I just finished reading this great article on what goes in to making semiconductors, and what goes in to constructing the buildings where they are made.
I was struck with how amazing modern semiconductors are, how amazing the process is to build them, and how amazing the buildings are. These buildings need freight elevators that can lift multi-ton, very expensive machines. If the building must be in an area with seismic activity, the very soil the building is on can be isolated from neighboring ground. The volume of gases needed requires a plant on-site that splits air apart. When you hear that semiconductor design today requires dealing with individual atoms, you probably think, "wow, that's pretty impressive, I'm sure there are some smart scientists that have worked a long time to figure stuff like that out." On the other hand, your intuition, or your knowledge of "what is hard", probably doesn't have a very good basis for how difficult building enormous freight elevators is, or how much research went into isolating a building from seismic activity. It's easy to overlook or hand-wave those details away, but this article, by walking through the detail, really does a great job at shining light on those things, and making the reader appreciate how difficult some of these problems are.
I was also struck by the evolution of the process over what is, by modern standards, a long period of time. The modern semiconductor facility didn't just pop up. The article points out that back in the 80s, a chip designer could manufacture their own chips, because the facilities were relatively simple. Today, chip designers probably can't afford to build their own facilities, so they fall back on foundries who make their chips and others' chips and can afford to build a modern facility. Of course, the requirements have changed over time. But, 50 years is a long time, when today it seems like new models come out every day.
A feeling I've had for awhile, then, is that in today's world, there is a zeitgeist of shortsightedness. Or a feeling that speed is all that matters. Or, that hard things are easy.
The focus is on how we can retire at 40. How AI can do most of our work for us, faster. How we can market a mid idea to a VC to get a couple million bucks so we can find a less-mid idea to achieve product-market-fit to get a few million more bucks to hire more people to grow a business that's worth a lot on paper.
Well, who cares? There is a lot of dry powder (investable cash) sitting at VC funds. They need to spend it on something. And, everyone knows how the game works at this point. Money to many startups. A couple work out. Many sound very promising early on, on the surface.
I guess I care, in that I fear people who used to solve the problems like the ones this article details, are now focused on not missing out on the well-oiled make-a-quick-buck machine. Smart people feel like they need to tie themselves to the next big model, to equity that will skyrocket, and then they are set for life. Maybe worse, they think it's just that easy, and their mental model for "hard work" and "success" re-calibrates to a false reality - one where being in the right place at the right time is what matters, and where doing hard things over time does not.
In that model, there is a real lack of durability. There is a big problem in achieving a certain level of technical achievement and a bigger problem in sticking with the evolution of things over time. Most people are fine with that, which begs the question of how we'll continue to achieve things like a modern semiconductor facility.
As someone who doesn't have any real desire to retire early and who enjoys going deep into technical problems, the glossing over of difficulty and the idea that a three-to-five year time horizon is plenty to get in and out, is a tough environment to thrive in.
To me, it's nice to see reminders that (i) things really are hard and (ii) great things do evolve over longer periods of time. When I struggle on a problem at work, or in a technical interview, it's nice to remember that not every thing should take 60 seconds and be perfect. It becomes easier to stomach making incremental improvements to a system over the course of the year when you can see how far it's come, and realize that a year isn't very long at all.
It's hard to imagine that we'll go back to working most of our careers at the same company or that the news cycle will extend. But, I hope that we can step back and appreciate how amazing some things are today. How long they took to get to that point. And, what that means for our own work going forward. Even with better assistance and higher levels of abstraction, things are hard, things do take time.
Pair with: Nobody cares.