Institutions to constrain chaotic robots: why generative AI needs blockchain

With Sinclair Davidson and Jason Potts. Available at SSRN

Abstract: Generative AI is a very powerful new computing technology, but the problem of how to make it economically useful (Alice: “hello LLM, please send an email to Bob”) is limited by its inherent unpredictability. It might send the email, but it might do something else too. As a consequence, the large language models that underpin generative AI are not safe to use for most economically useful and valuable interactions with the world. This is the ‘economic alignment’ problem between the AI as an ‘agent’ and the human ‘principal’ who wants the LLM to interact in the world on their behalf. The answer we propose is smart contracts that can take LLM outputs and filter them as deterministic constraints. With smart contracts, LLMs can interact safely in the real world, and can unlock the vast economic opportunity of economically aligned and artificially intelligent agents.

The exchange theory of web3 governance

With Jason Potts, Darcy W E Allen, Aaron M. Lane and Trent MacDonald. Published in Kyklos,  June 2023. Working paper available on SSRN

Abstract: Blockchains have enabled innovation in distributed economic institutions, such as money (e.g., cryptocurrencies) and markets (e.g., decentralised exchanges), but also innovations in distributed governance, such as decentralised autonomous organisations. These innovations have generated academic interest in studying web3 governance, but as yet there is no general theory of web3 governance. In this paper, we draw on the contrast between a ‘romantic view’ of governance (characterised by consensus through community voting) and the ‘exchange view’ of governance from public choice theory (characterised by an entrepreneurial process of bargaining and exchange of voters under uncertainty). Our analysis is the first to argue that the latter ‘exchange view’ of governance is best to understand the dynamics of governance innovation in web3, providing the foundations for a new general theory of governance in this frontier field. We apply the ‘exchange view’ of governance to three case studies (Curve, Lido and Metagov), exploring how these projects enable pseudonymous, composable and permissionless governance processes to reveal value. Our approach helps illuminate how this emergent polycentric governance process can generate robustness in decentralised systems.

Large language models reduce agency costs


With Jason Potts, Darcy W E Allen, and Nataliya Ilyushina. Available on SSRN.

Large Language Models (LLMs) or generative AI have emerged as a new general-purpose technology in applied machine learning. These models are increasingly employed within firms to support a range of economic tasks. This paper investigates the economic value generated by the adoption and use of LLMs, which often occurs on an experimental basis, through two main channels. The first channel, already explored in the literature (e.g. Eloundou et al. 2023, Noy and Wang 2023), involves LLMs providing productive support akin to other capital investments or tools. The second, less examined channel concerns the reduction or elimination of agency costs in economic organisation due to the enhanced ability of economic actors to insource more tasks. This is particularly relevant for tasks that previously required contracting within or outside a firm. With LLMs enabling workers to perform tasks in which they had less specialisation, the costs associated with managing relationships and contracts decrease. This paper focuses on this second path of value creation through adoption of this innovative new general purpose technology. Furthermore, we examine the wider implications of the lower agency costs pathway on innovation, entrepreneurship and competition.

The problem of ubiquitous computing for regulatory costs

Working paper on SSRN

The benefits of regulation should exceed the cost of regulating. This paper investigates the impact of widespread general-purpose computing on the cost of enforcing of regulations on generative artificial intelligence (AI) and decentralized finance (DeFi). We present a simple model illustrating regulators’ preferences for minimising enforcement costs and discuss the implications of regulatory preferences for the number and size of regulated firms. Regulators would rather regulate a small number of large firms rather than a large number of small firms. General-purpose computing radically expands the number of potentially regulated entities. For Defi, the decentralized nature of blockchain technology, global scale of transactions, and decentralised hosting increase the number of potentially regulated entities by an order of magnitude. Likewise, locally deployed open-source generative AI models make regulating AI safety extremely difficult. This creates a regulatory dilemma that forces regulators to reassess the social harm of targeted economic activity. The paper draws a historical comparison with the attempts to reduce copyright infringement through file sharing in the early 2000s in order to present strategic options for regulators in addressing the challenges of AI safety and DeFi compliance.

Why airdrop cryptocurrency tokens?

Abstract: A cryptocurrency token airdrop is a novel means of distributing rights over a blockchain project to a community of users and owners for free. The market value of these airdrop giveaways is often upwards of hundreds of millions of dollars. This paper considers why projects might choose this unusual and costly means of token distribution. It considers a selection of high-profile airdrops as case studies between 2014 and 2022. This is the first comprehensive analysis of the rationales and mechanisms of Web3 token airdrops. We find that two primary rationales for airdrops are marketing (to attract new users and to maintain a community) and decentralisation of ownership and control of a project (building community, providing regulatory protection, and enhancing security). The paper contributes to an understanding of business practice and strategy in the emerging cryptocurrency and blockchain industry.

Author(s): Darcy W. E. Allen, Chris Berg, Aaron M. Lane

Journal: Journal of Business Research

Vol: 163 Year: 2023 Article: 113945

DOI: 10.1016/j.jbusres.2023.113945 and manuscript version at SSRN

Cite: Allen, Darcy W. E., Chris Berg, and Aaron M. Lane. “Why Airdrop Cryptocurrency Tokens?” Journal of Business Research, vol. 163, 2023, article 113945.

Continue reading “Why airdrop cryptocurrency tokens?”

The Case for Generative AI in Scholarly Practice

Available at SSRN

Abstract: This paper defends the use of generative artificial intelligence (AI) in scholarship and argues for its legitimacy as a valuable tool for contemporary research practice. It uses a emergent property rights model of writing to shed light on the evolution of scholarly norms and practices in academic practice. The paper argues that generative AI extends the capital-intensive nature of modern academic writing. The paper discussing three potential uses for AI models in research practice: AI as a mentor, AI as an analytic tool, and AI as a writing tool. The paper considers how the use of generative AI interacts with two critical norms in scholarship: norms around authorship attribution and credits for contributions, and the norm against plagiarism. It concludes that the effective use of generative AI is a legitimate research practice for scholars seeking to experiment with new technologies that might enhance their productivity.

Review: Voting over a Distributed Ledger: An Interdisciplinary Perspective

Published in Journal of Economic Literature, 61 (1): 295–97. Available at the Journal of Economic Literature

A review of: Voting over a Distributed Ledger: An Interdisciplinary Perspective. By Amrita Dhillon, Grammateia Kotsialou, Peter McBurney, and Luke Riley. Foundations and Trends in Microeconomics, vol. 12, no. 3. Boston: Now, 2021.

Fraud is a stubborn problem in democratic elections, and the history of voting technology is a history of attempts to reduce it. Votes might be manipulated at the ballot box, while voting records are in transit to counting centers, at the counting centers themselves, and at the point of broadcasting the result of an election. As Jones (2006) shows, many political reformers of the nineteenth century were technologists who designed voting machines that they believed would not just make elections more efficient, but also reduce real and perceived threats to the legitimacy of election results.

In their monograph, Voting over a Distributed Ledger: An Interdisciplinary Perspective, Amrita Dhillon, Grammateia Kotsialou, Peter McBurney, and Luke Riley explore and propose a significant upgrade to the technical infrastructure of voting: electronic voting over the internet using blockchain technology. Blockchains, the distributed ledger technology that underpins cryptocurrencies such as bitcoin and ethereum, use a combination of cryptography and economic incentives to achieve consensus over the state of a shared ledger without the need for a central authority to maintain and enforce that consensus. Dhillon et al. set up the terms of their investigation as follows. There are many intuitively appealing characteristics of voting over the internet. Voters could vote from their home devices or on mobile phones, potentially enhancing voter turnout (particularly among those who, for disability or socioeconomic reasons, find it difficult to turn up to a ballot box). Fully electronic votes could be counted immediately and at far lower cost.

Yet, as the authors detail in chapter 2, online voting has a wide range of risks that can threaten the real or perceived legitimacy of an election’s results. Some of these are cybersecurity risks: the software on a voters’ device might be corrupted through malware that manipulates the vote as it is cast, or the infrastructure that communicates votes to a counting authority could be targeted by adversaries with a denial-of-service attack. A counting service could go offline, possibly causing confusion that would lead to claims about illegitimacy. A more severe potential problem is that the counting authority might itself be corrupt. It is often observed that algorithms can be a black box to all but their creators; the difficulty of auditing a counting system might allow fraud to pass undetected. These potential risks are why many jurisdictions still use paper ballots or machines that print paper receipts-each of which present an audit trail manual verification by election supervisors or the voters themselves.

In chapters 3 and 4 the authors provide a straightforward description first of distributed ledger technology as a general category of decentralized systems, and then of blockchains as a special case of decentralized ledger. They distinguish correctly between “permissionless” systems, such as Bitcoin, where any user can join and participate in the consensus mechanism, and “permissioned” systems that will typically have a closed (and known) set of validators.

Chapters 5 and 6 are the core of the monograph, where the authors describe a high-level design of a blockchain voting system. Voters would be given a unique identity and vote by submitting a transaction directly to a node on the network. The node would then propagate that vote across the network. A voter could verify, using their private keys, that their vote had been recorded correctly and was included in the total vote count. Distributing control of the voting system across multiple nodes would reduce the risk of corruption by a central authority as well as provide a path for external observers to validate the correctness of the final count. Vote secrecy and anonymity could be assured using frontier cryptographic techniques such as zero-knowledge proofs that would allow anyone to check if a vote is valid without revealing any information about the vote other than its validity.

Voting over a Distributed Ledger is an effective argument for the feasibility of blockchain voting, but it is surprisingly modest. Some short sections suggest a much more ambitious vision. As the authors note, the step change in efficiency and reliability offered by blockchains could allow for the adoption of alternative voting schemes. Better technologies for voting open up possibilities for innovation in democratic form. It is hard, for example, to imagine a quadratic voting scheme (Weyl 2017) being built on paper ballots alone. In work with colleagues (Allen, Berg, and Lane 2019), I have explored a form of blockchain-based “liquid democracy” where voters could delegate their voting power while at the same time reserving the ability to directly vote on more personally salient issues.

Indeed, we already can see some of the possibilities of democratic innovation in the way that blockchain-based communities have innovated on the voting systems they use for organizational governance. Delegative democracy where voters retain an option to override their delegate’s vote is a core part of the Cosmos SDK, one of the major classes of blockchain protocols. More sophisticated systems are widespread in decentralized autonomous organizations (DAOs). One compelling example is the time-weighted voting systems that grant DAO token holders additional voting power for having locked up their tokens for a period of time, used most famously by the Curve protocol. As these examples show, reducing fraud, lowering costs, and increasing the speed with which elections are counted are important goals, but blockchains offer something else that should excite social scientists: a vast new design space for building systems of collective decision-making.

REFERENCES

Allen, Darcy W. E., Chris Berg, and Aaron M. Lane. 2019. Cryptodemocracy: How Blockchain Can Radically Expand Democratic Choice. Lanham, MD: Lexington Books.

Jones, Douglas W. 2006. “Technologists as Political Reformers: Lessons from the Early History of Voting Machines.” Paper presented at the Society for the History of Technology Annual Meeting, Las Vegas, NV, October 13. https://homepage.divns.uiowa.edu/ jones/voting/SHOTpaper.pdf.

Weyl, E. Glen. 2017. “The Robustness of Quadratic Voting.” Public Choice 172 (1-2): 75-107.

How Web3’s ‘programmable commerce layer’ will transform the global economy

World Economic Forum, 28 November 2022. Originally published here. With Justin Banon, Jason Potts and Sinclair Davidson.

The world economy is in the early stages of a profound transition from an industrial to a digital economy.

The industrial revolution began in a seemingly unpromising corner of northwest Europe in the early 1800s. It substituted machine power for animal and human power, organized around the factory system of economic production. Soon, it created the conditions to lift millions of humans from a subsistence economy into a world of abundance.

The digital economy began with similarly unpromising origins when Satoshi Nakomoto published his Bitcoin white paper to an obscure corner of the internet in late 2008. We call this the origin of Web3 now – with the first blockchain – but this revolution traces back decades as the slow economic application of scientific and military technologies of digital communication. The first wave of innovation was in computers, cryptography and inter-networking – Web1.

By the late 1990s, so-called “e-commerce” emerged as new companies, which soon became global platforms, built technologies that enabled people to find products, services and each other through new digital markets. That was Web2, the dot-com age of social media and tech giants.

But the actual age of digital economies was not down to these advances in information and communications technologies but to a very different type of innovation: the manufacture of trust. And blockchains industrialize trust.

Industrial economies industrialized economic production using physical innovations, such as steam engines and factories. Such institutional technologies organize people and machines into high production. What the steam engine did for industry, the trust engine will do for society. The fundamental factor of production that a digital economy economizes on is trust.

Blockchain is not a new tool. It is a new economic infrastructure that enables anyone, anywhere, to trust the underlying facts recorded in a blockchain, including identity, ownership and promises represented in smart contracts.

These economic facts are the base layer of any economy. They generally work well in small groups – a family, village or small firm – but the verification of these facts and monitoring of how they change becomes increasingly costly as economic activity scales up.

Layers of institutional solutions to trust problems have evolved over perhaps thousands of years. These are deep institutional layers – the rule of law, principles of democratic governance, independence of bureaucracy etc. Next, there are administrative layers containing organizational structures – the public corporation, non-profits, NGOs and similar technologies of cooperation. Then we have markets – institutions that facilitate exchange between humans.

It has been the ability to “truck, barter and exchange” over increasing larger markets that has catapulted prosperity to the levels now seen around the world.

Information technology augments our ability to interact with other people at all levels – economic, social and political. It has expanded our horizons. In the mid-1990s, retail went onto the internet. The late 1990s saw advertising on the internet. While the mid-2000s saw the news, information and friendship groups migrate to the internet. Since their advent in 2008, cryptocurrencies and natively digital financial assets have also come onto the internet. The last remaining challenge is to put real-world (physical) assets onto the internet.

The technology to do so already exists. Too many people think of non-fungible tokens (NFTs) as trivial JPEGs. But NFTs are not just collectable artworks; they are an ongoing experiment in the evolution of digital property rights. They can represent a certificate of ownership or be a digital twin of a real-world asset. They enable unique capital assets to become “computable,” that is, searchable, auditable and verifiable. In other words, they can be transacted in a digital market environment with a low cost of trust.

The internet of things can track real-world assets in real-time. Oracles can update blockchains regarding the whereabouts of physical assets being traded on digital markets. For example, anyone who has used parcel tracking over the past two years has seen an early version of this technology at work.

Over the past few years, people have been hard at work building all that is necessary to replicate real-world social infrastructure in a digital world. We now have money (stablecoins), assets (cryptocurrencies e.g. Bitcoin), property rights (NFTs) and general-purpose organizational forms (decentralized autonomous organizations (DAOs)). Intelligent people are designing dispute-resolution mechanisms using smart contracts. Others are developing mechanisms to link the physical and digital worlds (more) closely.

When will all this happen? The first-mover disadvantage associated with technological adoption has been overcome, mostly by everyone having to adopt new practices and technology simultaneously. Working, shopping and even entertaining online is now a well-understood concept. Digital connectedness is already an integral part of our lives. A technology that enhances that connectedness will have no difficulty in being accepted by most users.

It is very easy to imagine an interconnected world where citizens, consumers, investors and workers seamlessly live their lives transitioning between physical and digital planes at will before the decade concludes.

Such an economy is usefully described as a digital economy because that is the main technological innovation. And the source of economic value created is rightly thought of as the industrialization of trust, which Web3 technologies bring. But when the physical parts of the economy and the digital parts become completely and seamlessly join, this might well be better described as a “computable economy.” A computable economy has low-cost trust operating at global market scale.

The last part of this system that needs to fall into place is “computable capital.”

Now that we can tokenize all the world’s physical products and services into a common, interoperable format; list them within a single, public ledger; and enable market transactions with low cost of trust, which are governed by rules encoded within and enforced by the underlying substrate, what then?

Then, computable capital enables “programmable commerce,” but more than that – it enables what we might call a “turing-complete economy.”

Repugnant innovation

Abstract: Repugnant innovation is a form of evasive entrepreneurship that occurs in repugnant markets. Repugnance is an informal institution – controlled by long-lived norms, attitudes, customs and traditions – and repugnant innovation acts to shift institutions at the lowest level of the institutional stack. The paper considers three examples of repugnant innovation: e-cigarettes, online gambling, and webcam modelling. Each repugnant innovation challenges the complex mixture of material and moral concerns that contributes to repugnance in their respective markets. The paper adds to and expands on a body of evidence about innovation in apparently unsupportive institutional environments.

Author(s): Darcy W. E. Allen, Chris Berg, Sinclair Davidson

Journal: Journal of Institutional Economics

Year: 2022 Pages: 1–12

DOI: 10.1017/S1744137422000364. Working paper at SSRN.

Cite: Allen, Darcy W. E., Chris Berg, and Sinclair Davidson. “Repugnant Innovation.” Journal of Institutional Economics, 2022, pp. 1–12.

Continue reading “Repugnant innovation”