Past P2P digital cash projects failed primarily due to a lack of adequate decentralization. It all boils down to inspiring trust in market participants that your blockchain ledger is safe from a wide gamut of threats. One important factor is to ensure that the rules within the blockchain can’t be changed arbitrarily (e.g. total supply of coins, issuance rate, etc.). A second is that there needs to be a robust mechanism for enforcing the agreed upon rules. A third is wide distribution (in terms of total # of copies and geographical distribution) of all the data contained in the blockchain ledger such that all transactions are immutable and essentially impossible to adulterate after the fact. Decentralization can generally be described as the magnitude of how well the above factors (among others) are implemented in a blockchain.

Adequate blockchain decentralization is extremely hard to achieve. Not only are there technical barriers, but it also requires adoption, cooperation & loyalty from the end-users (see section 12.5). In a blockchain, enforcement of the algorithm rules is the responsibility of the “full-validating nodes.” Here, validating nodes are not the same are miners (see section 12.2). Instead, think of them as autonomous police cars, who’s only job is to enforce the rules within the blockchain algorithm. Each validating node is equipped with the entire ledger (hence the name, full-node) along with all the algorithm rules allowing them to independently identify/quarantine bad-actors.

You want as many police cars (full-validating nodes) as possible operating in a blockchain network, spread more-or-less evenly across the globe. As such, running these nodes should ideally be cheap and easy. The primary incentives to run your own validating node are: 1) to support the network you’re invested in, 2) have a voice in protocol-level changes in the algorithm where your vote is your choice of the software version you are running and, 3) gain the power to independently verify/validate every input/output from your personal wallet(s) — without trusting anyone.

As the complexity, and the amount of data within a blockchain balloons however, so does the hardware costs and expertise required to run a full-validating node. ETH currently has ~3500 full nodes, with full archival data totaling ~8,300 Gb, growing at a rate of ~200Gb/month. Just from a data storage point of view, running a ETH full node can be expensive for most. There are number of services (e.g. AWS, infura, etc.) where one can rent ETH full nodes, but end-users and hobbyist are unlikely to pay the fees to use these services. In fact, rented ETH nodes are largely used by for-profit companies for their operations. Moreover, ETH development pathway includes significant alterations (ETH2.0) to their blockchain that is likely to increase the technical complexity of operating a node. Reliance on AWS, Infura and others to makeup ETH’s full-node network will likely increase moving forward.

Although it is important to note that hobbyists can run a ETH “light node” that doesn’t have the entire blockchain history, only a curtailed snapshot of the current blockchain state. Here a light node must rely on another “trusted” full node for the entire blockchain history. Ready-made ETH light nodes can be found here and here.

Now, on to the rhetorical question: to what extent are you decentralized when you have a large number of full-nodes running but, a significant percentage is operated by centralized for-profit companies? This of course, can be a risk; one phone call from the right person can have a significant impact.

Today, blockchains are a dime-a-dozen and all of them throw around “decentralized” simply as a buzzword. In reality, very few blockchains are actually decentralized to any appreciable degree. ETH and Cardano (ADA) are the second and third most decentralized, respectively. They both lag far behind BTC (qualitative figure below).

There is an estimated ~12,500 “reachable“ full-nodes on the BTC network (actual node-count could be as high as 100,000, as of 2019). The key difference here is the BTC ethos of — anyone should be able to run a full-node. You can run a BTC full-node on any old computer or, raspberry pi (<$100+cost of ~500gb storage) using little-to-no energy & bandwidth. It is simple and cheap, follow instructions here. In addition to the number of nodes, it is also important to determine who are running the nodes? There is little to no need to pay AWS or Infura to run a BTC full-node that a $100 used laptop can easily run. Most of these reachable entities are largely individual end-users concentrated in stable democracies of North America, Europe and Asia.

Decentralization is likely the most important factor for why BTC commands such a premium in the crypto market. BTC’s degree of decentralization is its moat, along with its network value, name-recognition and community ethos that fiercely defends this fundamental property.

Although decentralization is the highest priority for BTC’s purpose as a store of value (more on it later on), it simply not as necessary for most applications. In fact, newer blockchains (Avalanche, PolkaDot, Cardano, Solana, Binance Smart Chain, etc. etc.) are competing to provide functionality that’s possible only when you compromise on decentralization (more on section 7). Comparing BTC to Solana for example, is akin to comparing a miata and a dumptruck; they both have wheels, but that’s about it.