On technical knowledge

Over my 8 years in California, I’ve met many immigrants who, like me, found a way in the US thanks to a “generalist education.” Being a generalist (or at least multidisciplinary) in an increasingly specialized workforce is an incredible edge. That same dynamic (combined with a lack of courage) led to general contractors billing $150/h (rightfully).

In high school, I learned technical drawing on paper, doing manual projections then a few simple shapes in SolidWorks on crappy classroom computer (windows XP memories anyone?). I was told: do your best with little, you’ll do great with better when you’ll be working a “real” job. The current generation jumps straight into professional software because that’s what employers are looking for based on (bogus) job descriptions you can find on LinkedIn. I didn’t have a laptop. I was 16. Constraint breeds something. I didn’t end up using that specific skill but I definitely believe it shapes the web designer I am today.

Everyone knows not to expedite or skip the warm-up. There is a lot to it, even (especially) intellectually. It’s what we call trusting the process. It’s what the seniors mean when they say “Easy! How did you get that?”. They can smell shortcuts. They have the scars of having tried them themselves. They honed their craft for years. Corporate culture is centered around shortcuts. MVP, minimum effective dose, scrappy, bias toward action… it has many names.

The “Move fast break things” ethos bleeding across industries is scary. If a server crashes, nobody dies but when planes or rockets explode, it’s a different story. Engineering as a discipline is crossing a critical milestone. The seniors are retiring or (rightfully) milking the corporations as consultants. It is felt across many industries. I believe that Sustainability as a goal in the corporate world is largely the result of the “tech debt” induced by the lack of senior mentorship. Design systems in the design/tech space are a great example of an attempt at reducing debt and risk due to turnover. Today, AI promises to store all humane  knowledge and make it available and digestible to newcomers. Can it?

I see a parallel with Zen philosophy where the concepts are relatively simple but can’t replace first-hand experience. Enlightenment is a simple idea, sitting for hours, for years of doing cryptic rituals is something else. Experience requires time and effort that few are willing to put in… or foster. Education as an institution is mixed on how to solve this, but it seems to be aware of the loss of intellectual autonomy of each generation. AI is the next leap, similar to the transition from the slide rule (hand-operated mechanical calculator) to electronic calculators in the 70s. No organization wants to slow down progress. At the same time, knowledge about the different levels of abstraction of each technology gets lost along the way. I don’t know how to use a sliding scale. I don't even know half of the functions of a scientific calculator. Each of us knows less and less about the stuff we depend on. Some of it is normal, some is imprudent. We are witnessing an example of this phenomenon as America is trying to bring home chip manufacturing. Can we do it? Should we do it? 

Engineering is a fascinating window into a culture. I love all the stories about Japanese software development. They are so wild. Often about low-level programming languages like assembly, involving massive teams, at large scale, with high standards for test and quality. No tolerance for failure. Suicide rates and insane mega corporations are only a tiny subset of the many facets of Japanese culture.

The US is the complete opposite. With massive capital, it can afford failure and has embraced it (It paid good dividends). Like yin and yang, both approaches make sense. The existence of one justifies the other. Nevertheless one relies a lot more on material resources while the other on human capital. The economic pressure for innovation to fuel growth makes me favor autonomy and human capital.

Another example: The Apollo 11 Guidance Computer (AGC) source code for the command and lunar modules was 3.47 MB got man on the moon. All were coded in assembly written in 1969. Today apps are thousands of times heavier than that. I’d argue that few have benefited humanity anywhere near on a Byte:Byte ratio.

The counterargument to my techno-skepticism is simple and strong: incremental gain requires optimization. Something needs to be sped up, abstracted, or both. We can’t rely on mere knowledge accumulation at the individual level. One cannot know everything there is to know about electrical systems from the chemical composition of transistors all the way to the latest data transfer protocols (Though some freak out there are pretty close). Personally, I’m equally scared by catastrophic (hypothetical) scenarios as I’m appreciative of the comfort technology provides. So I (delusionally) defuse my fears with podcasts and blogs hoping to vaguely know “how stuff works”. On the flip side, higher education keeps getting less and less accessible. Because proper transmission of knowledge takes time and resources. 2 things money can buy. For how long?

← Index / Published on 2025-01-12