Distributed temporal mutations - Consensus-systems as novel temporal regimes

This work is the condensation of research that has occurred over the past two years surrounding the creation of novel timezones based on techno-social processes, which culminated in a series of sculptural objects entitled ‘Temporal Secessionism: Timezones 1 - 3’ presented at ‘Iskra Delta’, the 34th Ljubljana Biennale of Graphic Arts and the 7th Athens Biennale, as well as the digital-only ‘Timezone 4’ presented as part of Projekt Atol’s ‘Plaza Protocol’. This work simultaneously stands as the initial step in deeper theoretical research into the contemporary temporal regime, and the technologies that are giving rise to it, exploring points of tension raised in a chat-only reading group hosted on Urbit.

“If for the moment we assume that computers are equivalent to clocks in this instance then we are 100% living in the world of CPU time.” ~tontyl-pitret, ~sovnub-banleg/temporal-secessionism

1. Distributed temporal mutations The points covered here began as reactions to Anna Greenspan’s PhD thesis Capitalism’s Transcendental Time Machine, around which much of our research orbited when developing Temporal Secessionism. However in the 21 years since its publication, digital timekeeping infrastructure has shifted from centralized and human-centric timezones and scales based on the historical powers of the Nation State and Greenwich Mean Time (GMT) to decentralised clusters of increasingly machine-oriented timezones serving corporate or inter/extranational networks outside of Westphalian groupings. As such, some of the core theses of Greenspan’s work had to be re-examined in light of these developments, and questions asked regarding the nature of these technologies on timekeeping practices and the nature of the digital time(zones) they create.

The focus here specifically is the still-nascent use of blockchains and Distributed Ledger Technologies (DLTs) as replacements for centralised server-client software architectures. As an increasing amount of digital infrastructure begins to rely on these distributed temporal architectures, we examine what some of the broad-strokes of a consensus-based temporal regime might look like in light of Greenspan’s work, and tensions that arise from her conclusions when applied to these novel technologies.

Greenspan’s thesis focussed on establishing “a connection between abstract thought and material practice” [pg1], highlighting how modern capitalist time is a categorically distinct and primarily synthetic conceptualisation of timekeeping compared to previous timekeeping systems. Disassociated as it is from the calendric cycles of history and instead reliant on fungible, qualitatively identical units of time defined by the controlling structures of the working day, the “modern conception of time - produced in both transcendental philosophy and by the culture and technics of capitalism - involves a distinction between time as a formal structure and the historical variations which occur inside it”. [pg 184]

The ability for this conceptualisation and production of timekeeping as we know it - according to Greenspan - is rooted in the transcendental philosophy of Kant, with the movement taken by his ‘Copernican Turn’ freeing time from its conceptual subservience to space, and positioning time as the deepest of inner senses within which all else occurs. Detached from the “cyclical patterns of astronomy […] thus emerged a formal time conceived of as a purely quantitative framework that was considered to be separate and distinct from the historical content which happened to fill it”, which was instantiated by “the culture and technics of modern capitalism” [pg184].

Thus the “features of formal time are not static, but rather the result of an ongoing machinic process that produces the formalization of time”, with the the formal time of post-Y2K capitalism being “instantiated by the digitization [sic] of both time and money that takes place in cyberspace”. [pg 187] 2. From command-and-control to consensus regimes The development of timekeeping practices from sundials through to mechanical and digital clocks, alongside the ever-increasing precision of digital clock-time, can be mapped to progressive instantiations of globalised capitalism’s control mechanisms in a manner that is both fairly non-controversial and the subject of extensive historical analysis and research. As such, this will not be our focus here.

The development and instantiation of blockchains and DLTs - the focus of our efforts in this brief text - appear to have upending the power distribution of the post-Y2K digital, globalised time of Greenspan’s cyberspace: time, instead of being passed down a tree of hierarchies unilaterally from centralized timekeepers to other digital machines in a network of temporal command-and-control, is now produced as the outcome of equally permissioned network nodes participating in a series of games in order to agree on a shared history, referred to as consensus.

Consensus-time thus appears as a sidestep from the veneer of capitalism’s ‘smooth progression’; perhaps the beginning of a novel regime of techno-mediated temporality defined not only by a change in power distribution, but also in its machinic self-contained-ness, as well as its ability to forgo the need for a single globalised standard to function, instead arising as networked instances of mutually-independent, decentralised time systems.
2.1. Unix as temporal command-and-control The digital time of cyberspace that Greenspan laid out in her thesis is also known as Unix Time, which is a 32-bit integer representing the number of seconds that have occurred since the Unix Epoch - 00:00:00 UTC on 1st January 1970. “Early Unix engineers picked that date arbitrarily because they needed to set a uniform date for the start of time, and New Year’s Day, 1970, seemed most convenient.” [https://kb.narrative.io/what-is-unix-time#:~:text=January%201st%2C%201970%20at%2000%3A00%3A00%20UTC%20is,%2C%201970%2C%20seemed%20most%20convenient.]

A strange hybrid of the calendar and the clock, Unix Time is the first instance of digital time. How Unix time is distributed is fairly simple: computers ping timeservers as they come online, receiving as a response the current Unix Time. This is necessary for them to be able to properly interact with network infrastructure such as mail servers. Timeservers are themselves simply servers which respond to these requests, and organised into a strict hierarchy at the top of which is a ‘master’ atomic clock.

The power structures baked into Unix Time’s software architecture means that the time is unilaterally fed from timeservers to any other networked machines, with a global structure of Stratums and complex protocols such as Network Protocol Time (NTP) facilitating this global synchronisation.

“With common computers, we take timekeeping for granted. However, there is a rigorous mechanism that works behind the scenes. The Network Time Protocol (NTP), for instance, addresses the timekeeping issue using a hierarchy of servers distributed globally. This includes up to 15 Stratums the routing paths of which are developed to synchronize in the most optimized manner. This is also enabled by the construction of a Bellman-Ford shortest-path spanning tree that decreases both latency and transmission time inconsistencies.” [https://iohk.io/en/blog/posts/2021/10/27/ouroboros-chronos-provides-the-first-high-resilience-cryptographic-time-source-based-on-blockchain/]

There is a clear power differential between the master time servers and client computers, with a simple and unidirectional flow of information occurring from the former to the latter (not to mention that atomic clocks are only feasibly able to be run by governments and/or government-funded research labs).

Furthermore, NTP must take into account the locations of both the timeserver and the client, given variables such as network latency and the physical constant of the speed of light. Unix time is not trivially spatialised, but relies on entire networks of strictly hierarchically organised nodes providing an intermediation between the speed of light and spatial distribution of nodes, all in order to facilitate the homogenous and globalised temporality of post-Y2K Capitalism handed down from atomic clocks to all clients in the network.

2.2. Technological infrastructure development since Y2K: the shift to consensus-based machine-to-machine temporalities As previously mentioned, there have been substantial developments in networking software since the turn of the millennium, a subset of which we wish to explore as sources of novel temporal regimes: systems which rely on decentralised consensus.

“Decentralized systems are distributed systems where a group of independent but equally privileged nodes operate on local information to accomplish global goals. These systems lack a central controller that exercises governance, supervision and control over the system, thus allowing power to be distributed over the network in a more uniform and fair manner. […] Consensus is a shared view of reality that is agreed upon between different parts of a system” [https://medium.com/orbs-network/blockchain-and-decentralized-consensus-108845a091cb]

Consensus exists as an answer to the question of how to decide on an order of what has happened within distributed or decentralised networks, where there is the necessity of achieving a ‘total global ordering’ of transactions, but no ‘god’s eye view’ privileging one actor with seeing through the fog of cyber-consensus trickery. Put otherwise, consensus mechanisms (the different ways in which networks achieve consensus) are utilised to create a global state that is agreed upon by the majority of the network’s constituent nodes, and are the backbone of technologies used to maintain technologies such data centers (Google’s Raft, and Paxos algorithms) and cryptocurrencies, as well as their underlying ledgers, blockchains.

All blockchains rely on a decentralised consensus mechanism in order to achieve consensus in an adversarial environment - i.e. one where the honesty of any other node in the network is suspect - but the salient point for our work here is to imagine consensus-machines as a decentralised clock, the ticks of which are blocks of transactions (network events), each writing a block of shared consensus history. Time becomes little more than a series of linked hashes broadcast between the network nodes, constructed from the activities that make up the network itself, with backwards-computability subordinating the ticking of the clock as the arbiter of history - don’t trust, verify!

“There are different timestamping techniques used across a range of blockchain ledgers, however, they aren’t necessarily very accurate. For example, Bitcoin uses timestamps for consensus security reasons, but not primarily for timekeeping; and in Ethereum, on-chain timestamps are determined by miners whereas the consensus won’t technically block or verify those for validity.” [https://iohk.io/en/blog/posts/2021/10/27/ouroboros-chronos-provides-the-first-high-resilience-cryptographic-time-source-based-on-blockchain/]

Unix Time is utilised in these systems as a secondary layer of consensus manufacturing and game-theoretical assurance as to the validity of the incoming transactions, repurposed from the single source of centralised truth to merely a useful tool for optimising consensus finding via difficulty management (for more on this, see Wassim Alsindi’s contribution to this publication).

3.0. Decentralised consensus as the crux of a new temporal regime From this move - from Unix’s reified hierarchies mirroring political tradition to decentralised consensus - we can glean aspects of the future we’re beginning to inhabit: 3.1. Consensus negates spatialisation An important crux point around which the new architecture of time pivots is that consensus exists to negate the spatialisation of nodes in a network which must agree on the ordering of things in a decentralised manner - nodes which create a shared history, producing time as the outcome of consensus games. Within these decentralised networks of crypto-temporality, time production is no longer globalized, it is non-spatialised, and exists to partially negate the spatial issue of differing network latency in a manner that doesn’t rely on centralised keepers of the clock.

The development over the last decade, from Blockchain 1.0 (Bitcoin) to Ethereum as a ‘world computer’ to the present of application-specific, interoperable internets of blockchains appears as evidence of the increasing specialization and fragmentation of temporal infrastructure: time is splintering, globalized time is no longer solely necessary, and pockets of non-localized timezones functioning as pockets of machine time are beginning to operate our infrastructure, standing in stark contrast to the networks of formal time such as NTP. 3.2. Time != Money, Time = Consensus
“In a regime in which time is equated with money, power must develop “a new technique for making profit out of the movement of passing time."" [pg131]

Not only did cyberspace time demand ever more granular temporal subdivisions in order to function as frictionlessly as possible, but it led to “[t]he convergence of time with money” [pg186], as formal time was able to be quantified and each atomic unit of time accounted for. Indeed, Greenspan goes so far as to claim that ‘time = money’ was “perhaps the most important synthesis in the capitalist regime” [ibid.].

Consensus-time, however, could be said to not only equate time and money, but is actually constituted a list of new additions to the shared global state, the motivation for creating which is the creation of time itself in the form of blocks (at least in Proof of Work systems such as Bitcoin, Ethereum, and Monero), with the ‘winning’ miner of said block gaining all the transaction fees contained within it.

This is yet another abstraction laid over the form of synthesis Greenspan already highlighted as constitutive of capitalist time. Just as the “homogenous” time of capitalist societies - that have “assimilated natural time into the machine”[pg85] - has become “a synthetic operation which combines the hour, minute and second with the calendric anchors of day, month and year.” [pg119], decentralised time production stands as a transcendental technology of time which refuses any calendric synthesis, instead relying on the ordinal march of blocks of transactions.

If the temporality of capitalism relied on “the universal equation of time with money, an abstract process of production which mobilizes both the time of the clock and the time of the calendar into a particular synthetic regime”[pg119] in order to function, the synthesis of decentralised time production - a universal equation of time with consensus - subordinates money into the mechanism of temporal production itself.

You don’t have to look too hard in the depths of BitcoinTalk to find ‘zombie’ blockchains - projects in which no transactions took place and/or no blocks were mined for rewards, and thus time ceased. For those able to kick-start the incentive to mine, however, time appears as the product of a perpetual consensus machine. 3.3. The Shift to Ordinal Reality Consensus-time as the aggregation of transactions into blocks is an instance of ordinal time dislocated from the cardinal ticking of the clock in cycles of hours, minutes, and seconds. This is a machine-to-machine (m2m) time of semi-autonomous infrastructures of trascendental time which stand in reaction to the temporal command-and-control infrastructure of GMT.

Far from being a machine dystopia however, consensus-time offers the possibility of freeing us from the atomised grind of Unix and NTP, instead allowing for participation in a nexus of decentralised transcendental timezone production. WAGMI.