Scalability of value in decentralized networks

Decentralized networks existed before blockchain, BitTorrent being one example. But the key innovation of blockchain is its ability to hold economic value, as pioneered by the Bitcoin network. Since the inception of Bitcoin its ideas underwent many rounds of modifications. Currently, the blockchain space holds promise for a broad range of decentralized applications in many areas, including commerce, social media, content sharing, internet services, logistics, and even government.

At CoinFund, as we contemplate the growth of the blockchain space we conclude that such decentralized networks are a potential disruptor of society unlike any that humanity has seen before. We also ask ourselves what is still missing in terms of the blockchain’s ability to penetrate and transform existing markets and social structures. In this article I will expose one aspect of this question, that of scalability of value.

Scalability of value is different from scalability of transaction throughput or data capacity, which is addressed by recent and upcoming developments such as Lightning Networks, State Channels, sharding systems, and others. Scalability of value is more subtle yet, in my view, more fundamental.

According to my estimate the cost of attack (COA) on the Bitcoin network is on the order of 150 million dollars. For Ethereum this number is much lower. For smaller POW-based networks it is tiny by comparison.

As exemplified by the recent number of double-spend attacks on smaller clones of the Ethereum network, the COA is an important measure of safety of the network as a gauge of its systemic risk. Fundamentally, this is a scalability issue, because the motivation to attack the network grows as its value (expressed as total market capitalization) increases. Consequently, in order for the network to scale, the COA must grow commensurate with the value of the network.

How big should the cost of attack be? An attacker motivated by profit has to perpetrate a successful double-spend in the value exceeding the network’s COA. Obviously, a large double-spend attack will be quickly detected and, consequently, the network will lose value. The double-spend must be performed before this happens. The viability of such an attack depends on trading volumes of a given cryptoasset, which in any given day are a small percent of the network’s total value.

On the other hand, attacks are not necessarily motivated by profit. Some actors may desire to destroy or cripple a decentralized network because of the dangers it poses to their core business or the way it disrupts their interests. In this case, a minimum required cost of attack is quite hard to estimate. For example, PayPal, if it desires to destroy Bitcoin, can afford to pay the current COA.

This gets more complicated for smart contract platforms and those that allow users to define their own tokens. In these cases the total value of the network is not limited by the market capitalization of its underlying cryptotoken, but becomes equal to the total value of all tokens carried on this network. In general, a value of a single smart contract can become arbitrarily large if the contract is used for important tasks, such as governance, or to create complex financial instruments, such as derivatives.

For Bitcoin, in order to maintain its current proportion of total value to COA, miner rewards in proportion to total value must grow or remain the same. This means that assuming a constant relationship between Bitcoin market cap and its transaction volumes, the transaction fees must eventually rise to about 1%, as block rewards wind down. But Bitcoin’s COA is not currently sufficient to protect it from attacks by sovereign entities which may quite easily shell out the funds and equipment necessary. I argue, in fact, that the network’s COA must match or exceed it’s value and/or contain safeguards of non-economic nature.

For Bitcoin network to increase it’s COA from the current 1.5% of market cap to 10% will make the required transaction fees entirely untenable. What alternative is there? Delegated proof-of-stake systems allow to potentially increase COA up to the total market cap of the cryptotoken underlying the network. They also have a nice feature that the COA will increase as the cryptotoken’s value grows.

But this improvement remains an economic safeguard. It does not provide sufficient protection to nascent networks whose value is small from actors who want to insure a future without free finance. To introduce better protection networks must issue and use some kind of tokens that can not be simply purchased.

This may be based on social identity, even if anonymous. Urbit, for example, has a system of “digital land” that is similar to social identity. Transfers of digital land are not frictionless as compared to a true cryptocurrency. Consequently they carry the benefit that usurping the power offered by owning the majority of these tokens becomes untenable. Schemes like this may be required to create a decentralized network that can withstand the long-term challenges of value scalability.

As the space matures we will inevitably have to address this problem. Otherwise, we are sure to see a vast number of networks eventually fail due to safeguards insufficient for the increasing value they are holding. This may be just how things are. But I suspect there are simple solutions that are not being considered simply because the space is too young for this type of scalability problem to come into the spotlight. I imagine with some care we can make sure it never will.