of Stake and other plans for Ethereum 2.0, the absolute cost of Operation will certainly still be at least one order of magnitude higher than that of a centralized system, and due to the Conservation of Cost principle, the cost of using the network will still severely limit use cases.

Vitalik Buterin has been quoted as saying that ETH 2.0 would reduce the Ethereum network energy use by 99%, and even though that is a huge step forward, that would still make the Ethereum network 2 orders of magnitude less efficient instead of the current 4 orders of magnitude.

“Are people willing to pay even 100x for the same functionality, just decentralized?”

In my opinion, the only way that a general purpose decentralized blockchain network and the dApps relying on them will become viable is for the Primary Use Problem to be solved by either finding a way to not artificially limiting supply or some other means and for the Conservation of Cost problem to be solved by finding a way to have the system run at an efficiency approaching that of as standard centralized system.

Both of these fundamental problems are very far from being solved currently, and the upcoming technologies such as Ethereum 2.0, while making huge improvements, do not suggest getting to a point of true viability is anywhere in sight.

Arie Trouw — If you enjoyed this, please follow me on Twitter: @arietrouw

Entrepreneur, husband, father, Dataist, engineer, human.