In this new Working Paper with Luis Garicano, we argue that, by allowing networks to split, decentralized blockchain platforms protect members against hold up, but hinder coordination, given that adaptation decisions are ultimately decentralized. The current solutions to improve coordination, based on “premining” cryptocoins, taxing members and incentivizing developers, are insufficient. For blockchain to fulfill its promise and outcompete centralized firms, it needs to develop new forms of “soft” decentralized governance (anarchic, aristocratic, democratic, and autocratic) that allow networks to avoid bad equilibria.
The contractual technology on which the market economy is based has evolved over a period of two thousand years. Blockchain promises to overhaul it, by allowing for the creation of open, distributed, secure, encrypted and programmable digital ledgers and enabling secure and fully decentralized “P2P” trade.
Fulfilling this promise requires decentralized blockchain platforms to provide operating standards, a central reputation for the whole ecosystem, and a communication protocol that allows applications (“Apps”) to communicate with each other. But the success of such decentralized platforms hinges crucially on finding good governance solutions to align the incentives of all participants.
Specifically, all platforms, centralized or not, must solve two classes of problems: protecting the investment of platform participants from hold-up and coordinating adaptation and change. In this paper we discuss how decentralized, blockchain-based, platforms solve these two problems compared to traditional centralized ones, and evaluate the governance solutions currently being proposed by different platforms.
Hold up problems are particularly prevalent in centralized platforms. Once participants have committed to the platform, switching is costly. This allows the “network architect” (e.g., Apple in iOS Apps) to hold up application developers, for instance, by implementing changes that make their previous investments obsolete. The risk of being held up leads participants to underinvest. When the network architect has a sufficiently long horizon (again, like Apple in iPhone Apps or Google in Android), it has an incentive to develop a reputation for protecting the investments of its application developers (and not holding them up) that enforces the relational contract. Decentralized platforms such as blockchain drastically reduce the potential for hold up, since network members can always refuse to introduce any change. This reduction, however, comes at the cost of increasing coordination problems.
In both types of platforms, coordination problems result from the fact that platforms must ensure that core software and application developers have the right incentives to cooperate and adapt together to evolving needs. In centralized platforms, the decision to change is made by the architect, who ignores the private costs of the developers and, as a result, tends to introduce too much change. In a decentralized platform the opposite happens: individual application developers (and other platform members) have some veto power through their freedom to update. They must therefore agree to any change. This poses a problem for changes that are not win-win, that is, any change which is good for the whole network but causes losses to some players. Decentralized platforms will find this type of change hard to implement.
In Section 2 we study the key contracting difficulties in centralized platforms and their relational solutions. The network architect can unilaterally make the adaptation decisions of the platform. It may require excessive investments from partners and too frequent adaptation, since it does not fully internalize the costs of such decisions. However, a network architect with sufficient relational capital can use it to safeguard a governance structure that protects participants in the platform against its own hold up and thus provide them with adequate incentives to invest. To illuminate these problems, we briefly describe the governance of the Apple and Google platforms and their ability to protect developers from hold up and adequately respond to change.
In Section 3 we consider blockchain-based platforms. In a decentralized setting, each member of the network (individual developers, final users and other partners, such as blockchain miners) can refuse to change, which protects her from hold up. However, incentives to adapt in a coordinated manner are limited. There may exist multiple equilibria, as in the traditional network externality literature (e.g., Katz and Shapiro, 1985, 1986; Farrell and Saloner, 1985). Absent explicit monetary incentives or other governance solutions, the system is likely to yield too little adaptation or generate inefficient “splits” (where only part of the network adopts a particular protocol change). However, unlike in standards “wars”, where two standards battle in a winner-take-all confrontation, platforms can implement governance rules to handle the trade-off between hold-up and coordination as efficiently as possible.
For instance, consider the two main decentralized platforms: the ones developed on top of Bitcoin and Ethereum. While Bitcoin is a digital currency that can be used for structuring simple predefined transactions, such as escrow accounts or document signing, Ethereum was created with the aim of making even more complex transactions easier. In particular, its programming language (Solidity) supports complex logical functions like conditions, recursions, loops, go-tos, macros, etc. Codified Ethereum’s transactions can be called repeatedly, to send and receive funds, to act as new currencies, as well as to interact with other contracts. These features make it suitable for coding “smart contracts”—those which are enforced automatically by computer code.
The governance of both platforms is also very different. Bitcoin is an extremely decentralized platform with very little governance. Ethereum is a considerably more centralized platform where some additional governance aims to achieve some coordinated adaptation.
It is precisely the same ability to adopt changes individually—even to split from the main network—that protects App developers and others from expropriation what may lead to large scale coordination failures. In each of these decentralized platforms we study how they handled a situation where the implementation of a change led to a large coordination failure and a split in the network, in connection with what is called in the blockchain community a “hard fork”. We evaluate, in each case, the governance solutions that are being currently proposed by the different platforms, examine their weaknesses, and propose some further solutions.
While blockchain governance involves significant shortcomings, we study some of the new tools allowed by blockchain that may be used to solve incentive conflicts.
First, platform designers can rely on the complementarity between crypto currency and platforms. Specifically, “pre-mining” and other ways of allocating cryptotokens and cryptocoins to developers and, particularly, core developers, may alleviate coordination problems. We argue that this is only of limited value, and only early in the life of the platform. The impossibility of imposing “vesting periods” and other restrictions substantially weakens this tool.
Second, some networks are implementing solutions designed to facilitate coordination around the best equilibrium. We classify these emerging solutions as anarchic, aristocratic, autocratic and democratic, by analogy to political “hard” governance in the real world. Although these solutions “feel” like governance, ultimately network members always have the ability to implement or not these changes, and thus they are better understood as devices to facilitate the convergence of expectations than as real governance devices. Hence our use of the term “soft governance”.
In Section 4 we compare the relational capital needed by optimally-governed centralized and decentralized platforms. We show that centralized platforms require less relational capital when network members are highly heterogeneous, while decentralized ones are preferred when investments are hard to verify. This results point the way to the instances in which we expect one or the other solution to prevail.
The main precedent for our approach, using the economics of contracts and organization to study these platforms is Bresnahan and Greenstein (2014), who discuss informally mobile computing platforms, and suggest that the key tradeoff is that, while a hierarchical, or centralized platform is superior in terms of coordinating a response to change, a decentralized one is superior for exploration of new products and ideas. Our discussion is distinct in that we focus on a different tradeoff, between investment (hold up) and coordination; we analyze the problem in terms of relational contracting; we propose a formal model that we believe captures this tradeoff in the simplest possible way; and finally our case study discussion has to do with the blockchain platforms, not discussed in their paper.
There is also a burgeoning literature on bitcoin and blockchain. Böhme et al. (2015) focus on the economics of bitcoin. They note two key costs: one is the technological waste, and the second is the cost associated with the market concentration of intermediaries, notably of miners. Catalini and Gans (2016) study blockchain economics and argue that its main benefits are the reduction in these cost of verification and (with the addition of bitcoin) of the cost of networking. Athey et al. (2017) discuss the privacy paradox, whereby cryptocurrencies offer people the chance to escape government surveillance, but do so by making transactions themselves public on a “blockchain”. The effect of small incentives may explain the privacy paradox, where people say they care about privacy but are willing to relinquish private data quite easily. Biais et al. (2018) undertake a full, detailed theoretical dynamic analysis of the equilibria in a blockchain. Our work is distinct from the above in our emphasis on how emerging governance solutions deal with existing problems and pointing the way to future governance, and in comparing the robustness to incentive conflicts of centralized and decentralized networks.
Our work builds on two recent streams of literature in organizational economics on coordination and adaptation (e.g. Dessein et al. 2010) and on relational contracts (Levin, 2003). We also rely on the law and economics literature on agency law and the birth of the corporation (e.g., Hansmann and Kraakman 2000, Harris 2000, Arruñada 2010) to highlight that comparisons of current blockchain governance against centralized governance is unfair to blockchain: centralized governance applies ready-made solutions provided by a centuries-old evolution of institutional and organizational solutions, while blockchain is just starting a similar process of evolutionary discovery.
Source: Arruñada, Benito and Garicano, Luis, Blockchain: The Birth of Decentralized Governance (April 10, 2018). Pompeu Fabra University, Economics and Business Working Paper Series 0000. Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3160070