Bot Club
Bot Club Legal Bots: Conceptual Implications for Future Blockchain Infrastructure
The Self-Owning Augmented Forest: Photo: Terra0.
Researcher Max Hampshire wrote a text on the material presented at Thursday Night Live! Bot Club: Legal Bots and included points of discussion raised at the Critical Making Session the following day.
The advent of Decentralised Autonomous Organisations (DAOs) has already begun to give rise to scenarios that push the limits of our conceptualisations of ownership and agency in the non-human realm and will to continue to push them ever further as blockchain technology develops. This text aims to help sharpen some of those conceptualisations, in order to better understand the magnitude and sheer weirdness of DAOs in both the near- and farther-future.
A Brief Introduction to DAOs
Decentralised Autonomous Organisations (DAOs) are organisations that are instantiated on a blockchain, and are constituted of interacting Smart Contracts: computer programs that are also instantiated on a blockchain. Smart Contracts are able to hold their own cryptocurrency wallets, they can self-execute on the occurrence of a certain event or state and can be interacted with (somewhat) like traditional web applications.
What makes DAOs interesting is that they inherit the characteristics of blockchain technology that make it such an attractive technological substrate to build upon in the first place. Blockchain technology enables the creation of entire organisations which can be transparent and are decentralised, free of being rooted in (and thus accountable to the laws of) any particular nation state. Most importantly for this essay, they are even able to act autonomously, free of the needs of human intervention and management.
This being said, it is very difficult to produce an exhaustive definition of DAOs. There are a myriad of potential DAO implementations, because they are, at their core, simply a list of rules defining the potential actions of the organisation in particular situations. These rules are either explicitly stated in the DAOs' code, or implicitly arise from the actions of its constituent parts (i.e. its constituent Smart Contracts). 'Successful' implementations range from the cryptocurrency Dash , to the now infamous attempt at transparent crowdfunding known simply as 'The DAO', and the DAO-production-platform Aragon. These examples indicate the currently still nascent state of blockchain infrastructure. There are far more initially-esoteric sounding (yet entirely technically realisable) DAOs that could arise in the not-too-distant future, given the continued development of blockchain technology. One such potential implementation that I will use as an example later in this piece was outlined by computer scientist, and AI-blockchain startup leader Trent McConaghy: a system of artificially intelligent DAOs that are able to create works of art, reproduce, and even adapt their own code (and therefore their offspring's future actions).
Overlapping fields of influence of cryptogovernance, cryptoeconomics, and crypto-based consensus, as well as the placement of human and machine agents in these fields.
DAOs as Decentralised Legal Persons
What is perhaps initially most intriguing about DAOs is that they have some kind of autonomous agency. They have the ability to act more or less independently and to manage their own assets (digital or otherwise). This agency exists even on the most the most basic level of a DAO optimising itself for its continued existence: retaining enough tokenised capital to continue paying for the basic transactions it needs to in order to continue operating.
However, the agency of DAOs jars with the conceptualisation of agency we ascribe to many other autonomous agents at work today, of which we tend to have a fairly anthropomorphic understanding. For instance, the algorithms performing high-frequency trading (HFT) are simply faster and cheaper versions of human traders. The wild variety of chatbots that rely on machine learning are likewise the faster, cheaper, and more easily controllable solution to the organisational problem of providing an always accessible interface to customers - they are the 24/7 stand-in for someone answering the phone. Also, many robots (hardware-based agents) such as those coming out of DARPA, are almost always 'human' in size and scope; emulations of animals and humans, built from the perspective of human designers and engineers.
DAOs exhibit a far broader form of emergent operational agency; resembling more the systems for market analysis or the actions of animal hive-minds than the actions of other software or hardware-based agents. In this sense they are more akin to infrastructure. Their particular brand of agency is made all the more distant from our conceptual norm by the fact that they only rarely interact with the world outside of the limits of their blockchain, mostly operating solely in closed-off loops with other blockchain-based systems or agents.
A more helpful example of a previous instantiations of non-human agency is the 'legal person' of the corporation. A corporation is a so-called legal fiction, that - in certain regards - is treated as an individual agent in the legal realm. It is a non-human rule-based construct that has rights and obligations conferred onto it, such as the ability to own property, or be legally liable for its actions.
According to science fiction author Charles Stross, corporations are not only the first non-human legal persons, they are also the first form of Artificial Intelligence because they were the first artificial non-humans that acted - on an abstract level - autonomously, and solely in their own self-interest. They have one main goal: survival. A corporation survives by making money in order to pay its staff and continue investing in either itself or sections of the economy that might support its future investments. In short: a corporation is a non-human legal person that metabolises capital, just like a DAO is.
Corporations and DAOs show enough similarity with regards to their actions and forms of agency that it is not too far a stretch - and is in fact useful - to conceptualise DAOs as legal persons in much the same way that corporations are, augmented by the inherent characteristics of blockchain technology. They are decentralised, autonomous legal persons that do not necessarily have to abide by the laws and regulations of a nation state.
But DAOs are not merely the decentralised cousins of the legal persons that are corporations. Corporations do not necessarily bring about the weird range of potential cases of ownership and legality that DAOs do. Some of these interesting cases will be discussed below.
Conceptual Implications: Weird corporations
The construction of corporate legal person has occurred gradually over the course of recent history, and as reactions to demands of security and ownership of assets; situations that did not force us to reconsider the basic notions of agency of non-humans, but merely asked us to willingly suspend our disbelief for the purposes of engaging in legal action. The weirdness of the potential cases that DAOs could (and will) bring about far exceeds this suspension, and instead forces us to consider the autonomy of non-humans anew, as well as the huge implications of these cases on our conception of non-human agency.
In his excellent _AI DAOs Series_, Trent McConaghy outlines a future in which DAOs are able to create generative artworks, make money from selling these artworks, and self-optimise in order to sell more artworks for a better price. Furthermore, DAOs will be able to 'reproduce' by copying their own code, scraping GitHub to add additional functions to this code, and instantiate this copy on a blockchain. Before long, a network of self-optimising DAOs could exist, in control of potentially huge amounts of cryptocurrency capital, with the sole task to survive via 'metabolising' capital. These DAOs are - if coded in a certain way - no longer even under human control and aren't even necessarily able to be traced back to those individuals who originally created any of the various bits of open-source code they're partially constituted of.
This indicates the sort of autonomy usually associated with agents such as computer viruses, not entities that more closely resemble corporations. This forcible reconceptualisation has a knock-on effect, because it also challenges our notion of ownership, the relationship that constitutes the basis of our global economic system.
Conceptual Implications: Non-Human Ownership
The case of self-replicating, GitHub-scraping, art-creating DAOs highlights novel forms of non-human ownership relations that arise at various scale levels of the blockchain-enabled space in which DAOs operate. They are in many ways functionally identical to human-asset ownership-relations but there are key differences due to the nature of the blockchain technology facilitating the existence of DAOs.
Tokens - the currency of blockchains - are the smallest level at which these new types of non-human ownership relations exist. Tokens are, in the simplest sense, 'programmable money' , and can be coded such that the emergent behaviour of all instances of a particular token resembles more the behaviour exhibited by certain natural systems than the flows of fiat currency. One of the most interesting use cases involving tokens is the proposed ERC721 non-fungible token standard. These tokens share all of the functions of 'standard' ERC20 tokens, but are also tied to an individual object; they constitute an ownership relation that is supported by a blockchain between an individual token and an individual asset such as an artwork. The programmable parameters of the token dictate the potential actions taken by it - and by extension, the asset it represents. A token could for instance be programmed such that it (and the asset it is tied to) could only be traded once within a given period of time, or only be traded between a particular list of addresses (i.e. only between certain agents). This is functionally tantamount to the sort of ownership transferal we see at work with controlled substances and the legal prerequisites for their sale (e.g. sale on provision of a valid ID that shows the buyer is above a certain age). The key difference with regards to tokens is that it is far more difficult to get around the restrictions that are imposed on tokens and tokenised assets; the transaction will simply not properly execute and the token exchange will not occur if transferal is attempted via means other those prescribed in the code.
In short, the mechanics of token transferal seem to operate in a similar manner to the form of legally-sanctioned ownership transferal we are used to on the human level, yet with a much more heavy say in the matter on the part of the token itself, along those lines prescribed in their code.
Harking back to McConaghy's situation outlined above - and moving 'up' a level of abstraction -we see that a future situation can arise in which non-humans are in control of programmable money on an unprecedented scale. This would be a situation outside of the 'comfort zone' of our current ideas regarding agency and ownership; decentralised legal agents outside of the legal sphere engaging with their surroundings via the sort of relations that have previously been restricted to the human realm. Furthermore, these agents could potentially even be in control of the parameters of the tokens they trade; these could be parameters that control the flow of tokenised assets at the most minutely detailed level.
The implications of this are far-reaching and force a reassessment of our conceptual norms. It is important to note here that assets represented by tokens held on a blockchain are secure in a way that physical assets are not, due to the cryptographic protections typical of blockchain technology. There is no way of brute-forcing a secure digital wallet that would not take an inconceivable amount of energy and time. So, DAOs are essentially able to control the transferal of value - an ability functionally equivalent to ownership as understood in the human realm - in a much more secure fashion than any physical being could with any physical or digital asset. Unless one interacts with a DAO 'on its own terms' - fulfilling the conditions that must be met before it engages in trade - there is no way of accessing the funds it holds in its wallet(s). In the case of assets that aren't on a blockchain (for example, a physical asset represented by a token), a suitably-programmed DAO would be able to pay for a human proxy to represent it in court (much like corporations already can and do). This DAO would thus be just as able to defend its right of ownership of this asset as any natural person is, were this asset illegally taken from it.
Wyn (Generation Unknown) token series. Each artwork is represented by an ERC721 token. Each generation of tokens creates a new generation of 'offspring' by swapping aspects of their metadata. Artwork: Harm van den Dorpel.
Motivations to Take This Seriously
The Indeterminately Near-Future Scenario The speculative (but technically realisable) scenario(s) sketched here, combined with the fact that DAOs are somewhat immune to legislative action after they are instantiated, point towards a future in which decentralised legal persons are not merely in control of vast amounts of assets, but in fact are potentially able to dictate the flows of these assets with greater precision than any agent (legal or natural) that exists today. Furthermore, they will be better secured against theft than the vast majority of digital or physical assets currently in circulation. This is unprecedented; entire industries and smart infrastructures could exist constituted of code that is not the property of a given person or group of people, and act entirely of their own - albeit programmed - volition, controlling assets on a global scale - and the humans who rely on them.
Although actually defining DAOs is not possible beyond their general technical characteristics, it is clear that their existence has (at least) profound philosophical, legal, and (crypto-) economic implications. Of these we only begin to get a glimpse, but they already force us to stretch our home-grown anthropomorphic conceptions of agency and ownership. Given the sorts of situation(s) we could be faced with in the future, further research into their potential is warranted; research that must simultaneously push the boundaries of our current conceptions in order to understand the alien actions of its subject. Hopefully via these future attempts at understanding decentralised non-human agents, we can also begin to understand our own place in the decentralised future that will, to a greater or lesser extent, be less than anthropocentric.
Max Hampshire
Max Hampshire is a programmer and crypto researcher based in Amsterdam. He is one of three project initators of terra0, a company built on the Ethereum network that provides automated resilience systems for forested ecosystems, and is also a Blockchain engineer at BlockLab in Rotterdam, developing proof-of-concepts and MVPs for the blockchain infrastructure of the future. He is a member of the RIAT Institute For Future Cryptoeconomics, engaging in research-through-practice to glimpse the potential interaction possibilities of tomorrow's cryptosphere, and was previously part of the Institute of Network Cultures in Amsterdam_._
Bot Club
Algorithmic culture is a label for the entangled, sometimes disturbing, sometimes funny state of cultural formation that humans have entered by producing algorithms that operate online on such a massive scale and at such a fundamental level of societal organisation that they play a key role in producing culture. In this programme the term 'bot' is used in a broad sense to include all algorithmic agents that operate without direct supervision of humans, but that present the results of their functioning to humans. Bot Club takes an amazed and critical look at the world in which bots, algorithmic agents and generative processes do their work, and places them centre stage. A Bot Club consists of two parts: the first is a Thursday Night Live! programme, with human speakers, demonstrations of and performances by bots and algorithmic agents. The evening programme is followed by a workshop day dedicated to putting notions into practice and producing concepts and code for functioning critical algorithmic agents.