Author’s Note | October 2026

I originally developed this essay in May 2022, while working on the publication of my book, The Weakest Link. My concern was that our increasing dependence on computing systems was creating something I described as cognitive debt: the gradual loss of the knowledge, skills, and independent capabilities that these systems replace.

The argument extended beyond individual users to organizations and society. As computing systems become increasingly integrated into everyday activities, they don’t simply change what we do; they change how we think and what we remain capable of doing without them.

I subsequently extended this argument to artificial intelligence in a May 2023 essay, Protecting Tomorrow from Today’s AI. With the growing adoption of generative AI, these concerns seem particularly relevant today.

The essay below reflects my thinking in 2022, with minor edits for clarity.

 

Cognitive Dependency on Computers

We often think of computers as tools for instrumental action with hardware and software elements. But computers are more than mere instruments. They are generative systems that, through the interactions between hardware, operating system, applications, designers, organizations, and users, lead to emergent outcomes that are more than the sum of these parts. 

Many of these outcomes expand what we do with our minds–such as sort through vast data, process volumes of information, and glean patterns from them—and even surpass our individual mental performance in speed and scope. Outsourcing these to computing systems, relieve our minds, allowing us to evolve in other areas. This has led to an upward cascade in human activity and with it the emergent complexity of modern computing systems. 

Take an individual application, accounting software for instance. The basis of accounting is the double-entry bookkeeping system, developed the merchants of Venice for keeping records of their trade, and documented by Benedetto Contrugli. It evolved into a formal system, with specific rules of capturing transactions on ledgers and journals that were adopted by businesses the world over. The process was tedious, requiring care and repeated entries. The earliest computing systems were built to speed this up. 

One of the first was the hand cranked predecessor to the modern calculator, the adding machine, invented in the 1880s, and a mainstay of accountants. Then came punch-cards, that helped standardize how numbers could be read by machines, followed by magnetic tapes, that increased how many numbers could be captured on mainframes. None of these changed the book-keeping approach. The ledger-based, double-entry system was still being used by certified accountants the world over. 

This began changing with the development of spreadsheets and software such as Peachtree and Quickbooks that worked on personal computers. They led to new standards and approaches for structuring data. Businesses could capture more data and track it easily. The ledger-based paper and pen approach moved from just book-keeping of sales into an integrated financial modeling, payroll payment, and supply chain management system. This created new process efficiencies that weren’t possible using the ledger-based system. Thanks to the system-wide integration of accounting, lean manufacturing approaches like Just-In-Time were possible, as were process improvement approaches like Six Sigma, which needed continuous steams of data. These had to be integrated across the industry, and with global supply chains, across the world. Computerized accounting systems were no longer just keeping trade records—they encompassed all commercial transactions.  

Today, it is impossible to run a large business without using an integrated accounting system. It has led to unique processes that cannot be construed or executed without it. The system has replaced not just what businesses do, but also how they think. It is a cognitive dependency on computing for business survival—and it is this system that Colonial’s hackers hold hostage.   

We have never studied our cognitive dependency on computing systems, which is positioned to get even deeper as AI-based systems, which already surpass our mental pattern recognition abilities, further their analytical capabilities. Our cognitive dependency on computing systems warrants further examination for four major reasons. 

1) It emerges out of a tool-first approach.

This makes our dependency on computing systems distinct from the sort of philosophical paradigms that Thomas Kuhn wrote about. Those aren’t tool dependent; these are. There isn’t a democratic process, but an engineered imperative based on technical and utilitarian considerations driving the developed of computing systems. Take something as simple as Microsoft’s PowerPoint—the dominant format that is emulated by all presentation software. No designer ever considered the impact that it has on the quality of decision making. While sources ranging from the Pentagon to the many board-roams have banned its use because of its negative impact on information presentation and deliberation, companies, even schools, the world over, continue to use PowerPoint (or some analogous software), for everything from instruction to decision making because of its availability and ease of use.  

2) It is driven by market considerations.

Because they have an organization or promotional entity behind them, the widespread adoption of computing systems is actively pursued with goals that aren’t egalitarian. There is a market dominance agenda driving their diffusion and the goal is to create dependancy. Take single sign-ons, where users can sign into new services using their Google or Facebook credentials. While these are globally promoted by these platforms for their convenience, in actuality, these offerings allow the platforms to track and mine user data across various services. They also make users more vulnerable to hacking by creating a single point of failure, where the loss of one of these passwords or a hack into one of these services could lead to a compromise across all of them.

3) Its effects permeate multiple systems.

Because computing system generatively create more systems, each emulates the original to foster easier adoption, leading to ever more of the same, but in zones of human activity we’d never expect. This spill over effect is seldom considered. Ever wonder what would have happened if the iPhone had removable batteries? Would Samsung and everyone who emulated them also have them? Would Tesla, whose founder has admitted to being inspired by Apple Inc., have them too? Keep in mind that GM’s EV1, the first ever electric vehicle to hit commercial production, had removable and user serviceable batteries.

Unsplash

 

4) It causes a cognitive debt.

As computing systems advance and our dependency on them increase, the cognitive distance between what it does and its physical analogs gets wider. The original computing system that was designed for instrumental efficiency gets layers of ideas baked upon it. No longer can the user discern or even relate to the original. Nor can they recreate what was done in the past, without using the system as a reference point. This is already happening with AI programs, whose analysis cannot be discerned without using other computer programs. In this loop, the system is essential. People cannot think without them, and in time, as generations are raised using them, won’t even know to think outside them. This price in our ability to think outside the system is paid forward. It is a Cognitive Debt that is incurred as we adopt computing systems without considering their generative systemic impacts. And, it is this debt that was paid by Colonial. 

 

Stopping this must happen at multiple levels. At the most immediate, we need to understand how users cognitively orient with computing technology and its components. Email, messaging, social media, and individual applications, all foster ways of orienting, rituals of actions, and beliefs and biases. These need to be examined at a system level to see where the gaps in cognition and action exist that are being exploited by hackers. A method for identifying the gaps is in my recent book, The Weakest Link: How to Diagnose, Detect, and Defend Users from Phishing. It helps organizations understand how users’ cognitive-behavioral factors interact with system level factors (computing design, work norms) and macro environmental factors (federal policy, breach notification laws) during technology encounters. Armed with this knowledge, IT departments can identify where the fissures in these surfaces exist, how hackers develop attacks that take advantage of them, and which users are most vulnerable to these exploits.

At another level, we need to study the outcomes of computing more comprehensively. We have been addressing individual tools in isolation, rather than their generative impacts on people, processes, and society. For instance, there is little research on how or why presentations made using PowerPoint hinder learning. Fixes are commensurately ad hoc. Many teachers routinely use presentation software for teaching, but entities from the Pentagon to CEOs like Jeff Bezos disallow its use. We need to understand the source of the problem. Is it the software, is it the approach to presentation, is it the system of discourse, or is it the decision that is not amenable to software-based presenting? Answering such questions requires us to take a systemic look at how individual software-driven processes influence the way we think about information presented during decision making, and how this has a downstream and upstream influence on commerce and society.

Finally, we need an approach for capturing business processes that are being replaced by computing so as not to incur a cognitive debt. Just like linguists chronicle endangered languages, we need to accumulate the knowledge from business processes, especially ones that are being replaced by new computing systems. Take the case of the Antikythera mechanism, found in a shipwreck off the coast of Greece. Built around 200 BC with a complex system of interlocking gears and dials, this mechanism accurately predicts the movement of celestial objects. But no one knows who built it or how. There is no Rosetta Stone for deciphering it. Its scientific basis was lost, as was the system that developed it. Other ideas, which perhaps appeared superior at the time, replaced it, obscuring both the science and the system that gave rise to this marvel that many consider the world’s first analog computer. It took 1500 years for science to develop another astronomical clock of its kind and another 500 years for us to create computers. This is the ultimate threat of computing system dependency. That in our zeal to adopt technology, we end up replacing great ideas merely because another one appears better at that point in time, and it takes a millennia to realize it.