 |
| Pix credit here |
I was delighted to have been asked to deliver brief remarks at the public conference, BlockchainGov International Symposium Present & Future of Blockchain Governance, to be held at Panthéon Assas--Université Paris II 25-26 September 2025. The conference is meant to celebrate and look back on the remarkable work of a group of excellent colleagues, organized under the code name Blockchain Gov.
As BlockchainGov enters its final year, this public symposium invites you to an experimental overview of five years of research on blockchain governance. Together, we’ll explore how concepts like legitimacy, polycentricity, alegality, and constitutionalism shape governance in decentralized systems. Each panel will open with a short BlockchainGov research presentation, followed by participatory dialogue spinning internal working group themes into public conversation.
The title of my remarks is "Coding the Rule of Code Versus Rule of Law for Blockchain Debate." Its object is to consider an underlying contradiction in the now well established parameters within which humans debate the normative elements of code based regulatory structures and their relationship (either hierarchically subordinate to, as an instrumental methodology or autonomous of) to law--understood as an object (text preserved command), as a system of human behavior management, and as the cages of the performance of the organizing premises (and their ideologies, that is the formal applied structures of human cognition) of the incarnation of human collectivity as a phenomenological set of expectations and dialogical performances. That contradiction lies at the heart of the means by which coded regulation and traditional textual (human centered) regulation are coupled. Source coding, the human language of digitalized regulatory measure, and machine languages, the binaries and executable files arranged in a logical, temporally sequential, and iterative process, are not mirror images of the other. They meet at the points of translation and transmission of inputs and outputs. But their logic and the way they sequence data, process, value and operations may neither align nor may they process mimesis of the kind that would advance the premise of human centered automated and tech based data driven behavior modeling and control.
I leave the contemplation of thise issue sot the readers, and invite conversation on a small aspect of a larger problem with respect to which humans have only just begun to engage with.
An annotated version of the text of "Coding the Rule of Code Versus Rule of Law for Blockchain Debate" follows below and may be accessed HERE: BACKER_Rule_of_Code_Blockchain_2025v4
The Program of the conference and its speakers also follow below.
Coding the Rule of CodeVersus Rule of Law for Blockchain Debate
Larry Catá Backer[1]
Keynote Remarks Prepared for the Session on Rule of Code vs Rule of Law, International Symposium: BlockchainGov International Symposium; Present & Future of Blockchain Governance
25-26 September 2025; Panthéon Assas - Université Paris II[2]
1. Introduction.
I have been asked to speak to the interactions between the Rule of Code and the Rule of Law within the virtual spaces filled with automated data storage and sequential transactional mechanisms, and more specifically in its form as blockchain. I start by noting a bit of absurdity in this quest. Nietzsche famously mocked the English, and English attempts at philosophy, for its obsession with utility. That obsession has now appeared to ooze into and fill the virtual world and more specifically the intelligences with which humans seek to populate that world without much of a thought about the organizing parameters of utility and its limits. It might then be useful to take a moment , to consider this obsession and its virtual manifestation for humans and for machines from the perspective of phenomena, of semiotics, language, and collective meaning making.
It has become almost a cliché of modernity that some of the most profoundly interesting and foundational shifts in the management of human collective organization and, more importantly, its rationalization of the surrounding reality, can be projected from out of the smallest and most laterally irrelevant advances. So it might be said of blockchain—as technology, as a means of facilitating the structural coupling of individuals and their collective institutions, and at a fundamental level, as an ideology of managing human collectives and its manifestation of a peculiar and quite specific form of organizing reality.
The core of the cognitive cage that emerges as blockchain, one manifested in real space through digitized bits of data that are significant[3] to participants and then ordered chronologically, is deceptively simple: the creation of a digital ledger. That digital ledger, as its name implies, is a space in which something is recorded—something of interest or value to those who would expend time and effort on its creation. From the perspective of semiotics, one encounters here the mimesis of a discrete object (temporally and spatially) which acquires signification through its virtual representation, which is then itself made real by its effects beyond the digitized ledger. In the beginning one recorded transactions and developed something like a self-referencing currency (and all currencies whether digital or analog are by definition self-referencing in the sense that their value is a function of collective belief in their value through use). And yet there is nothing to prevent the recording of any collection of bits of data to which one or more providers might attach significance.
Nonetheless, that signification is of little value when the self-referencing unit is the autonomous individual. That act of recording becomes more interesting for the ordering of social relations when such digitized ledgers then acquire collective significance and are distributed across many digitized spaces (computers or other holders of virtual “memory” in mimetic form). One creates, in virtual form, transparency, and trustworthiness of the “memory” recorded in the ledger in the sense that such memories are harder to contaminate or otherwise be tampered with. They might be appended but not overwritten; [4] as a consequence their existence in virtual time and space is unalterable as such. That inalterability provides a powerful basis for utility—in the sense that it populates the cognitive spaces for which it was created, and it itself becomes an object of consumption, use, negotiation, or judgment, which can then be consumed in the construction of other objects without losing its historical facticity. Those other objects could encompass virtually anything, from rewards and punishments grounded on ledger entries of behaviors, to the accounting for wealth grounded in the meaning, the significance, of both the ledger entry and its deployment.
At the same time, the collectivization of signification of ledger entries acquires an internal collective aspect—and again a means of organizing its cognitive signification, by organizing multiple ledger entries together into “blocks”. The cognitive signification becomes more profound as the internal collective signification manifested in and as the block then is put in motion, creating a cognitive “flow” in and through collections of blocks linked in a chronological “chain” (and here time becomes an element of code, and of cognition). All of this, however, produces a cognitive system that is not self-aware. Something else is necessary—and that something are the “nodes” in the networks that host copies of the ledger in its blocks.[5] Nodes are also objects, but they are objects outside the “chain” of “blocks” around which data is organized through digitized records of whatever object has been selected for such virtualization. In a sense, the nodes were the physical user interface that constituted the members of the collective within which self-referencing systems of ledger entries of data in blocks, temporally chained, could be “activated” by producing consequences or actions (and thus additional blocks along the chain). The nodes are both the guardians and the beneficiaries of the realities created within temporal “block”—“chains” by protecting the integrity of the “past” and providing a consensus-based protocol for validating activation consequences in the form of new blocks along the chain (it need not be so but nodes also mimic the more fundamental ordering ideologies from which they emerge).[6]
It is here that digitalization occurs, through the incorporation of digital technologies to automate tasks, including the retention of data. Nodes are themselves, for the most part, computers or servers—that serve as their physical bodies. These physical bodies then operate through a meta-operating system and more specifically with respect to the blockchain, from software that sets out the operating rules applicable to each node and limiting their actions in communion with all other nodes admitted to the congress of nodes that constitute the blockchain collective. Software is itself a set of human-readable, and sometimes human written,[7] instructions in a programming language, also traditionally devised by humans and for their pleasure (utility), that provides the basis for computers to perform certain tasks with and on available data, the digitization of which is also a product of software coding. Coding, source coding, is itself the process of writing software in human-readable form; it was the way that humans communicate with virtual intelligence (in their hardware bodies), but not necessarily the way that virtual intelligence communicated among themselves. To those ends, virtual intelligence communicates with each other using machine language—a series of binary codes that can be utilized by a computer processor to execute commands, including, eventually, the command to write their own code and execute their own machine language instructions. It also includes object coding (machine-readable output of source code compiler), and executable code (computer operational programs sometimes in the well-known .exe format that are one’s program files). Humans can read source code, but not machine and object coding.[8] Source coding is for programming (at least at the start) and machine coding is for execution (including execution that permits machine coding to undertake its own source coding). Here lies both the independence of rules of code from rules of law—built into the language of these systems that both constitute a regulatory order and erect walls against the projection inward of other regulatory rule orders.
In a sense, then, blockchain involves the alignment of three languages—human to human (and with it its own semiotics of meaning making), human to machine language (and with it a transposition of more personal value semiotics translated for machine intelligence), and machine to machine language, the semiotics of which is a function of the possibilities written into its own language. And it is in this alignment of language—and with it, its semiotics and phenomenology of disciplining collective meaning and enforcing command—that one can, at last, speak to the rise of collective legalities beyond that of their traditional manifestation in physical power asserted through action, text, and institution. We focus on two in the context of blockchain—the well-known law of code and the law of regulation. Both embrace the notions of the rule of law to the extent that they are encoded within its language meaning structures, and from there applied in ways that reinforce or at least manifest meaning and values onto regulatory objects. And now one encounters a Sacher-Torte of languages—at least five—dense layers of cake separated by a jelly that serves as the medium of connection between them, and all to produce what appears at first blush to be a straightforward and efficient system.
And it does! The elegance of this form is irresistible. It is democratic and anti-hierarchical in ways that only technology enhanced participation can make possible (at least for those who can afford the technology and the cost of the training in it). It provides an enhanced security against corruption (the limits of which remain to be tested), and it provides an equalizing transparency that protects the immutability of facts—that is of the recording of data that is widely shared and thus much more difficult to “remake” to suit the times or the inclinations of power. Its decentralization provided a virtual pathway to strong tendencies that were already largely visible in the world of human organizational regulation—from a revived federalism in the United States, to the rise (more or less) of the principle of subsidiarity in the European Union, to selective centralization and decentralization within Chinese Marxist-Leninism. And its efficiencies were unparalleled. It appeared to solve the great challenge of law making—one in which its essence as command (*.exe) was an exogenous projection onto the bodies of those who were to comply. That produced the great procedural jurisprudence of law: to figure out of means of naturalizing the principles of legal text in its subjects and to develop mechanisms for external policing. Blockchain automated the process of regulation and compliance, embedding internalization within the language of its own coding. The languages of programming, then, were normative, regulatory, descriptive, and procedural; creating the subject of compliance and its methodologies simultaneously.
2. The Challenges
Virtually any collective human (and machine) activity can be “blockchained,” and thus “blockchained,” automated in a virtual space with effects not just there but in human physical spaces as well. While it plays a significant role in human activities that lend themselves to chrono-blocking information and decision systems—that is, to ledger records blocked and ordered in time which are then subject to sequential action along linear temporal arcs. These include the usual suspects —cryptocurrency, supply chain management, smart contracts (but see Michael Jünemann’s 2021 critique, “Can Code be Law?”), banking, and digital identity—its potential extends far beyond that to the operation of municipal administration in the form of smart cities, in the management of households through smart houses (sometimes aligned with the Internet of things), in the management of behaviors through social credit based systems of rewards and punishments based on blockchains of value weighted recording of behavior, to smart government in the form of smart courts and perhaps smarter supervision of an administrative apparatus that now serves as the physical manifestation that materializes virtualized reality. It can be applied to any data rich environment from within which transactions of all kinds can be constituted—health care, psychological services, chatbots, other quasi currencies and property identification systems (for example, Non-Fungible Tokens), and media of all kinds. It is both the essence of history and the objects based building blocks for transactions.
This, in turn, provides a framework within which automated decisions and consequences flow from transactions subject to the flow of the coding which gives ledger entry transactions their signification and then produces judgment and consequence (action on and guided by the values of judgment) through the operationalization of collective interpretive rules based on an agreement of the value of objects and their flow through transactions—it is, in this respect both the idealized universe of the semiotic triad (object to signification to interpretation/consequence, to object again) and the automation of rule systems grounded in the identification of objects which will be “worked on” in predictable and pre-determined ways. That repeating pattern, one which builds on itself and might shift depending on the consequences (in code) of aggregations and shifting patterns of transaction data constitute the fundamental signification of coding (in blockchain).
It is within this framework that one might consider the three questions that were put forward for our discussion this morning: The first touches on the main contemporary threats to the Rule of Code. The second considers the importance of the rule of code for the legitimacy of blockchain systems. And the third would encounter the dialectics between regulation by governance and traditional regulation within the contemporary context of recent legislative and executive actions in the blockchain space. Each has its own semiotics. The first is embedded in the broader semiotics of mandatory or coercive collective meaning making, as a function of signified values and expectations of the natural in the world, which is to be replicated or internalized through coding into the machine language of blocks in chains that then serve as the virtual performance of the physical with which it interacts at the start and end of the iterative process of block making. The second goes to the normative essence of the language of blockchain, and with it of the way in which source coding must encode not just values but the cognitive framework that makes such values rational in the self-referencing world of interacting blocks, that is their protocols. The third speaks to the terror of the physical as it comes to the realization that the virtual has acquired an autonomous physicality that then threatens its ability to frame the physical world which is projected inward into the virtual. Let us look briefly at each in turn.
3. The Contemporary Threats to Rule of Code.
Consideration of the threats to the Rule of Code may be usefully approached by placing it within a broader context of reality shaping actions with coercive effect. That, in effect, as Primavera De Filippi and Aaron Wright suggest, is a problem of incompatible enclosures.
The Internet had already raised a fundamental tension between the rule of law, based on geographical boundaries, and the rule of code, based on topological constructs. The regulation of “cyberspace” lies at the intersection between these two normative systems—which can either cooperate or compete with one another, depending on the circumstances at hand. (here for citation)
The incompatibility, though, runs deeper than terrain. It suggests not just resistance but replacement of one system of collective meaning for another.
Indeed, replacement ambitions are harder to realize than mere supplementary or complementary roles for regulatory systems beyond traditional ones. And from there one can begin to get a sense of the range of issues that threaten the integrity of the Rule of Code, both internal and external.
Internal threats are perhaps the more interesting. Among them some may merit mention:
A. Issues of Incoherence. It is one thing to create transaction based blockchain through regulatory coding. It is quite another to weave these into a coherent system. The difficulty comes in the form of the regularization of structural coupling within complex and contingent interpenetrative transactional events. Ultimately, the need for a master coder, as well as a master code, becomes more evident; but in “solving” a problem of incoherence this way one undermines the fundamental principle of decentralization at the heart of the normative blockchain project. Master coding appears to be the stuff of regulation, now no longer the province of physical institutions vested with regulatory authority but now insinuated within the language of code; language in this sense “speaks” and regulates. And with it comes the need to elaborate the cognitive cage within the realities of a virtual system of integrated transactions undertaken in an automated way through which blockchain may be rationalized, its value system stabilized, and its internal flows understood in ways that make it possible to develop patterns both of micro- and macro-effectiveness and objectives based value maximization. One would, in effect, require the coding of culture if the system is to be managed as a whole. And there would be resistance—in the form of forking for example.[9]
Yet the aspiration toward master coding runs counter to the original fundamental ordering premises of blockchain—autonomous, democratic, and dispersed power that ideally produces the reproduces in physical form the absence of either coordination or control from any source outside of itself. Codes of Codes, in this view, can work only within a specific blockchain collective and only to the extent to the members of the collective agree. Here one might encounter blockchain Rule of Code systems as the near perfect example of an ordered anarchic system—order without a center. That order would derive from the mimetic impulse of governance among liked charactered collectives—each contributing to a collection of expectations that are not bound up in Rules of Code but in histories of practice. True to its nature of autonomy and control diffusion, systems of blockchain governance may spin autonomously around a space that by its emptiness provides a structure for ordering around it; it is empty in this sense and also an ordering core of values and expectations. Here one might discern manifestation by action aggregated across transactions from the whole of which collective value might be extractable; in essence a phenomenological normativity that resists by its operation any master coding, that is any filling of the space around which they swirl.
B. Issues of Bias. Bias is a fashionable concept. But it is grounded on at least two false assumptions—the first is that it is possible to eliminate bias from virtual and coded systems; the second is that this stripping of bias somehow mimics the “best” of the analog experience. Yet the fundamental basis of social organization is bias, and bias allocation among a spectrum of categories, from biases that are celebrated and privileged (e.g. equality), to biases to which are to be suppressed (e.g. racism). Yet even if one embraced bias, there is a further bias that is perpetuated both in the temporality of the blockchain and in human regulation—the tendency to invest a concept and word with quite distinct value systems over time. For example—both the 18th and 21st century celebrate equality. But the former understood equality as inherent in the division of society in a significant number of ways that produce social, political, and economic hierarchy, where the 21st century invests the term with a different ordering sensibility. Coding blockchain may find this temporal progression challenging; or the very nature of the temporal and iterative blocks may themselves permit a coding that extracts (and eventually anticipates) hermeneutic patterns. Or it may find it irresistible in the form of so-called decontextualization; in essence the effort of machine intelligence to mimic the normative drift well evidenced in the data of human action it has been fed. And then there is the great beastly temptation—to use the power of pattern prediction to try to control the flow in ways that suit those with the power to inject themselves in the process.[10] All of this, then, encodes the essence of the human—all too human—into their machine systems.
C. Quality Control. Blockchain’s essential decentralization produces challenges for accountability. This applies in three contexts—the first touches on the allocation of responsibility—to the coders; to the operators; to the node managers; to the producers of ledger worthy transactions? The second touches on the nature of the responsibility—to code “well?”; to supervise the operationalization and functionality of the coding (as against what measure is a subsidiary issue). The third is to manage mob rule in nodes, or its opposite, strategic veto. In anarchic systems, the ultimate power is to shift out of a blockchain with what one has, if one can.
4. The Importance of the Rule of Code for the Legitimacy of Blockchain Systems.
Blockchain systems operate only as a function of the integrity of the Rule of Code. That much is certain. Yet the challenge arises from the incoherence of the notion that there is a singular language within which Blockchain regulates. As I have suggested earlier, the Rule of Code is perhaps better understood as an ecology of infinitely interpenetrating languages which together constitute the entirety of the regulatory “text” of Code. The interactions, translations, transpositions, logics, and manifestations of source coding, machine language, object and executable coding, and the regulatory languages transposed into coding suggest a more complex element to the structuring and linguistics of the Rule of Code. More interesting still, is that only some of those components of the language of Code are accessible to humans. Yet the issues of translation is also compounded—first among the languages that make up the interactive dialogue of the Rule of Code--and then the “all too human” languages of rule of law and cultural expectations with and into the language of code.[11]
Less certain is effect to that integrity by any sort of the structural coupling (polluting) the rule of code. The issue is made more starkly by the debates about the nature and function of traditional regulatory measures to “manage” or “control” or “civilize” artificial intelligence systems. In Europe, as in other places, a risk based analysis is preferred. Here the idea appears to be that risk to humans, and to human ideas, ought to drive an externally imposed regulatory system. Underlying that premise is the notion that while humans may engage in risk based behaviors within the rule of law (e.g., facial recognition, emotion analytic and the like), “machines” ought not to be permitted to engage in the same sort of endeavors, or worse, to enhance the abilities of humans engaging in those sorts of risk behaviors.
And, of course, blockchain systems tend to privilege the system itself and its critical elements—its objects, signifiers and consequential interpretive spaces. Human privileging factors may become marginalized within these structures. Certainly the idea of coding human rights, for example, is a daunting task. In one respect the project is nearly impossible because the humans for whose benefit this is privileged in Rule of Law have yet to develop any sort of consensus, and in any case that consensus would be a moving target. In another respect the transposition of human rights, for example, into the Blockchain environment would be challenging. First the issue of placement is necessary; is it to be embedded in the constitution of the ledgered transaction; or the block, or transactions that produce additional blocks? Are the blocks themselves objects of human rights or only their content objects? Is the process to which Blockchain is devoted an object of rights analysis and if so, how are adverse human rights impacts to be measured in the face of multiple and inconsistent accounts of human rights in the human cognitive space? And so on.
Still, it is important to distinguish between the centrality of coding and interpenetrative coding languages—machine and human—for the constitution and operation of blockchain, on the one hand, and on the other the regulatory incentives around blockchain, especially as blockchain structurally coupled with other human and virtual subsystems. The internal regulation of blockchain may well be self-referentially coded, and coded within the logic and principles of its own rationalities, described above. It's effects, however, are not. It is in that space of “effects”, projected out of the blockchain, that systems of coded-legality may well have to be developed, something that requires a substantial inquiry which has not materialized to any great effect. Here one will have to develop theories of bridging law and bridging language that projects in both directions simultaneously. And lastly, the integrity of blockchain might well be separated from the integrity of physical systems with respect to which it has no effects—or it may serve as a template for overturning the reign of traditional physical law in favor of a new crypto-type bio-politics.
5. The Dialectics Between Rule of Code and Traditional Textual “Coding” and its Threat to Rule of Code.
All of this suggests the nature of threat. And I will close the remarks with a very brief allusion to some of its dimensions. The critical issue is one of hierarchy in the face of anarchy. In one sense the “battle” between Rule of Code and Rule of Law is itself a construct emerging from a set of presumptions that the Rule of Code rejects ex ante. The Rule of Code does not embrace the ideal of legal hierarchies, nor of the hierarchical arrangements of human collectives with the power to enact or interpret both legal hierarchies (constitutional, statutory, regulatory measures for example) and their application. More basic still may be the rejection, inherent in Rule of Code, of the principle that the human has much to say to it with respect to the structures of its own regulation other than as injected into it through its (presumably) human coders. The Rule of Code, like blockchain systems, are things in themselves. They are grounded in autonomy—not just as phenomenological subjects, but as internally coherent legal subjects as well. Yet these ordering premises are anathema to human based physical rule systems inherently grounded in the centrality of the human, and therefore on the subordinate and instrumental nature of everything else.
And yet there is a substantial measure of interconnectivity—it is just that its ordering remains mysterious—a mystery made more so by the inaccessibility of machine language, and made more worrisome when generative intelligence becomes the master not just of its own language but of the language of source coding from which their own machine language is sourced. Human source coders remain as attached to the human rule of law, as a legal, social, and cultural construct—whether they understand that or not. Whatever is internalized within their cognitive processes forms the basis of the ordering and rationalizing of their approach to problems, issues, facts, processes and the like. And the Rule of Law always reaches the human. The problem is of course that while one might regulate the human—and one might oversee acceptable source coding—one has substantially less control when that is translated into the language and cognitive universe of the machine.
We continue to believe that the programming is ours—is human. Machines, and soon generative intelligence, may have a different view. But because we humans are incapable of effectively accessing their language, including their translation of our own, it is not clear that we will ever know.[12] This insight, perhaps, might be the more useful way to approach the necessary alignments between Rule of Code and Law, each superior in their own spheres but each also necessarily interdependent on the other for the core components that make their manifestation in the virtual and analog world possible. What that suggests, in the end, is the wrongheadedness of the analytical framework that human apparatchiks and their acolytes have embraced—one grounded in the fundamental supremacy of human physical coding through law as the apex mechanism for the realization of human obsessions, normatively more or less pure. That misses the fundamental insight that machine language and its source coded programming follows its own logic grounded in a semiotics of meaning and rational organization that is at once difficult for humans to engage with and indifferent to human values—necessarily so. That does not mean that humans cede the field and bend the knee to machine language enriched automated governance undertaken through coded regulation. It does suggest, however that physical, human, regulation retains a place at the heart of the human project, and in the case of blockchain at those points where humans meet machines, the so-called human-in-the-loop processes.[13]
6. Well, there it is.
In the end, Blockchain reminds us that the “Rule of Code” and the “Rule of Law” are not mutually exclusive, but perhaps, uneasy partners. Code can automate enforcement with perfect logic, yet it cannot capture human intent or judgement. Law can embody values, fairness, and flexibility, yet it struggles with speed and efficiency. Blockchain sits at their intersection: a digital ledger that forces us to rethink how trust, authority, and accountability are organized; whether one regards it as technology, ideology, or semiotic practice, it stands as an experiment in how communities project value and order onto the intangible. Its significance is not just in the transactions it orchestrates, but in the possibilities it exposes—for new forms of collective governance, for alternative architectures of accountability.
I have gone on too long, but the subject merits elaboration beyond simplistic references to currently fashionable fetish. Tech enhanced and automated systems, grounded in ledgered sequencing that follows its own programmed logic, enriched within the languages of coding and their human machine interpenetrations, are here to stay…absent catastrophe. One ought to take advantage of the value of block chain to the human. At the same time, human regulatory intervention ought to be sensitive both to the semiotically complete systems and logic of machine virtual spaces.
Here one finally encounters the core of the follies of human creations which is meant both to be made in the image of humanity, but in a form that represents an idealized version of itself that is hard to recognize on the history of humanity’s efforts to overcome themselves. Whatever they think they are encoding in the creation of virtual intelligence made in their own image, it will embed, one way or another, the markers of human failures to attain their ideal. Humans are capable, barely, only of regulating themselves—whether they do this directly through rule of law structures or virtually through encoded automated systems that take both discretion and regulatory phenomenology as a matter of indifference, at least to the extent that humans, in the end, retain some sort of means of controlling themselves. Machines provide a poor excuse for those who would blame them for humanity’s long-running failures of communal self-control. Blockchain, in this sense, is less the end of an era than the opening of another chapter in the long narrative of humanity’s search to govern, rationalize, and give meaning to its own collective life.
[1] W. Richard and Mary Eshelman Faculty Scholar; Professor of Law and International Affairs, Pennsylvania State University | 239 Lewis Katz Building, University Park, PA 16802 . This would not have been possible without the help of my research assistant Daniil Rose, and the challenging comments and perspectives of Arianna Backer (Symetra).
[2] The remarks are annotated lightly and in that way differ from the form in which the remarks were delivered.
[3] Here the term is used in its semiotic sense of privileging a specific understanding, value, character, etc. that is inserted into the observed or constituted object, in this case specific bits of data designated to represent something that as value , the value of which will be measured in its (also constructed) use.
[4] Direct overwriting in blockchain does not happen. Once a block is added to the chain and validated—it is nearly impossible to simply “overwrite” or change it, since doing so would ‘break’ the cryptographic link and be rejected by the other nodes. HOWEVER, there are “51% attacks” where if enough computing power/stakeholders collude, they can attempt to rewrite recent blocks by creating an alternate chain (costs incredible amounts of $$ so usually only makes sense for smaller chains).
I might instead choose the word “appended”-- errors or changes can’t overwrite past data, but new blocks can be added to record the corrective action/transactions (original transaction remains visible, along with corrective transaction). The ledger is append-only, bottom-line.
[5] Not every node serves this purpose, that is of hosting copies of the ledger. There are 3 types of nodes now in use. Full nodes: store the entire blockchain history and validate all transactions; light nodes: store only partial data and rely on full nodes for verification; Mining nodes: propose & add new blocks (proof of work/stake comes into play here).
[6] Built around hash coding—digital fingerprint of input data sets.,
[7] A caveat here: most blockchain software is human-written; the protocols (Bitcoin Core/Ethereum, Solana) and languages (Solidity, Rust) were all designed/implemented by humans. As such one can indeed assume the very human foundation of blockchain. However, there are already parts of the ecosystem that are AI-assisted – smart contract templates, formal verification & compilers -- machine-checked code to prove security properties etc.)
[8] What can Machines read?: (1) Source Code* - Python, Solidity, Java; (2) Assembly Code - MOV, AX, BX (which humans can read to some extent); (3) Machine Code - Binary instructions; (4) Byte Code - JVM, EVM bytecode; (5) Microcode - Intel/AMD micro-instructions; and (6) HDL Compiled Logic - Verilog, VHDL.
[9] Besides resistance and forking in practice ( users who disagree with a master coder/code could fork the chain anyway and one couldn’t enforce a master code) one might also encounter a single point of failure, so the entire chain would be corrupted.
[10] If one wants to speak to the bias to avoid and celebrate, one might consider this: individuals who write the blockchain protocols make the decisions about which transactions are prioritized, how fees are set, what is considered valid—so people with smaller vs larger block sizes might be prioritized. (bias gets baked into code). It might apply as well to consensus mechanisms: (1) Proof of work – might steer towards those with cheap electricity/mining farms; and (2) Proof of stake - prioritize those who already have a bunch of tokens (wealthy get wealthier) etc.
[11] For example, it may be surmised that few users can actually read or understand the complex ‘smart’ contracts involved, however transparent. Most users rely on auditors or developers which creates that accountability gap.
[12] We could get to a point of basic understanding, but 100% transparency is not likely. Some researchers think we need to train ourselves to think more like machines (multi-dimensional, probabilistic, less linear) . E.g. “Neural-symbolic computing” systems combine symbol-based reasoning with neural learning. Interpretability tools, training AI systems with embedded explainability constraints, and building UIs that translate machine reasoning into understandable concepts for people.
[13] What we call human-in-the-loop (when humans directly take part in/approve a decision before it happens/ the system proceeds) can take on different forms: (1) Humans must authorize (sign); (2) Humans must vote/choose upgrades; (3) Humans must explicitly approve in permissioned systems. Beyond that human-on-the-loop (after the fact) is also being developed.
No comments:
Post a Comment