Foreword for 2023 book edition of Anna Greenspan's PhD thesis 'Capitalism's Transcendental Time Machine'
Tempus fugit, armor manet. So the saying almost goes. Time is all around us, watching our lives and clocking our hours as we grind to dust under insatiable formations of capital. Ever-increasing precision of timekeeping has led to the atrophy of the human attention span, as productivity flatlines. Who built all the clocks? This article sets out to draw parallels between historical insights regarding the ontological status of time and emergent phenomena observed recently in decentralised systems, namely Bitcoin and Ethereum. Cryptocurrencies are timestamping systems at their core. A discretised, linear data architecture sometimes known as ‘the timechain’ provides a reasonable degree of assurance that the network will respect a particular set of transaction orderings which, when chained together manifest a canonical historicity. Are these new legends also being written by the winners? Does capital formation necessitate a temporal polity? The transcendental mechanism known as proof-of-work leaderlessly transmutes immanent, continuous network activities into ordinally sequenced batches of confirmed transactions. As a byproduct of this process, temporal paradoxes are encoded in the beating heart of cryptocurrency network logics. This combination of polity and paradox leads to unintended consequences as chronobandits run wild in the dark forests of the timechain.
“I am separated from myself by the form of time”
Gilles Deleuze, Kant’s Critical Philosophy, p. ix
Time, as the most common noun in the English language, is familiar to all of us. At one level of phenomenological experience at the very least it provides a cumulative and sequential framework of continuity and progression that human conscious experience is ostensibly anchored in. A distinction is drawn by some between the objective, precise and measurable rhythm of devices marking the passage of time (c.f. Plato or Plotinus’ ‘moving image of eternity’, Chronos) and the subject's experience of temporality (c.f. Bergsonian ‘duration’, Kairos). Spiral, cyclical and abstract folk ontologies of time offer alternatives to the western and colonial hegemony of linearised traversals between past and future.
A line (or perhaps an arrow?) may readily be drawn between the symbolic reductions of Newtonian physics and Kant’s transcendental idealism. Issac Newton championed the notion of ‘t’ as a universal, absolute time to render tractable many hitherto unaddressed natural mysteries. He rationalised this by proposing that humans could only perceive relative time indirectly through the movement of for example celestial objects. In so doing, he initiated a paradigm of time-dependent science which many of his contemporaries, not least Gottfried Leibni(t)z vigorously opposed. A century or so later, Kant proposed a radical re-architecting of the ontological hierarchy of the universal dimensions, anointing time as ‘a priori’ and ‘pure internal intuition’ that is no longer subject to movement and correlated to but not subordinated by space. Kant’s transcendental criterion necessitates the existence of a boundary to the "outside", with an immanent plane of experience "inside".
“Time is impassive, more animal than human. Time would not care if you fell out of it. It would continue on without you. It cannot see you; it has always been blind to the human and the things we do to stave it off, the taxonomies, the cleaning, the arranging, the ordering.”
Lauren Groff on David Rooney’s About Time, NYRB
Temporality is often found imbricated with influence, hierarchy and control. The clock signifies precision, consistency and reliably verified measurement, but is simultaneously a tool of control and discipline. By whom does the bell toll? Temporal polities. Organised religion still uses solar and lunar calendars to regiment behaviours and decide when certain rituals take place. When sundials and water clocks were first installed in Carthaginian, Roman and Greek settlements, local bureaucracies harnessed their new timekeepers to regiment workers’ routines. As curfews began to be set by the clock rather than Sun, our newfound temporal regulators became enforcers of social order. Factory owners built crooked clocks to extract further productivity from their exploited labour forces.
Embedded within the device is a question of power: decisions were made to fund, design, construct, deploy, calibrate and enforce the logics of a timekeeper. In enshrining the chronological restraint of temperance as a virtue, does dyschronia imply iniquity? Who gets to decide what time it is? Can there be fairness when the relationship between past, present and future is in the hands of a chrony cabal? Who regulates whom? Do the bounded logics of communication necessitate the existence of insider asymmetries, through time as well as space? If we live by the clock, must we also die by the clock? It is this article’s premise that new forms of durational abstractions - be they machinic, algorithmic, cyberspatial and/or cryptographic - are being used by entrenched powers and capital for essentially the same ends as the old timers. The affordances, limitations and externalities of timechain systems extend into time and space in new and unexpected ways. But the same temporal polities appear to emerge as with many asymmetric scenarios. Regardless of its ontological status, time as a socio-cultural construct functions sufficiently effectively to act as control infrastructure.
“How can one capitalize the time of individuals...in a way that is susceptible to use and control? How can one organize profitable durations? The disciplines must be understood...as machinery for adding up and capitalizing time.”
Michel Foucault, Discipline and Punish, p. 157
As increasingly sophisticated clocks marked the passing of days, months, years and decades, Newtonian and Kantian theoretical approaches to understanding time’s place in the Universe remained largely untested until the advent of relativistic and quantum physics. First, general relativity did away with the notion of temporal homogeneity. There is no universal and absolute time ‘t’. Time is lumpy and uneven, locally consistent and providing partial ordering of events but far from the untouchable ‘inner form’ that Newton and Kant imagined. With quantum mechanics, phenomena such as wave-particle duality, non-localisability, discretisation, entanglement and uncertainty made it apparent that time was subject to and governed by similar factors as energy and space. An observer - or rather, a measurer - could affect the state of a system and cause a superposition to collapse into a particular configuration. This brought into sharp focus the subjectivity inherent in our spatiotemporal conception of reality, despite the impressive formalisms of classical physics and German idealism.
“Time has taken on its own excessiveness.
It is out of its joints.”
Immanuel Kant, Human Dignity, p. 12
Time is often regarded with a thermodynamic lens, to the extent that entropy is considered synonymous with time's arrow. This is largely due to the readily observed gradual increase in disorder within closed systems, such that entropy appears aligned with the passage of time. In some sense classical thermodynamics uses entropy as a thermal clock, though there is still a lack of commensurability with the quantum mechanical conception of entropy (see Orly Shenker & colleagues here and here). Ongoing work on quantum theories of gravity suggests that time is largely an illusory relational construct and/or artifact of conscious experience.
Bitcoin is at its heart a distributed system whose goal is to achieve a leaderless consensus as to the ordinality of a series of occurrences. Bitcoin is a decentralised timestamping server, and the transactions are simply messages changing the effective balances that each network participant has access to. These balances are denominated in the native unit of the system (BTC), and are used to pay transaction fees to miners and function as the de facto currency with which value is redistributed amongst the users of the network. Satoshi Nakamoto themselves used the word ‘timestamp’ fourteen times in the Bitcoin whitepaper. Bitcoin is an abstract timekeeping daemon incarnated through cryptography and thermodynamics.
“In this paper, we propose a solution to the double-spending problem using a peer-to-peer distributed timestamp server to generate computational proof of the chronological order of transactions.”
Satoshi Nakamoto, Bitcoin whitepaper Abstract v1, 2008
Originally referred to in the source code of early Bitcoin software client versions as ‘the timechain’, it is more commonly referred to today with the rather less appealing and uninformative moniker of ‘the blockchain'. Let us attempt to address this: the timechain is a chain of blocks, which are themselves batches of transactions, pages in an abstract accounting ledger. The timechain functions as an information repository that allows anyone to confirm the order and details of transactions without needing to be permanently online, with a new batch of network actions enshrined in a block and being added to the timechain, thereby updating the shared record of balances approximately every 600 seconds.
Scholars of idealism find in Bitcoin a metaphysical temporal fabric with its own heartbeat. A new timekeeping system - indeed, a new kind of time - made manifest by the timechain. Welcome to the chronaissance. This literal quantum leap in abstract time is divorced from celestial influence and mostly unphased by the increasingly subdivided temporal constructions of modern Homo Clockonomicus in pursuit of efficiency and productivity. A new ordinal domain where time reigns supreme. But this is not the end of the story: there is no happily ever after on the timechain. A new breed of necroprimitivist chrononauts exerting control whilst deriving religious zeal and economic gravitas from these energetically costly upstart instruments of temporal synthesis and mutation are in the ascendance.
Timechain technology does not afford a unified, singular chronognostic regime but rather a schizotemporal duality. A continuous cyber-clock mode exists where nodes propose transactions in ‘real-time’ and a discrete block-clock mode ticks to the sequential cadence of confirmed blocks. Proof-of-work is the transcendental mechanism which leaderlessly transmutes immanent, continuous-time network activity into ordinally-sequenced batches. A burnt offering for an indifferent god. In the collapsing of one into the other the opportunity for arbitrages, slippages and other chronically adversarial behaviours emerge. Transactions are confirmed by miners selecting the user-broadcast proposals they wish to include in an upcoming block, typically prioritised by the size of transaction fees paid. The miner finding the next block in effect decides which messages are canonised in the timechain, whilst having a comprehensive view of network-wide proposed transactions. There are a number of strategies that a greedy miner can utilise to extract value beyond the protocol-expected payoff of network-issued mining rewards and user-supplied transaction fees. These approaches mostly require engagement in behaviour not initially anticipated by Nakamoto’s protocol design. These include transaction reordering, virtual resource arbitrage, hostile forks and 51% attacks. These occurrences are rising in frequency, which suggests that the implicit social contract between network stakeholders might be breaking down over time.
Aspects of cyber-clock time (hereafter cyber-time) and block-clocktime (hereafter block-time) were characterised by Anna Greenspan in her doctoral thesis ‘Capitalism's Transcendental Time Machine’ (2000) in reference to a more general conception of cyberspace-time. Greenspan’s thesis predated Bitcoin by approximately a decade, and one might speculate as to a different exemplar of an Aeonic occurrence - in Bitcoin rather than Y2K - were it written a decade later. Greenspan considered cyberspace-time to be inhumanistic, mechanically simulatory and implying quantisation. As cyberspace is nonlocalisable, its regime of time would be transglobal or postglobal - today we might use the term decentralised. An immanent machinic culture (p2p), cyberspace-time would measure nothing outside of its domain of orientation (hard-bounded). As an abstract-yet-empirical method of timekeeping, cyberspace-time would require a larger paradigm shift than the clock was to the calendar. Taking this further, Nick Land characterised Bitcoin as ‘chronogenic process’ (Crypto-Current §0.83, 2018) and the timechain as a ‘transcendental reality criterion’ (Crypto-Current §1.15, 2018). One might even venture as far as to say that Bitcoin is a deus ex nihilo. Benjamin Noys proposes that Bitcoin exhibits ‘stable dematerialisations’ (AHR, 2020), though one might counter that the perceived stability is somewhat illusory due to the eternal contingency of timechain probabilistics (see Zeno-sum Games) and that a metastable dematerialism would be more fitting.
The cyclically rhythmic and discretised temporality of cryptocurrency networks - the block-clock mentioned earlier - is hardly something to set one's watch by. As proof-of-work is a random process involving searching a possibility space iteratively using brute-force computational repetition, the time between candidate blocks that fulfil the network-mandated validity conditions - valid transactions and block construction, hash value below the difficulty threshold - is variable. This is a stocha(o)stic process, and as a result the time between blocks is unpredictable and can differ widely. The network periodically recalibrates difficulty: the probability of a given hash satisfying the conditions for block creation, which in turn serves to adjust the inter-block cadence. In Bitcoin, a median inter-block cadence of 600 seconds is targeted, but it is entirely feasible to take twice as long to find a block, with the next block following just a handful of seconds later. Naturally, this durational volatility smoothens over a larger sample size but for predicting the real-world timing of a timechain event (such as the subsidy halving in Bitcoin, discussed later) can be challenging. A mitigation which is taken in Bitcoin to deter attacks employing deliberately false timestamps also has a side effect of helping make longer-term unions of block and clock times such that temporal averaging measures are routinely used in slow-block networks such as Bitcoin.
Median-Time-Past (MTP) takes the median of timestamps of the previous 11 blocks as a trailing time average, to avoid any issues with inaccurate timestamps, be they accidental or malicious. As Bitcoin does not have an endogenous mechanism to access clock-time, the network relies on miners including temporal attestations inside their proposed blocks. Due to the widely distributed and leaderless nature of cryptocurrency mining, propagating information across the node population may not be trivial due to long distances and variable communication infrastructure. Thus it is entirely feasible that a late block (with a higher block number) may have an earlier timestamp than the preceding block and protocols typically allow some chronological leeway for these reasons before considering blocks invalid. This temporal affordance is two hours in Bitcoin, which corresponds to twelve times the target inter-block cadence. However, the timestamp of the latest block must always be greater than MTP. Thus, MTP is the monotonically incrementing temporal machinery at the heart of Bitcoin’s chronaissance. The timestamps used in Bitcoin employ the Linux format, being 32-bit unsigned integers starting on 1st January 1970. They are not vulnerable to the Epochalypse bug in 2038, when 32-bit signed integers using the original Linux timestamping system will experience temporal overflow. The Bitcoin network will instead ‘run out of time’ ceteris paribus in 2106.
Some years ago in rather unrelated work, I was working towards a series of meta-epistemic frameworks for an approach to classification of cryptoassets and networks called TokenSpace. As a passing comment regarding the lack of price elasticity in the supply of Bitcoin, a formulation of the intrinsic value of Bitcoin measured solely with reference to time (either block-time or clock-time) that was later identified as an example of a Carnapian Ramsay Sentence. It is included below with a slight modification, as the Bitcoin network has undergone an additional subsidy halving since the time of publishing, thereby altering the terms of the thermoeconomic bargain that miners must adhere to in order to continue playing the Bitcoin game.
“Considering the functionality of the Bitcoin network, the current value of one bitcoin may be understood implicitly as the value of 96 seconds of the computational resource directed at defending the network from thermodynamic attacks and providing a high probability of assurance that the integrity of the canonical ledger will continue to be maintained.”
Wassim Alsindi, TokenSpace, 2019 (edited)
The significance of this definition is that it renders explicit the chronoeconomic jeopardy inherent to Bitcoin’s continued security. In order for the defence budget to be met as the subsidy attenuates stepwise by 50% every 210000 blocks, the external market-determined price of Bitcoin must continue to increase. As many acolytes have been saying for years, Bitcoin is ‘zero or moon’. It must continue to escalate in price or it will fail. The difference in the amount of security-time that a single Bitcoin has to purchase before and after each subsidy halving is a simple doubling: in clock-time the most recent halving took this security-time-per-Bitcoin from 48 seconds to 96 seconds (assuming target mean inter-block cadence of 600 seconds), whilst in terms of block-time this number has incremented from 8% to 16% of the inter-block cadence. In essence, the subsidy halving embeds at the heart of Bitcoin (and by extension all Bitcoin-like systems) a Zenoeconomic paradox which creates the conditions for a zero-sum game pitting capital and ecology against each other. Another quirk of Bitcoin’s latent Zenoeconomics is that 50% of all the Bitcoins that will ever be mined were distributed in the first few years of Bitcoin’s life, with a very small set of beneficiaries. All of this effort expended, so that only a few may print time, mint money and decide history.
“Even inside the networks, it’s markets all the way down. Including a transaction in the next block is in reality a market process too. One must out-bid others to motivate a miner to include your transaction in a block, and even then the corollary market for transaction ordering might scupper you with miners extracting ‘your’ value. What we see in today's conception of timechain is the apotheosis of neoliberal deterritorialization. Literally everything: the church, the state, the corporation is subsumed into the network. Everything is marketised.”
Wassim Alsindi, edit of interview quotation in Awham Magazine, Issue 4, 2021.
Front-running and Miner-Extractable Value (MEV) are emergent dystemporal phenomena that many people new to the world of timechains might not be unaware of. A well-established tactic in traditional financial markets, front-running is essentially queue-jumping. In High Frequency Trading (HFT), trading firms would buy up tiny slivers of real estate as close to the Chicago Mercantile Exchange as possible. Proximity reduces response time to key pieces of information, thereby allowing traders to react ahead of competitors.
In the timechain setting, front-running exists due to a mismatch between the temporalities of network and ledger as discussed earlier (Dysclocksia). By acting as temporal mediators, miners get to ‘see into the possible futures’ of the network regardless of whether they win the lottery to find the next block. Miners can 'extract value' beyond the implicit social contract of mining rewards and user-supplied transaction fees at the expense of innocent users by inserting their own versions of previously proposed unconfirmed transactions. In many ways, timechain front-running resembles the ‘Payment For Order Flow’ that HFT firms such as Citadel Securities pay handsomely for to have ‘foresight’ and ‘priority’ over retail investors at mass trading outlets such as RobinHood.
MEV has been on the rise on the Ethereum network in recent times. In addition to the obvious issues it poses for the viability of timechain economics and infrastructure, additional consequences exist. By generating network congestion and pushing transaction fees higher, MEV appears to be an accelerant of ‘timechain gentrification’. Taking a leap further than transaction-level MEV, a recent version of an Ethereum software client promised miners rewards for on-demand chain reorganisations. Block-level MEV is another emerging threat to the perceived structural affordances of timechain networks.
Various mitigation strategies to MEV have been proposed and some implemented in the wild, which provide centralised and/or decentralised solutions by circumventing the need for users to publicly broadcast their transactions to the network before being included in a block. This has been used in the wild in cases such as art collectors or speculators wishing to mint or accumulate multiple NFTs at the floor price of a collection without wanting other parties to be aware of their actions in advance. Without dealing directly with miners, there would be no guarantee that their transactions would go through ‘atomically’ in a single block, thereby raising the prospect that MEV ‘solver’ bots would detect their behaviour and intervene to accumulate floor price tokens. There are a number of pro-MEV arguments, chiefly that arbitrage makes markets more efficient or that protocol builders should build mitigations into their DeFi systems, such as orders which can only be filled by nominated addresses. It is also notable (praise be to Kairos) that MEV has risen to prominence recently in concert with reward-attenuating changes to Ethereum’s proof-of-work initiated by the EIP-1559 upgrade. Are miners taking back control after developers changed the network constitution to reduce their share of the network’s bounty? The keepers of time and money will always have outsize sway.
As alluded to in the previous section, the wildly volatile expense of using public timechain networks is well known and documented. In Bitcoin, the variation of transaction cost scales up or down orders of magnitude depending on demand to be included in the next block. With Ethereum-like networks which afford more generalised computational capabilities, the cost to transact - and space available in each block - are measured in terms of computation. The market for smart contract fuel - known colloquially as ‘gas’ - follows analogous logics to the market for any other scarce resource. Buy low, sell high: as an efficient commodity arbitrageur is wont to do.
As Ethereum-type networks allows for facile tokenisation, and gas can be stored within smart contracts (and therefore tokens), Project Chicago created multiple variants of a GasToken Solidity contract which allows for the minting of tokens which store gas, and do little else. One can envisage a series of ‘gas-bots' which monitor the demand for computation on Ethereum-based networks (as referenced to their long-term averages) and mints GasTokens when network demand is low, then burns them to reclaim, use or sell the computation proxy when prices are high. This might happen when major network events take place - forks, subsidy changes, token launches and so on. A relatively benign use of GasTokens would be as a hedge against volatility in computation prices - just as power suppliers or transportation companies engage in resource futures markets. Miners have had the option of filling their blocks with gas-storing self-transactions, and now a wider spectrum of network stakeholders in principle also have this capability.
Due to proposed changes in the Ethereum network the risk of serious impact by GasToken is not very high, though this may not be the case on other networks. There have been reports that GasToken has had a significant impact on Ethereum Classic and was being held responsible for state bloat. The Ethereum Classic timechain had been growing in size faster than would otherwise, pushing externalities onto node operators in the form of increased storage requirements. Further, some centralised trading venues experienced griefing-type attacks where the failure to limit computation caused some unforeseen value leakages. The GasToken imposes burden upon collectively managed distributed network storage, and as well as potentially pushing computation costs higher (due to the creation of a new secondary market), and storage costs might also rise. GasToken also wastes block space due to inefficiencies in the mechanisms it uses (gas refunds are only partial in most cases), and could be rendered ineffective by networks removing the gas refund for clearing contracts or storage. Indeed GasToken can be thought of as a cross-commodity mispricing of resources by connecting the market for storage with the market for computation, with time as the arbitrage medium. Only those who know the secret may pass to the hallowed land of riskless arbitrage, others must pay in time or space.
Due to the inability of Bitcoin to trustlessly synchronise with external timekeeping systems, Bitcoin cannot verify the accuracy of miner-submitted timestamps and averages timestamps of recent blocks to avoid issues with inaccuracies. There is a long-known theoretical vulnerability in Bitcoin - and Bitcoin-like systems also employing proof-of-work - which uses false timestamps maliciously to game the network. This has come to be known as a ‘time-warp attack’, however has never been knowingly observed on Bitcoin itself. Examples of the attack have however been observed on both Bitcoin testnets and other proof-of-work cryptocurrencies.
The attack starts with a miner submitting blocks with timestamps which are inaccurate but still within network tolerance (up to 120 minutes ≈ 12 Bitcoin blocks). This has the potential to affect the mining difficulty adjustment as it retargets the likelihood of a valid proof-of-work hash to satisfy the network’s conditions as an acceptable block. Proof-of-work is essentially a Sybil-resistance mechanism, it prevents malicious actors from flooding the network with spam or noise by making block creation costly through the hashrate-mediated difficulty adjustment. But when false (early) timestamps are submitted and accepted as valid, the network’s difficulty might adjust downward thus making mining easier. Therefore the attacker can build a longer & stronger chain with more accumulated work than the current one in private, only broadcasting found blocks when they are able to claim canonicity of their alternative history. The reduced network difficulty would potentially allow the attacker to create blocks faster than they can propagate across the network, thereby orphaning many other candidate blocks and essentially monopolising mining. As the work would be lower on the time-warp chain, this would have to be a much longer chain to claim canonicity but the block subsidy would potentially offset the work required. On Bitcoin, a time-warp attack would also necessarily require 51% of the hashrate which presents its own issues for the security of the network (see above).
That this attack has not been observed in the wild on the Bitcoin network suggests that would-be exploiters - despite economic and/or ideological motivations - consider this attack impractical or unprofitable compared to rational modes of mining. Due to the long inter-block cadence on Bitcoin (600 seconds) and the wide window for difficulty adjustment retargeting (2016 blocks ≈ 14 days) the start of an attempt to conduct such a strategy would be immediately apparent to those monitoring Bitcoin mining, either from studying inaccuracies in the timestamps themselves, or by noticing the divergence of median-time-past with respect to clock time. Miners who do not wish to collude with such an attack can also take action, by refusing to build upon blocks with inaccurate timestamps. It is conceivable that users and economic stakeholders in the Bitcoin network would apply pressure to large mining constituencies (farm and pool operators) to boycott suspected time-warp blocks.
These behaviours have been documented on other networks. Firstly, Bitcoin testnet3 - which is a clone of the Bitcoin protocol network but with valueless tokens intended for R&D - uses the same proof-of-work algorithm as Bitcoin but due to lack of token value has a very low network hashrate. As a result, an entity desiring a large quantity of testnet-Bitcoins could direct hashrate to the network. This would ordinarily increase the difficulty, but by using inaccurate timestamps the difficulty can be kept low. There have been cases where a time-warp attack has incapacitated testnet3, until a rival miner directs sufficient computational resource to increase the network difficulty. Testnet3 also has some quirks related to mining difficulty which make it easier to game. More than one circumstance - such as a long period without blocks - allows miners to create a block with the lowest difficulty possible.
More salaciously, the small proof-of-work network Verge (formerly DogecoinDark) was subject to multiple time-warp attacks in early 2018. The exploit approach mirrored that theorised above, but with an important addition that enabled them to conduct the exploit (and claim several million USD worth of XVG tokens at time of attack) as a result of an additional complexity present in Verge as opposed to Bitcoin. As an attempted mechanism to engineer ASIC-resistance - now considered a largely futile way to ensure that no specialised mining equipment can be created for a proof-of-work network - Verge employed a naively-constructed sequential cycling of five different hashing algorithms, each with their own difficulty parameter. As some of the five algorithms already had specialised mining hardware (which may have been privately available if not purchasable by the wider public), the time-warp attacker merely focused their efforts on Scrypt mining, changing the ‘51% condition' to between 0.5 and 10% of the network hashrate. Further, Verge's network difficulty retargeted every 30 minutes (rather than the 2016 block ≈ 14 day window in Bitcoin) making the attack very rapid to get going and allowing attackers to cement control of the network mining. Indeed the retargeting window was several times smaller than the tolerated timestamp drift, making the attack even easier to launch. The attacker achieved a difficulty reduction of 99.999999%, making it approximately one billion times easier to create blocks (low difficulty and mostly empty) and claim the mining rewards on many more blocks than would normally be found under typical network conditions. Lastly, due to further inheritance of code and characteristics from a legacy network which had a completely different architecture, Verge’s consensus mechanism accepted the longest chain by simple length without discriminating for the amount of work accumulated. This meant that the timewarp attacker could maintain their timeline despite much less work going into it than an honest miner’s rival timeline.