The first and perhaps most viscerally felt singularity is temporal – the compression and acceleration of events beyond human cognitive capacity. This is not simply about life moving faster; it is about a fundamental breakdown in our relationship with time itself.

Consider how political decisions once unfolded. The Cuban Missile Crisis of 1962 played out over 13 days, with leaders having time to deliberate, consult advisors and consider consequences. Today, major policy decisions can be made, announced, reversed and forgotten within a single Twitter cycle. The median time from crisis emergence to first ministerial tweet in OECD governments fell from 72 h in 2009 to 5.5 h in 2024 – a 13-fold contraction. The Syrian chemical weapons debate of 2013 lasted 2 weeks; by 2020, equally momentous decisions were cycling through social media in hours.

Philosophically, this compression weaponises kairos – the ancient Greek concept of the opportune moment, the time when chronological time [chronos] breaks open to reveal possibility. Once, kairos was a gift, a rupture in routine that allowed for transformation. Now it arrives externally authored by algorithms, a manufactured crisis every hour demanding immediate response. We live not in time but under temporal tyranny, where each moment is flattened into discrete, commodified units. This represents our devotion to acceleration as what we might call a negative theophany – a revelation not of the sacred but of the idol we have made of efficiency itself.

In practical terms, this means that the acceleration of institutional response times has not improved decision quality; rather, it has replaced deliberation with reaction, transforming governance into a continuous emergency mode that precludes strategic reflection.

Indigenous time concepts offer a stark contrast. The Aymara people of the Andes conceived of the past as ahead (because we can see it) and the future as behind (because it is unknown) – a complete inversion of Western linear time that prevents the optimisation of an unseen future. The Māori concept of whakapapa connects past, present and future through genealogical time, making decisions based on relationships across generations rather than quarterly reports (Lai & Gershman, 2021).

The acceleration is measurable across domains. High-frequency trading now accounts for 55% – 70% of U.S. equity volume, executing millions of trades in microseconds, creating market movements that conclude before human traders can even perceive them beginning. The 2010 Flash Crash saw the Dow Jones drop 1000 points in 36 min – erasing $1 trillion before partially rebounding – a financial earthquake that was over before most people knew it had started. According to SEC data, average trade execution times have fallen from seconds in the 1990s to microseconds today, with some high-frequency trading firms operating at nanosecond speeds U.S. Commodity Futures Trading Commission, & U.S. Securities and Exchange Commission (2010). In the context of a meta-crisis approach, the ‘Flash Crash’ serves as an illustrative example of the various dimensions of the Agency Paradox. Following the flash crash that occurred in 2010, the speed of algorithmic transactions far outpaced the ability of human traders to intervene. Investors watched as the markets plummeted, marking a historical shift where decision-making transitioned from human judgement to machine logic. Despite the chaos, no single investor or institution was held fully accountable for the consequences of the crash (Borch, 2016). Responsibility was spread among the programmers, firms and regulators who developed the algorithms and systems for the markets, highlighting the displacement of agency within algorithmic systems. In the aftermath of the flash crash, market participants faced a challenge in adapting to the new reality of digitalisation. To remain competitive, they needed to design their own trading algorithms capable of operating at similar speeds. As a result, human investors had to increasingly ‘think like algorithms’ to stay relevant in the changing landscape of trading (Akansu Ali, 2017; Kirilenko et al., 2017). But the numbers only hint at the deeper transformation. We are experiencing what we call ‘temporal triage’: a state where we must constantly choose which accelerating processes to pay attention to while letting others run on autopilot. The average knowledge worker checks email every 6 min, not because they are obsessive but because the pace of workplace communication demands it. Microsoft Office 365 telemetry shows knowledge workers’ email dwell time averages just 3 min 12 s before task-switching. Gloria Mark’s research on workplace interruptions shows that interrupted work significantly increases fragmentation and cognitive strain (Mark et al., 2005).

Psychologically, this induces what researchers call decision fatigue, the deteriorating quality of decisions made after a long session of decision-making. But it goes deeper. We experience a form of learned helplessness specific to time itself. When interruptions blur the boundaries between past and future selves, our life story, once a coherent thread, shatters into a spray of disconnected moments. We lose what psychologists call ‘time perspective’: the ability to connect present actions with future consequences and past learnings (Holman & Zimbardo, 2009). Identity itself becomes fragmented, compressed into an eternal, frantic now, where the self struggles to maintain coherence across the blur of moments (Lai & Gershman, 2021; Lawrence, 2025).

The Jevons Effect on time is particularly cruel. Digital tools promised to save us time – email was faster than letters, smartphones more efficient than desktop computers. Yet each efficiency gain has only accelerated expectations (Freire-González & Puig-Ventosa, 2014). Where a letter might have expected a response in days or weeks, emails demand hours or minutes. We have not gained time; we have lost it to ever-escalating velocity (Hussain, 2024; Jevons, 1865; Waligóra, 2019).

The human impact goes beyond stress or busyness. We are losing what philosophers might call temporal sovereignty, the ability to control our own relationship with time. When algorithms determine our schedules, notifications interrupt our thoughts and the pace of events exceeds our ability to process them, we become temporal subjects rather than temporal agents. We live in time, but we no longer own it.

While temporal compression creates the conditions for the second transformation, it also necessitates it. As human cognitive capacity reaches limits, algorithmic delegation becomes not just convenient but essential – leading to our second singularity: The displacement of human agency. This temporal fracturing creates the perfect conditions for our second singularity – as human judgement becomes overwhelmed by pace, we increasingly delegate decisions to the only entities that can operate at digital speed: Algorithms.

The second singularity is unfolding so smoothly that we barely notice it: The saturation of AI into every aspect of human decision-making. This is not about robots taking over or superintelligence emerging. It is about something more subtle and perhaps more profound – the gradual replacement of human judgement with algorithmic optimisation (Lai & Gershman, 2021; Schaefer & Wickert, 2025).

The numbers tell part of the story. Algorithmic trading now accounts for 55% – 70% of equity trades (65% – 70% in foreign exchange [FX] markets). Nearly, all Fortune 500 companies use AI systems to screen resumes before human eyes see them – 75% of resumes are filtered by Applicant Tracking Systems before any human review. Forty-two per cent of Bloomberg’s first drafts are now AI assisted. Every major social media platform uses algorithms to determine what billions of people see, read and think about each day. Model interpretability remains below 20% in systemic finance. But these statistics mask a deeper transformation (Schaefer & Wickert, 2025).

We are beginning to think like algorithms. When we craft a LinkedIn post, we unconsciously optimise for engagement. When we write emails, we anticipate how they will be sorted by spam filters. When we apply for jobs, we keyword-stuff our resumes to pass through applicant tracking systems. Without realising it, we are internalising the logic of the machines we have created.

This creates what we call the ‘Agency Paradox’: The tools designed to enhance human capability are subtly replacing human judgement. We do not just lose the ability to make certain decisions; we lose the capacity to imagine that such decisions could be made differently. When ‘the algorithm decided’ becomes an acceptable explanation for everything from loan denials to hiring choices, we have crossed a threshold. At the core of the agency paradox discussed here is a systemic reliance on algorithms that undermines individual independence. In our technology-driven era, algorithmic systems are used in various applications, from search engines to financial scoring models (Fuchs, 2018). These algorithms subtly influence human behaviour in significant ways. For instance, activities such as listening to music, making new friends or applying for a financial loan are increasingly controlled by algorithms. The interactions that arise from our daily choices, both in life and business, as well as from systems designed to minimise risk, may threaten our autonomy. While we might feel that we are making free choices, our options are often regulated and influenced by algorithms driven by hidden logic. This leads to a more profound question: If our freedom of choice is shaped by these algorithms, can we truly be considered autonomous? For this reason, it is critical to implement responsible AI (Barredo Arrieta et al., 2020; European Commission, 2019).

What we term the ‘Agency Paradox’ builds on Bainbridge’s (1983) ‘ironies of automation’ – the phenomenon where automation removes humans from routine control loops yet leaves them responsible for system failures. We extend this concept by demonstrating how tools ostensibly designed to augment human capability systematically erode the situational awareness, practice and skills necessary for meaningful intervention. The paradox deepens as humans, attempting to remain relevant, begin mimicking algorithmic logic in their own decision-making – optimising resumes for keyword density rather than meaningful experience, crafting social media posts for engagement metrics rather than authentic expression. Thus, the tools designed to enhance human judgement instead reshape it in their own image (Carr, 2014).

Philosophically, this displacement commits what we might call an ontological crime violation against not just what we do, but what we fundamentally are. Heidegger warned of humans becoming ‘standing reserve’ [Bestand], resources awaiting optimisation. But what we are witnessing goes further: A re-ontologising of humans as machines-in-waiting, where consciousness itself is reduced to information processing. This is reverse Gnosticism – instead of spirit ascending from matter, we witness spirit being flattened into code, the soul compressed into algorithmic tokens. The sacred complexity of human judgement, with its intuitions, contradictions and wisdom, is reduced to decision trees and probability weights (Schaefer & Wickert, 2025; Whitehouse, 2025).

In governance terms, this re-ontologising means that human oversight becomes structurally hollow: The person nominally ‘in the loop’ lacks the situational awareness, practice and contextual judgement needed to intervene meaningfully when algorithmic systems fail.

The European Union’s (EU) AI Act (European Parliament and Council, 2024) embodies these principles in law, requiring high-risk systems to maintain ‘meaningful human oversight’. Canada’s Algorithmic Impact Assessment mandates friction – forcing agencies to slow down and consider consequences before deployment. These are not perfect, but they show optimisation can be legally constrained.

The impacts are starkly unequal. Artificial intelligence hiring tools penalise resume gaps, systematically discriminating against anyone who’s taken time for caregiving, illness or life transitions (Ajunwa, 2020; Dastin, 2018). Predictive policing algorithms direct increased surveillance to communities that are already overpoliced, creating self-fulfilling prophecies of crime. Those with the least power in society face the most algorithmic control, while those with resources can still access human judgement – for a price.

Psychologically, we experience what Albert Bandura termed ‘moral disengagement’ the deactivation of moral standards through the displacement of responsibility (Bandura, 1999; Bandura et al., 1996; Ke & Li, 2024). When we offload judgement to systems, we do not just lose agency; we lose the sense of moral weight that comes with decision-making. This fosters a particular form of alienation, especially acute in marginalised groups, where algorithmic biases compound historical trauma. The psychological burden is doubled: Not only do these communities face discrimination, but they must also contend with their attribution to ‘objective’ systems, making resistance feel futile (Ke & Li, 2024).

Perhaps the most troubling problem is the opacity problem. When AI systems become too complex for their creators to understand, we face decisions we cannot explain, made by systems we cannot interrogate, optimising for goals we may not even share. We are not building tools anymore; we are building alien intelligence that operates by its own logic while wearing the mask of objectivity (Whitehouse, 2025). The convergence of temporal acceleration and algorithmic delegation creates the preconditions for our third and most dangerous transformation: The fusion of previously separate risks into cascading system failures. The combination of temporal compression and algorithmic displacement culminates in our third singularity – a fusion of previously manageable risks into cascading system failures.

The third singularity may be the most dangerous because it is the hardest to see: The fusion of previously separate risks into what systems theorists call a ‘resonant threat complex’. Climate change, pandemics, financial instability and technological disruption – these are not separate challenges anymore. They have become aspects of a single, interconnected meta-risk where each crisis amplifies the others (Lawrence, 2025; Rowson, 2020; Whitehouse, 2025).

The 2021 Texas freeze provides a textbook example of cascade dynamics, which is compared in Table 1. An unusual cold snap was hardly unprecedented, but the system’s optimisation for efficiency had eliminated every buffer (Freire-González & Puig-Ventosa, 2014). Natural gas facilities, optimised for normal conditions, froze. The power grid, running at minimal excess capacity to maximise profits, collapsed. Water treatment plants, dependent on electricity, failed. Hospitals, already stressed by COVID-19, saw generators fail. Supply chains, operating on just-in-time principles, could not deliver food or fuel (Doerfler, 2021; Tran et al., 2024).

What killed 246 people was not the cold – it was the optimisation. Each system had been refined to eliminate ‘inefficiency’, which meant eliminating the redundancy that could have absorbed the shock. The deaths were not evenly distributed either; they concentrated in low-income communities that lacked the resources to create their own buffers against system failure.

Philosophically, these cascades represent a system attacking its own vital organs in pursuit of efficiency. What Texas revealed was not system failure but system success by the wrong metrics – metrics derived from extractive capitalism’s growth imperative rather than human flourishing.

Indigenous wisdom offers profound alternatives. Aboriginal Australian fire management, practiced for 50 000 years, creates mosaic landscapes through controlled burns – patches at different regeneration stages supporting maximum biodiversity. This looks ‘inefficient’ to Western eyes trained in monoculture. Yet it is precisely this inefficiency that prevented the catastrophic fires now plaguing Australia. The land was optimised for resilience, not yield – a fundamentally different target that emerges from relational rather than extractive ontology (Auzoult, 2021; Freire-González & Puig-Ventosa, 2014).

The COVID-19 pandemic revealed the same dynamic on a global scale. A biological risk became an economic crisis, which became a mental health crisis, which became a political crisis, which accelerated digital surveillance, which normalised new forms of control. Each intervention created new vulnerabilities. Remote work accelerated digital transformation but also digital exhaustion. Economic stimulus prevented immediate collapse but inflated asset bubbles. Vaccine development showcased human ingenuity while revealing deadly inequalities.

Psychologically, interconnected risks amplify what Durkheim called collective anomie – the breakdown of social bonds and shared meaning in times of rapid change. Yet crisis also catalyses what psychologists’ term ‘post-traumatic growth’. The communities that survived Texas best were those with strong social bonds, mutual aid networks and local knowledge – human systems that had not been optimised away. When we reframe cascading failures not as random catastrophes but as systemic revelations calling for relational resilience, we open space for transformation (Acharya et al., 2017; Lawrence, 2025).

The terrifying beauty of the Risk Singularity is that it makes traditional risk management obsolete. When risks are interconnected, addressing one can amplify others. When systems are optimised to their breaking point, small perturbations can trigger large collapses. When everything is connected, nowhere is safe – unless we change what we are optimising for (Lawrence, 2025).

The key concepts used throughout this framework are summarised in Table 2. Figure 1 visualises the convergent-singularities framework and the self-reinforcing role of the Jevons Paradox.

The drive to optimise is not new. The Industrial Revolution was essentially an optimisation revolution – factories dramatically increased output per worker, railways moved goods faster and telegraphs accelerated communication. But each wave brought hidden costs that compound into today’s meta-crisis.

Consider the enclosure movement in England, arguably optimisation’s first large-scale experiment. Common lands that had sustained communities for centuries were ‘rationalised’ into private property. Agricultural output increased dramatically. So did homelessness, as peasants lost access to the commons (Allen, 2004). The optimisation of land use created the first industrial working class – people with nothing to sell but their labour, forced into factories where their movements could be optimised in turn.

The early 20th century saw Taylorism reduce workers to efficient machines, with every motion studied and streamlined.

Frederick Taylor’s time-motion studies promised scientific management but delivered what critics called ‘the degradation of work’. Workers lost not just autonomy but the very knowledge of their craft, as thinking was separated from doing and concentrated in management. This was not a side effect – it was the point. As Taylor himself wrote: ‘All possible brain work should be removed from the shop and centred in the planning department’ (Adler, 1993; Cooper & Taylor, 2000).

The Green Revolution of the 1960s optimised agriculture, dramatically increasing yields through monoculture, synthetic fertilisers and pesticides. World hunger would be solved through efficiency (Allen, 2004). Instead, we got soil depletion, water contamination, seed dependency and the collapse of agricultural biodiversity. In India, traditional farming systems that had evolved over millennia – supporting hundreds of crop varieties adapted to local conditions – were replaced with a handful of high-yielding varieties dependent on chemical inputs. The result: Short-term abundance, long-term fragility (Evenson & Gollin, 2003).

Each wave of optimisation brought genuine benefits – more food, goods and capabilities. But each also created new problems that the next wave promised to solve. The pattern is consistent: Optimise for one variable (yield, speed, profit), create unexpected consequences in others (soil health, worker autonomy, community resilience) and then optimise again to address the new problems. The meta-crisis emerges when this cycle accelerates beyond our ability to understand or control it.

What is different now is that optimisation has become universal, simultaneous and self-reinforcing. Previous optimisation waves were limited to specific sectors and unfolded over generations, allowing time for adaptation and regulation. Today, AI serves as a universal optimisation engine, accelerating every domain simultaneously. The speed of change has outpaced our ability to understand it, let alone control it. Binding these three singularities together is a 150-year-old economic principle that reveals why each attempt at optimisation deepens our crisis.

At the heart of the meta-crisis lies a paradox identified by economist William Stanley Jevons in 1865. Observing the coal industry, Jevons noticed that more efficient steam engines did not reduce coal consumption – they increased it. Efficiency made coal cheaper to use, which expanded its applications, which increased total demand (Klitgaard, 2022; Waligóra, 2019). The very attempt to conserve resources accelerated their depletion (Hussain, 2024; Jevons, 1865).

This paradox now operates everywhere, but its effects vary by domain and culture. Not all efficiency gains trigger a full rebound. Danish home insulation shows only 30% rebound – energy savings persist because heating has natural comfort limits. But digital domains differ crucially: Near-instant elastic demand means every efficiency gain immediately becomes a new baseline expectation (Liu et al., 2025).

The Jevons Paradox takes different forms across contexts:

Digital Communication: Email made communication 100x faster than postal mail. Result: 1000x more messages. Slack promised to reduce email. Result: Another channel requires constant monitoring. Each efficiency improvement increases total communication load.

Transportation: Fuel-efficient cars were supposed to reduce gas consumption. Instead, people drove more miles, bought larger vehicles and moved farther from work. The SUV – a monument to Jevons – exists because efficiency gains made gas temporarily cheaper (Hussain, 2024; Jevons, 1865; Klitgaard, 2022).

Work-Life: Productivity tools promised 4-day workweeks. Instead, we had expectations of 24/7 availability. Home offices eliminated commutes but erased boundaries between work and rest. Every hour saved becomes an hour that can be optimised.

Learning: Online education democratised access.

Result: Continuous upskilling expectations, certificate inflation and the anxiety of never knowing enough. Khan Academy’s vision of personalised learning became LinkedIn Learning’s optimisation treadmill.

The paradox reveals something deeper than economic behaviour – it exposes the ideological structure of growth capitalism. Under a system that demands perpetual expansion, efficiency gains cannot be allowed to reduce activity. They must stimulate more. This is not a bug; it is the core feature.

Indigenous economic systems offer an instructive contrast. Pacific Northwest potlatch ceremonies deliberately destroy surplus wealth to prevent accumulation. The Inca mit’a system rotated labour obligations to prevent any group from optimising its advantage. These were not primitive practices but sophisticated technologies for preventing the concentration dynamics that efficiency enables.

Most perversely, the Jevons Paradox (Klitgaard, 2022; Waligóra, 2019) applies to problem-solving itself. The more efficiently we can identify and address problems, the more problems we discover and create. Better diagnostic tools reveal more diseases. Better communication reveals more conflicts. Better measurement reveals more inequalities. We are caught in an ever-tightening spiral where each turn of the screw demands another (Hussain, 2024; Jevons, 1865).