Build Your Own Tools
In Richard Feynman's famous 1959 lecture "There's Plenty of Room at the Bottom" he posited that there was great innovation yet to be unlocked in miniaturization. In a time when humanity was reaching unprecedented scale in their engineering projects Feynman offered a $1,000 prize to anyone who could build an operating electric motor that fit in a space smaller than 1/64 of a cubic inch. He hoped this would spurn some innovation of physics and science. Instead, the prize was claimed by an electrical engineer named William McLellan who used such ordinary and pedestrian tools as toothpicks and a watchmaker's lathe to painstakingly construct the motor with traditional machinist techniques and ingenuity. No new processes needed to be invented, only the conscientious and careful application of machining techniques honed across centuries of iteration and invention.
Machining was the crucible of innovation through the steepest growth of the industrial revolution; it was the medium through which advancements in physics and engineering could be realized. The human capacity for innovation may be boundless but the speed limit of invention is that which can be built. It was the machinists who pushed that envelope, expanded the theoretical sum total of machines that could be invented. The milling machine - one of the most consequential manufacturing tools ever created - was reinvented multiple times within the span of just a few years in the Connecticut River Valley.
Therein lies the mechanism by which machinists innovate; they build their own tools. They make precise metallic parts and the tools that enable ever more precise parts are also... precise metallic parts. The feedback loop follows the shortest possible path. The traditional machinist's apprenticeship initiates a process of building progressively more complicated tools and learning how best to apply them. At its culmination this process yields a master machinist with an overflowing toolbox of both skills and handcrafted tools and fixtures, precisely crafted for the job at hand.
There is a modern analogue, a trade that has unlocked similar leaps and bounds of innovation, has manifested theory into reality, all through almost pathologically obsessive toolcrafting. The modern machinist is the software engineer. The inventions of the milling machine in the Connecticut River Valley of the early 1800s and the proliferation of front-end app frameworks in Silicon Valley in the 2010s has a deeper parallel than aesthetics.
Software engineers also engage in the deeply personal process of tool selection and building. From the earliest days of the emacs vs vim tribalism to ongoing debates about the latest trendy programming language, software engineers care deeply about the tools with which they ply their trade. Open source tools and frameworks supercharge the individual agency of software developers, with any sufficiently motivated person empowered to wield the tools to build and innovate.
Likewise the industry at large iteratively builds and rebuilds their tools. Keeping up with the latest JavaScript framework or infrastructure best practices can feel extraneous and like a laborious churn. Yet within these trends is something undeniably powerful; the constant self-reflective reinvention of the field. The compounding benefits of this practice are so powerful that they've been integrated into the expectation of exponential progress of software. And the software industry has delivered on this growth curve over the decades, an amazing track record continually reinvigorated by the next powerful tool or abstraction.
Not all fields are positioned to build their own tools. It's an unfortunately rare treat, enabled by cultural norms as much as technological convenience. Any industry whose tools are composed of the same substrate as their output is well-positioned; the precise metal of machinists and self-assembling code are both common between what they make and how they make it. Time invested in building either process or production returns dividends to the other. The skills and tools are already readily at-hand to improve a process, putting that agency into the hands of every member of the field.
Cultural expectations are heavily influential as well. Software collaborates on their infrastructure and tools through the mechanism of open-source, a practice so normalized in the industry that it is at risk of being taken for granted. Machinists implemented collaboration through offline mechanisms; the Machinery's Handbook, the oral tradition of folk engineering omnipresent in the apprenticeship system, and even standardized interfaces like fixture and tool mounting systems. Each culture rewards innovators with social capital. The most lauded software engineers build infrastructure, the most respected machinists build the most ingenuous tools. Around either the break room lunch table or forums and GitHub Issues you can see respect being paid according to this ethos.
Industries that don't build their tools operate on an entirely different time constant. The state-of-the-art is speed limited by a feedback loop that includes symbiotic but distinct entities; tooling providers, industry consortiums, or standards bodies. The additional layers of deliberation and feedback slow the rate of innovation but the more consequential limitation is the compartmentalization of concerns. Those who build the products are most equipped to understand the process, to find flaws and friction, and to imagine a better way. Yet they're often the furthest from the arbitrators of process and best practice. Exploration is by definition a lossy process; many dead ends must be traversed and false assumptions disproven before a new abstraction or tool can be integrated. A fast feedback loop is the only mechanism to cut through the chaff and eliminate unpromising ideas. If that loop contains more than one entity the time constant will slow by orders of magnitude.
There's an opportunity for legacy industries to fix the feedback loop; they need to build their own tools. While internal toolmaking has historically only been available to a few fortunate industries with the alignments outlined above the application of practical engineering has shifted subtly and the opportunity for innovation is more democratized than ever.
Early in my career (in the late 2010s) I worked as an engine calibrator. The role required running hours or days of dynamometer testing, yielding mountains of data that had to be processed and distilled into calibration tables to characterize the machine. I was tasked with a particularly thorny calibration project and sent to the senior engineer who had last executed the process on a prior program. Happy to help, he dusted off a three-ring binder as thick as a phone book. He had printed out over 1400 individual time-series plots and spent days painstakingly selecting the correct signals with a pen. I was dumbfounded at the labor required! Even though the department had a tools and methodology group I realized that my own job was as much data scientist as calibrator. My implementation was a brief Matlab script that ran some signal analysis and generated a cal table in minutes. For years afterward, at each new job one of my personal onboarding tasks was creating a home for my tools: a new blank Microsoft Excel add-in, a repository for Python scripts, or whatever framework I was working with. The software skills required were somewhat hard-fought for a mechanical engineer with little coding practice but the benefits compounded and I fell down the software rabbit hole. Now younger generations of engineers have ever more coding exposure upon which to draw.
The most consequential engineering tools in every industry are all software. The digital realm precedes even heavy manufacturing with simulation, digital twins, and modeling. Even legacy parts of the process - like designing a physical part - now must exist in a digital realm for collaboration and alignment. And the entire engineering process must at least be representable in a software layer. The next great physical innovations will necessarily be enabled by a software layer. It's all digital now but simultaneously the power to make software has never been more universally available. Any technical person - whether they're a chemical engineer or a manufacturing technician - has had some exposure to software and code. The education and knowledge to expand that capability and make whatever arbitrary software deemed necessary is also available. This is not quite the seamless synergy of the golden age of machining or the realm of software engineering but the alignment is close enough for practical purposes.
Whether as an individual, an organization, or an industry the next innovation unlock is probably toolmaking. It's terrific if you can make tools that look a lot like your products but most likely your tools will be software and you'll have to learn some new tricks. Developing software skills will require an investment of time and effort but there's nothing with higher returns; making your tools restores sovereignty to your abilities. To discover what tools to craft be reflective about your process. Be self-critical about your day-to-day work. Be lazy! Abhor inefficiencies and repetition. Consider the end-to-end processes, at the organizational or industry level. And most importantly; share your tools. Build a collaborative culture of introspection and communal ambition towards better methodology. Just build your own tools.