What technical affordances does a Smart Contract have? With the definition of the token standard ERC-721 in 2017, the authors of the proposal probably unintentionally predetermined the formal technological definition of a work of art “on the blockchain” – a non-fungible token that points to an individual URL.
As a result of this definition, three years of NFT hype followed in which largely centralized platforms like Opensea used NFT’s primarily as a distribution mechanism for images and animations. The resulting duality between artwork and software, i.e. between predestined distribution mechanism and artistic medium, complicated the concept of smart contracts as an independent self-referential medium. Looking at the state of the NFT ecosystem, we can assess that the majority of NFT’s consist of largely unmodified template code pointing to a visual medium found at the endpoint of a URL. In this case, the only function of the smart contract is to implement the transaction rules of the ERC-721 and to reference the selected media. The platonic duality of form and appearance cannot be resolved – the visual medium and implemented code are fundamentally disconnected.
If we want to construct the possibility of a materialist counter-practice that not only comments on technology but also approaches it as a medium, it is not necessarily worthwhile to look at “media art” but to instead analyze the alienated immanent practices that consider code itself as a central form of expression. Coding as a socio-cultural practice of procedural sign rewriting can arguably be traced back till the beginning of the 19ths century – bound to the invention of programmable machines. With the later advancement of technology and the standardization of machine code and underlying hardware, “coding” as a practice became a foundation of modernity. Coding is therefore the production of programs which can be rerun on any emulation of a universal Turing machine. The resulting program is a formalized operation, utilizing the underlying hardware while being bound intrinsically by its internal scope and namespace. What does a turing machine do? It computes. The practice of coding becomes inherently self-referential since the performance can only be evaluated through the compiling and running of a program. The semantics of program code are novel insofar as they provide an immediate vector of utilization through the compiler itself. An idea becomes only relevant through the proof of successful operationalisation.
If we are in search of cultural references that consider the materiality of code as a post-conceptual medium, we do not necessarily have to use the lens of fine art: The most materialistic and experimental approach comes, as so often, from a scene that is largely ignored by fine art – the demoscene. Born in the 80’s with the advent and increasing availability of general purpose home computers, the demoscene produced real time demos of animations and music running on limited hardware like the Atari 800. The demoscene was only concerned with one specific question: How far can one push an existing soft- and hardware architecture to create art? This approach differs completely from a widely shared conception of art by decentering the artist in favor of the existent materiality of the machine.
“Basically, the classic media artist asks himself “What kind of technology would I need to realize my concept”, while the demoscener asks “What can I achieve with the hardware in my hands” (…) demos rely on the artistic materiality of computers. Their aesthetic values are not based on infinite digital flexibility, but on the physical restrictions of computer platforms. That’s why demoscene is not about an idealistic projection of future technology, but about the creative appropriation of present hardware. Thus, it overcomes the frustration of “media platonism”, as Kittler would probably say.” 
As Botz noted, the demoscene is concerned with the a priori of technology, treating the creation of code as a synthetic synthesis while dissecting the infrastructure of computation to expose the source of machine production. This specific notion of Code is flattening the symbolic and visual representation with its immediate vector of utilization (the runtime). Kittlers notion of technical materiality comes to mind: What is a priori given, is dependent on the technical modes of mediation. For Kittler, even cultural and epistemological structures are contingent in the formation of the technical a priori.
In demoscene applications one finds a strong focus on the conditions and limitations of the hardware – starting with selecting distinctive hardware architecture and utilizing ultra-specific optimizations most commonly written in low level programming such as assembly. Similar experiments have appeared in Netart with “pure” software as material – one example would be the approach of the art collective Jodi, which Alexander R. Galloway described in a similar trajectory as “Infrastructural modernism”:
“Jodi are thus stubbornly out of step with the dominant rhythms of contemporary art. Less obsessed with the cultural or social effects of new media, Jodi orient themselves toward the specificities of hardware and software. The resulting aesthetic is, in this way, not entirely specified by the artists’ subjective impulses. Instead, the texture of code and computation takes over, and computing itself—its strange logic, its grammar and structure, and often its shape and color—produces the aesthetic.” 
With the increasing balkanization of software and hardware, new and mostly turing-complete software environments are emerging, which exhibit special features and characteristics. In the context of blockchain art, the most relevant is conceivably the ethereum virtual machine(EVM). (to be continued …)