There was a period in my career when arriving at work on a Monday felt like arriving at a frontier. Not metaphorically — literally. The problem you were trying to solve had never been solved before. The approach you were taking had not been approved by a committee, because no committee yet had the vocabulary to evaluate it. The software you were writing would, if it worked, do something that had not been done before. And it worked, or it did not, and either way you learned something irreplaceable.
I am thinking of two periods in particular. The first: Marconi's Underwater Systems Division in Wembley, building the most advanced torpedo simulation platform in the world — real-time software at the absolute frontier of what was technically possible, running on hardware that had no margin for error, in a Faraday cage that literally had to be sealed against the outside world. I had come straight from university. I had no idea that work could feel like that. The second: the years that followed at Cray Systems, building command and control systems for ESA — for spacecraft that would, once launched, travel beyond the range of any intervention. Software that had to work, without exception, across an operational lifespan measured in decades. The technical demands were extraordinary, the constraints unforgiving, and the work was, genuinely, some of the most joyful of my professional life. Marconi took me to the frontier of real-time computing. ESA showed me what a frontier looked like when the stakes were measured in decades and the distance between you and your software was 150 million kilometres.
What it Actually Felt Like
The NHS National Programme for IT — for all the political turbulence that surrounded it — had a similar quality in its early years. We were building, for the first time, a single digital backbone for the entire UK health service. The Spine. Every hospital, every GP surgery, every pharmacy, connected. The scale was unprecedented. The technical challenges were genuine and unresolved. And there was a shared sense, across the team, that what we were doing mattered.
I remember a specific afternoon in Darmstadt, 1989, when the team confirmed that the Ulysses mission control software had handled a real-time command sequence correctly for the first time. The room was not large. The celebration was not elaborate. But the feeling was of having done something that had not previously existed in the world. That feeling is difficult to replicate in a governance review.
The best days were not the days when everything went smoothly. They were the days when a hard problem finally gave way — when the thing you had been trying to build for weeks suddenly worked, and you understood, for the first time, that it was going to be all right.
When the Balance Started to Shift
The shift did not happen suddenly. Nobody announced that the age of building things was over and the age of governing them had begun. It happened gradually, through the accumulation of entirely reasonable additions to the way organisations operate. A risk framework here. A compliance requirement there. An information security standard that, individually, made complete sense and, collectively, began to consume the time and attention of the people who were supposed to be building.
DORA arrived. The EU AI Act arrived. BaFin's guidance on operational resilience arrived. Each regulation is defensible — each addresses real risks that real institutions have failed to manage. But the cumulative effect on the working life of a senior technologist is that an increasing proportion of the job is now documentation, governance, audit evidence and committee attendance. And a decreasing proportion is building.
The proportion of the working week spent in meetings whose primary output is the preparation for other meetings has, in my observation, approximately doubled over the last twenty years. The proportion spent actually making technical decisions has halved.
What Was Actually Lost
I want to be careful here, because this is not a straightforward complaint. The governance frameworks exist because organisations failed. The compliance requirements exist because people were hurt by the absence of them. The security standards exist because breaches happened that should not have. The world is more complicated and more consequential than it was when the work felt most joyful, and the governance that now surrounds it reflects that reality.
But something was lost in the transition — something that is hard to name precisely and easy to undervalue. A relationship to the work itself. An immediacy of consequence. The feeling that what you were doing today was connected, directly and tangibly, to something that mattered.
The young engineers I work with now are technically capable, often extraordinarily so. But many of them have spent their entire careers in an environment where the governance layer is already fully established — where the interesting technical decisions happen within a heavily constrained space, and the work of navigating the constraints is as demanding as the work of solving the problem. Some of them have never experienced the specific feeling of being at a genuine frontier. I find myself hoping they eventually will.
The good times did not go away. They became harder to find, more constrained, more surrounded by necessary complexity. The frontier still exists — in AI, in quantum, in the problems that the next generation of technology will create and have to solve. But getting to the frontier now requires navigating a great deal of infrastructure that was not there before.
That is probably the right trade-off. It does not stop you missing what it felt like when the distance to the frontier was shorter.