We love a good efficiency drive. Faced with mounting pressure, spiraling costs, or a team that seems to be spinning its wheels, the instinct is almost universal: standardise, streamline, automate. Set the process, adopt the tools, get the outcome.
That logic works brilliantly in factories. On an assembly line, the variables are known, the tasks are repeatable, and removing friction genuinely speeds things up. But for decades now, organisations have been applying the same industrial playbook to a fundamentally different kind of work, and the results are quietly disastrous.
The numbers are alarming. Around 75% of employees report battling what researchers now call tool fatigue, the cognitive exhaustion that sets in when company after company layers new AI agents on top of nine or more existing apps. The result is not greater efficiency, it is a 65% increase in coordination work between team members. According to Asana's latest Anatomy of Work data, the average worker is still losing nearly five hours per week just searching for the right tools and data: the exact duplicated effort that AI was supposed to eliminate. We built the machine to reduce the load, and somehow the load got heavier.
We keep doing this. And nowhere is the pattern more visible, or more costly, than in urban planning.
The Problem with Planning's Efficiency Obsession
Today, conversations in planning circles are dominated by familiar themes: standardisation of tools and processes, digital workflows, and the adoption of AI to automate parts of the planning process. These are presented as solutions to an obvious crisis. The planning system is critically overburdened and under-resourced, forcing practitioners to meet escalating demands within shrinking timeframes. This pressure has triggered a retention crisis, with experienced planners leaving the profession in record numbers.
But what if the assumption is wrong?
There is a striking body of evidence that automation and process standardisation, when applied to knowledge workers rather than factory workers, do not reduce cognitive load. They increase it. And there is a rich historical record explaining exactly why.
Where This All Started: The Factory Mind Meets the Knowledge Worker
Knowledge work, the kind of work that produces judgment, strategy, and decisions rather than physical goods, has always been awkwardly fitted into frameworks built for manufacturing. When knowledge-intensive occupations expanded dramatically in the early twentieth century, the dominant lens for understanding productive work was still Frederick Taylor's scientific management: decompose tasks into measurable components, optimise for efficiency, supervise against output targets. It was a framework built for bodies on production lines. Applied to minds at desks, it was a persistent mismatch.
The gap became critical when computing arrived. To integrate digital tools into knowledge work, you needed some account of what knowledge workers actually did with their minds. The account that was available, and that became enormously influential, came from cognitive science, specifically from a programme of research launched in the mid-1950s. Allen Newell and Herbert Simon developed programs designed to simulate human problem-solving through symbolic manipulation. Their central claim was compelling. Intelligent behaviour, including complex reasoning, could be understood as a form of information processing, amenable to formal description and therefore to computational replication.
This idea did not stay in the lab. By the 1980s, it had translated into expert systems, software that encoded specialist reasoning into rule-based structures, and it quietly became the operating assumption behind nearly every digital tool built for knowledge workers since. If thinking is information processing, then the goal of technology is to make that processing faster and more efficient. The planner's judgment becomes an input to be structured, not a capacity to be extended.
It sounds logical. It sounds manageable. It is not.
The Reality: Streamlining Adds Cognitive Load, It Does not Reduce It
The information-processing model of mind is not wrong, it is incomplete. And its incompleteness has consequences that show up directly in the data on worker wellbeing and productivity.
Cognitive offloading, the idea that handing analytical tasks to computers frees up human mental capacity, turns out to be far more complicated than the theory suggests. There are biological constraints on how your mind processes information, weighs it, and reconnects it to the broader goals of decision-making. Creativity, judgment, the recognition of what actually matters, the capacity to hold incommensurable values in productive tension, these are not tasks you can hand off to a system. They are not decomposable into rules. They are, in fact, what knowledge work is.
In planning, the underlying assumption has always been that strategic planning is fundamentally an analytical problem, and that better tools for executing analysis would produce better decisions. Even those who criticize the conception of strategic planning as an analytical task have never explained or explored what the non-analytical looks like.
The Paradox of Choice in Modern Work
The same pattern plays out in every sector. Stack enough tools on a knowledge worker, impose enough standardised processes, and you do not free up their thinking, you colonise it. Every new system demands orientation, maintenance, context-switching, and a constant low-level negotiation between what the tool is designed to do and what the situation actually requires.
The result is what researchers have started to call digital fatigue, phenomena that look, superficially, like complaints about too many apps. But they are pointing at something deeper; a fundamental mismatch between how knowledge work actually functions and how the systems designed to support it are built. When a worker loses five hours a week not to doing their job but to navigating the infrastructure built to help them do their job, something has gone badly wrong at a conceptual level.
The productivity measures we inherited from factory management can not see this failure. They measure how fast the computer did the analysis, not whether the analyst was able to think. They measure process compliance, not judgment quality. They confirm their own assumptions, and so the assumptions go unchallenged.
Is There an Alternative?
Yes. But it requires rethinking the question entirely and that is what we are working on in Seal on the Beach.
The dominant model asks how do we make knowledge workers execute defined processes more efficiently? The alternative asks how do we build tools that extend and support the cognitive capacities that knowledge work actually demands?
The difference matters enormously. Cognitive offloading, handing tasks to systems, is one kind of tool. Cognitive scaffolding, creating conditions in which a person can think better, more creatively, more systemically, is something else entirely. A scaffold does not do the work for you. It creates the structural conditions under which you can do the work you could not otherwise do as well.
What would this look like in practice? It means designing tools around personal productivity, which, in knowledge work, is inseparable from agency, autonomy, and job satisfaction. It means recognising that organisational productivity in knowledge-intensive fields is not achieved by standardising individual thinking but by supporting it. It means tools that adapt to the specific needs and cognitive styles of the decision-maker, rather than forcing the decision-maker to adapt to the tool.
The price of continuing to apply industrial logic to knowledge work is not just wasted money on software subscriptions. It is planners, and strategists, analysts, researchers, and all the other professionals whose primary output is judgment, leaving their fields, reporting unhappiness, struggling with tools that were supposed to help them but somehow make everything harder.
Streamlining will cost you more in the long run. Not because efficiency is a bad goal, but because the tools built around a factory model of the mind are not efficient for knowledge workers, they are inefficient in a way that the metrics we use can not see. The problem is not the workload. The problem is that we keep building the wrong kind of help.
What would this look like in practice? It means designing tools around personal productivity, which, in knowledge work, is inseparable from agency, autonomy, and job satisfaction. It means recognising that organisational productivity in knowledge-intensive fields is not achieved by standardising individual thinking but by supporting it. It means tools that adapt to the specific needs and cognitive styles of the decision-maker, rather than forcing the decision-maker to adapt to the tool.
