Weekly digital disruptions have become a familiar feature of contemporary life. Flight operations in the United Kingdom face sudden paralysis multiple times within a span of a few months. In an even more bizarre incident, supermarket freezers across thousands of locations are remotely hacked and compromised. When these failures occur, the disruptions are severe, yet a consistent pattern follows. Almost nobody is held accountable. There is rarely any official identification of the culprit who caused the breakdown. Systems eventually return to normal, only for the cycle to repeat with another unexpected crisis.
This systemic vulnerability was starkly illustrated by the widespread global information technology outages involving the Windows operating system. In that instance, public attention quickly fastened onto Microsoft. Facing intense scrutiny and severe reputational damage, the corporation needed to deflect the negative press. An investigation was initiated, and a culprit was formally announced. The blame fell upon a little-known firm called CrowdStrike, an entity that the vast majority of the public had never heard of prior to the incident. While some observers welcomed this development as a sign that accountability could be enforced, analysts argue that the gesture served primarily to shield Microsoft from ultimate responsibility.
Stepping back from these high-profile episodes reveals a deeper and more troubling inquiry. How did society arrive at a digital world that is simultaneously fragile, opaque, unaccountable, and yet deeply sophisticated? The roots of this modern crisis do not lie in technological limitations. Instead, the answer is found within the mechanics of capitalist economics.
To understand this phenomenon, it helps to examine a traditional physical manufacturing process, such as making clay tiles. Producing a tile requires at least four distinct stages: preparing the raw clay, shaping the material, drying the product, and firing it in a kiln. When a production process is broken down into these discrete, well-defined stages, each phase can be optimized, streamlined, and cheapened. Producers can introduce specialized tools, machinery, and technoscientific advancements tailored to each specific step. This may involve procuring a dedicated shaping tool, inventing an internal mechanism, or designing a better way to transfer the semi-finished product between stages.
Dividing a production process into separate components makes it easy to analyze the exact level of skill, labor, and effort required for each phase. This optimization yields efficiency-oriented cost reductions. In economic theory, this concept is known as the Babbage principle, named after the nineteenth-century British scholar Charles Babbage, whom history remembers as the father of computers.
Applying this same economic logic to the digital realm reveals how modern software development operates. Consider a digital process like building an image-captioning artificial intelligence. Such a system requires identifying objects within an image, labeling those objects accurately, and then synthesizing those labels into coherent sentences using text generation technology.
A shrewd, profit-driven business operator will scan the web for existing tools corresponding to each phase of this process. They search for third-party object identification tools, external labeling services, and ready-made text generation utilities. Often, they find web services that perform these functions at minimal cost and can be accessed programmatically in real time via application programming interfaces.
While applying the Babbage principle to physical tile making requires identifying technology vendors, negotiating terms, and engaging in formal contracts that consume considerable time, the digital equivalent happens differently. In the digital world, a designer can reconfigure a complex process with just a few clicks within minutes. The software pipeline is rapidly altered to rely on external web services, immediately reducing production costs and boosting profitability.
Yet, this relentless drive toward cost reduction triggers a profound structural transformation. Once a digital process is reconfigured this way, it is no longer under the complete, autonomous control of its creator. Instead, it becomes fundamentally dependent on an array of external web services accessed across the network. If any single one of those external services fails, lags, or behaves erratically, the entire digital process can fail or produce bizarre errors.
The cheapened digital process is, by its very nature, a fragile digital process. It contains multiple points of vulnerability and numerous distinct categories of potential failure. When a crash occurs, it is rarely obvious which specific component broke down. The errors may not even be reproducible, because the failing web service might resume normal operations in the next minute or even the next second.
This environment is not merely fragile; it is structurally opaque and inherently unaccountable. Every single opportunity to cheapen a production process acts as an incentive to create a new digital link. These programmatic calls and API integrations multiply the layers between the software and its execution. With the creation of each new link, the fragility, opacity, and lack of accountability go up another notch.
Society often takes pride in living within an hyper-connected digital ecosystem where everything interacts seamlessly. Yet this is simultaneously a world where underlying failures cannot be deciphered or traced to a responsible human actor. Observers define this reality as digital network capitalism. The underlying crisis is not that engineers lack the capability to build robust and reliable digital systems. Rather, the social relations of capital demand constant cost reduction, and those economic imperatives ensure that reliability is consistently sacrificed for profitability.
[Deepak P, is Associate Professor of AI at Queen’s University Belfast and adjunct faculty member at IIT Madras]