You Don't Automate the Levers of the Past

What scale, automation, and creative destruction actually do to a business

1. Automation as Systemic Transformation

Elevator Logic

Automation demands transforming the entire workflow. The historical example of the elevator operator makes it obvious. When the profession disappeared, nobody built a robot to sit in the chair and move the manual lever, and nobody built a voice recognition system so passengers could tell a machine which floor they wanted. Both would have been expensive ways to preserve the inefficiency of the old system instead of replacing it. The lever became a button, the manual circuit became an integrated electronic system, and the components changed because the system changed. The lesson generalizes. An automated system can’t stay the same and just become “automatic.” It must evolve, or the industry’s progress will destroy it. You don’t automate the levers of the past, you redesign the system so those levers are obsolete. The workflow, the components, and the interface all change together, or the attempt at automation is just a dressed-up version of the old process running on more expensive hardware.

What the elevator pattern really shows

The loudest debates around automation treat it as a self-contained phenomenon, as if a single technology causes a single class of job loss. The reality is wider and older. People who used to make candles lost their livelihood to the spread of electric light. Telephone operators who spent their days connecting cables for callers were replaced by an electronic switching system that worked around the clock without a salary Bell Labs deployed the first electronic switching exchange, the 1ESS, in Succasunna, New Jersey in 1965. Did these people just disappear? No. Some suffered through the awkward years of unemployment. Most of them adapted, willing or not, because the world around them kept pushing forward and they had to keep up with it. That pressure is the general pressure of creative destruction, not a side effect of automation alone. Joseph Schumpeter used the term in Capitalism, Socialism and Democracy (1942) to describe how capitalism constantly displaces the old with the new. The old gives way to the new, not because anyone planned it that way, but because evolution doesn’t consult the people it displaces.

Automation is one of the more visible faces of this process, not the process itself. A profession disappears because the underlying activity stops being demanded, and a machine often shows up simply because it’s the cheaper way to serve whatever demand is left. Framing job loss as “caused by automation” misses the structural driver and tends to produce bad policy and bad business decisions built on a misdiagnosis.

2. The Trap of Premature Capital

Capital needs a critical distinction. VC money isn’t inherently “bad money.” But VC money used too early, before a company has proven its customer demand, enables a dangerous detachment from reality. That premature injection of capital encourages management to scale an inefficient business model. The real threat is scaling an inefficient system, not automating it. When you scale a system loaded with sync tax, the cost of running it doesn’t grow linearly, it grows exponentially, and the whole operation bloats as a result. It demands more meetings, more coordinators, more human glue, more tooling just to keep the pieces from falling on top of each other.

The talent side of this trap is brutal. The company recruits skilled professionals with tempting offers to keep the scaling machine fed. Those professionals walk in expecting to do meaningful work and quickly discover that most of their day goes to meetings, status updates, and hand-offs between teams that can’t talk to each other on their own. The job experience gets devalued. The talent stagnates. The company pays a fortune for it, both in infrastructure and in the silent cost of people who are slowly checking out while still drawing a salary. The Sync Tax doesn’t just eat profit. It eats the professional lives of the employees.

When the “just leave” answer fails

A common objection at this point is whether automation “frees” the humans trapped in such a system. In practice, that question misses the structural constraint. People are not slaves. Nobody is legally forced to keep showing up to a job that is slowly hollowing them out. The freedom to leave exists in principle, and in many labor markets it exists in practice too. The real damage is the loss of a meaningful job experience, not the loss of liberty. Sync tax devalues what people do all day, and the people who stay either accept the slow decay or suffer through it as if they had no choice.

The caveat matters, though, because the freedom to leave isn’t equal everywhere. In countries with strong protections against firing, the same laws that make it hard to let someone go also make it harder to get hired somewhere else. The cost of changing jobs rises, the duration of unemployment stretches, and the practical freedom to walk away from a toxic workplace shrinks. So the question of “why don’t they just leave” deserves an honest answer. In some regulatory environments, they can’t leave easily, and pretending otherwise ignores the structural cost those policies impose on the very workers they are meant to protect. Either way, the right move is to remove the sync tax that turned the job into a devalued experience in the first place, not to make staying in a devalued job more comfortable.

3. What Humans Still Have to Do

As deterministic data processing and repetitive mechanical tasks get delegated to machines, the essence of human work concentrates into two high-value jobs that, in their nature, have to stay human: creativity and leadership.

• Creativity: the ability to invent the machines and design the interfaces (or anything else) that move the world forward.
• Leadership: the ability to set the vision, inspire, and guide the transition from learning what works to executing it at scale.

Repetitive and mechanical work is still needed. The work was always meant for delegation, not destruction. Freeing that contribution up is the whole point.

What machines can and can’t do

In practice, machines don’t lead humans. They can coordinate, schedule, route, and even simulate authority, but the act of leading other humans through uncertainty, conviction, and responsibility isn’t a process you can encode. A machine follows its objective function, a human can change the objective function, and that distinction is the entire difference between management and administration.

The creativity claim needs more care because AI today sounds creative to many listeners. It can produce a poem, a logo, a strategy memo, a working piece of code, and any of those outputs can feel new to the person reading them. The output, though, is recombined from training data that already existed. The machine didn’t invent anything in the strict sense. A human looked at the world, noticed something nobody had noticed, and wrote a paper, painted a canvas, designed a product, or started a company that didn’t exist the day before. Only humans invent the way humans invent. AI emulates the surface of that process well enough to fool a lot of people, sometimes for a long time, but it’s still working from something that already exists. Anyone building a business on the assumption that the machine is the source of novelty is building on borrowed ground.

When the role itself is the problem

A company that successfully removes sync tax will discover, often with surprise, that a meaningful slice of its workforce was only there to manage the meetings, the handoffs, and the manual reconciliation work the new system no longer requires. The instinct in some leadership circles is to keep those people employed out of a vague sense of fairness. That instinct is wrong, and the reason is economic. There is no demand for the coordination work those roles were doing, because the new system does it for them. The whole point of The Sync Tax and The Architecture of Independence is that you don’t pay for things nobody needs. Keeping a useless role alive to spare someone’s feelings costs the company money it could spend satisfying actual demand. This equals less profit, which equals jeopardy. It also traps the person in a job that is being hollowed out from the inside, which is exactly the devalued experience the redesign was supposed to end.

The honest version of the message is simple. When a role is no longer demanded, the right move is to let it go. The person in that role gets freed to find work that is actually valued, even if it’s unpleasant and uncomfortable for some time, and the company gets freed to pay for work that customers are actually paying for. Demand rules, and that pattern repeats across product decisions, hiring, and org design, and actually everything related to humans. The job of a leader in a post-sync-tax organization is to make those calls quickly and humanely, not to pretend the work was never the problem.

Middle management is the structural layer that makes creativity and leadership possible at scale, not a layer of waste. The only honest distinction to hold onto is between middle managers who do real coordination work and the layer of coordinators who only existed because the underlying system couldn’t coordinate itself. The first one is demanded, and the second one is the one the previous paragraph is about.

Who decides what to build next

The same logic applies to who decides what the company should do next. A healthy company hires expert professionals and asks them what to do, not the other way around. If a CEO dictates every product project from above, the best product manager in the world has a single rational move, which is to quit that toxic workplace and find a company that will actually use his brain.

Any entrepreneurial venture starts in learning mode. The team investigates, tests, adapts, and chases the right problem to solve (the JTBD, the Jobs to be Done framework from Clayton Christensen’s The Innovator’s Solution) and the right solution that customers will actually pay for. That phase is emergent strategy, because nobody up front knows the answer. Once the demand is proven and the solution is working, the strategy becomes deliberate. The company now knows what to do, and execution takes over from learning. That is the natural transition.

The CEO also has a parallel duty, and it’s the one most often missed. He has to keep scanning for changes in the market that the deliberate strategy is no longer catching. As companies grow, the CEO loses direct contact with demand. The product department owns that scanning work, and the tech department owns the implementation. In a small company the same person can do all of it, but the moment the organization scales, the role splits. The product manager is looking at the present, at what customers are asking for right now and what the next quarter’s roadmap should look like. The CEO is looking at the future, at where the market is moving, which adjacent demand is about to open up, and which part of the current business model is going to be obsolete in a couple of years. Both jobs are real. Neither replaces the other, and the practical question of how a company should actually structure that split’s the kind of thing that gets decided badly in many growing companies, often by accident rather than by design. The principle, though, comes back to the one introduced in The Sync Tax. Demand rules, and the people closest to demand should be the ones deciding what to build.

For anyone who hears all of this and fears for their own job, the reality of creative destruction is simple. At least in the short term, AI won’t replace you, someone who uses AI will replace you (a saying that gained traction in early 2023, with no single clear origin). The replacement is a verdict on the value of doing the work the old way when a new way exists, not a verdict on you as a person.

4. How Modular Systems Take Over

Modularity is a pattern that shows up everywhere once you look for it, and it doesn’t belong to any company or any framework. Nature doesn’t evolve by making a single organism live forever. It evolves through newborns, each one more efficient and more resistant to threats than the last, while the old generation dies off. A company that wants to survive has to follow the same pattern in its own structure. A biblical metaphor captures the same point. You can’t put new wine in old wineskins Matthew 9:17; Mark 2:22; Luke 5:37-38. The new wine ferments, expands, and bursts the old container. Trying to force a new, agile business model into an integrated, interdependent architecture destroys both. A company that wants to live has to be willing to kill its interdependent business model, not just the software monolith, and replace it with a modular, efficient one.

As Clayton Christensen described in his work on disruptive innovation and modular architecture The Innovator’s Dilemma, 1997, integrated systems are optimized for performance during the early phase, when the components are still small and being tuned to deliver a working product. Modular systems are optimized for efficiency during the sustaining phase, when the priority shifts from making it work to making it cheap to maintain and evolve. Old machines are replaced by new ones because keeping the old ones around costs more than the value they still produce. The same logic shows up in genetics, in industries, and in software, because the underlying constraint is the same in each case.

5. Deregulation, with Enforcement

The C-suite often accepts high complexity costs as “normal” because they lack technical understanding or because the leadership below them is too submissive to push back. The answer is to give the system more freedom and more accountability at the same time. The historical precedent is clear. The economies that flourished after periods of shock therapy the term associated with Jeffrey Sachs and the Balcerowicz Plan in Poland (1990) did so through aggressive deregulation of the private sector, paired with strict enforcement of the law against abuse, and both halves were required. Deregulation without enforcement is just chaos with a marketing slogan, while enforcement without deregulation produces a system so over-regulated that nothing can move. The countries that got it right gave businesses room to operate, and then made sure that anyone who exploited that room faced serious consequences.

Applied to a company, the same logic holds. Independent modules get the freedom to innovate on their own side of the boundary, and the agreement between them (the interface, with its rules of who owes what to whom and in what format) is enforced strictly enough that nobody has an excuse for breaching it. If two departments or two modules need constant meetings to coordinate, the diagnosis points to a bad interface, not a need for more meetings. The time those humans are spending in meetings should be invested in redesigning the interface until the meetings are no longer needed. Constant meetings between two modules are a signal of over-regulation inside the company, a sign that the architecture is forcing humans to do the work that an interface should be doing for free. Once the interface is fixed, the internal economy deregulates itself, and the company’s ability to actually ship and adapt comes back.

The same principle explains the cost of C-suite blindness in a way the balance sheet never quite captures. A CEO who has learned how to raise large rounds and pick his CTOs will keep operating as long as the runway lasts, even when the technical debt is compounding. If the CTOs he picks share his blind spots, or simply lack the standing to disagree with him, nobody in the building has both the information and the authority to bring the consequences to light. The fix isn’t a louder dashboard but giving the people closest to the technical impact the freedom and the obligation to challenge the high-level decisions that are creating the cost. The CTO stopped going to the sync meeting he was never supposed to need.

References

1. Schumpeter, J. (1942). Capitalism, Socialism and Democracy. Harper & Brothers. Wikipedia
2. Bell Labs. (1965). First Electronic Switching System (1ESS) deployed in Succasunna, NJ. Wikipedia
3. Christensen, C. (2003). The Innovator’s Solution. Harvard Business Review Press. Wikipedia
4. Christensen, C. (1997). The Innovator’s Dilemma. Harvard Business Review Press. Wikipedia
5. The Holy Bible. Matthew 9:17; Mark 2:22; Luke 5:37-38 (NIV). Bible Gateway
6. Shock therapy economics. Jeffrey Sachs and the Balcerowicz Plan, Poland (1990). Wikipedia