Comment by jdw64

US aggregate productivity metrics fail to address this nuance. There is a fundamental difference in abstraction layers between a macro-system becoming more efficient and an individual enterprise experiencing operational failure. As a software engineer, distinguishing between these layers is critical. Your argument is akin to claiming that because the Google Play Store sees a higher volume of app releases (increased productivity), the intrinsic quality of individual apps has naturally improved.

In this analogy, the individual app represents a company, and the Play Store represents the broader US market. Silicon Valley’s highly liquid labor market allows talent to flow freely, which opens up and elevates the baseline of the overall market. However, that is entirely distinct from the fact that individual companies are suffering severe drops in internal quality and productivity.

Furthermore, in software architecture, 'productivity' and 'quality' are rarely directly proportional. With AI coding tools, we can ship an app orders of magnitude faster. Historically, it took me three months to write 60,000 lines of code; recently, I am generating that same volume in just two weeks. My productivity has undeniably spiked, but can I confidently claim the code quality is better than when I manually scrutinized every single line?

The real issue is not whether the broader economy has grown more productive since 1947. The core issue is whether a specific organization bleeds capability when the exact people who understand its real-world constraints, failure modes, and operational history walk out the door.

Both realities can co-exist: National productivity can trend upwards, while individual companies simultaneously suffer operational regressions due to botched migrations, failed refactors, or the loss of tacit knowledge.

I agree that 'institutional knowledge' is sometimes weaponized to defend unnecessary complexity. However, the opposite fallacy is treating all localized, domain-specific knowledge as worthless. While some of it is merely job-security folklore, the rest is literally the only surviving documentation of why the system functions in the first place