No city embodies the dual nature of the US economy quite like San Francisco. On one hand, it stands as a prominent figure in the cutting-edge realm of artificial intelligence, the cornerstone industry of the 21st century. Conversely, it also showcases evident weaknesses. As highlighted by Reason magazine:
San Francisco made headlines in October 2022 by allocating a staggering $1.7 million for a solitary public restroom in the Noe Valley neighborhood. The exorbitant cost was attributed to the steep construction expenses in the city and lingering supply chain challenges.
However, the state intervened shortly thereafter, abandoning the proposed restroom amidst public outcry over its hefty burden on taxpayers. Fast forward fifteen months, the public square where the restroom was slated to be remains devoid of a restroom facility—with no imminent resolution in sight.
While perusing this piece during my stay in Tanzania, a country boasting well-maintained public park facilities, I couldn’t help but ponder: why does a nation with a per capita GDP of \(1327 (\)3595 PPP) excel in constructing toilets compared to a city with a per capita GDP of $144,600? Surprisingly, according to Reason, San Francisco finds itself in a predicament where constructing such amenities seems financially unfeasible.
Of course, the term “afford” is not interpreted in a literal financial sense here; San Francisco undeniably possesses the financial means to undertake such projects. Rather, it reflects the conventional notion of affordability, akin to deciding against a lavish hotel stay due to its high cost. Given the substantial expenses involved, the venture appears impractical. Nevertheless, why does San Francisco grapple with a scenario where the cost outweighs the benefits, despite having resources exceeding Tanzania’s by over a hundredfold?
One plausible explanation could be that Tanzania’s efficiency in constructing toilets surpasses that of San Francisco by a significant margin. It’s conceivable that Tanzania could construct a functional single-stall toilet for under $17,000. But what drives this stark contrast? Isn’t building toilets a less complex endeavor than creating sophisticated language models?
Reason magazine posits that the root cause lies in a web of regulations, including mandates necessitating approval from meticulous design authorities, staunch environmental advocates, and the utilization of unionized labor:
San Francisco boasts the highest construction costs globally—a hardly surprising revelation. Erecting a public restroom in Noe Valley, even at a site already equipped with the essential plumbing for such an addition, entails navigating a labyrinth of regulatory hurdles. These obstacles range from securing endorsement from the Arts Commission’s Civic Design Review committee to complying with the California Environmental Quality Act, in addition to obtaining consent from the city’s Rec and Park Commission and the Board of Supervisors. Moreover, the project would be subjected to a phase of “community feedback.”
Even post-approval, the city isn’t at liberty to opt for the most cost-effective restroom solution—likely a prefabricated unit—and link it to municipal plumbing. As per a 2022 San Francisco Chronicle report, prefabricated bathrooms contravene the city’s Public Labor Agreement. Furthermore, the city is mandated to engage unionized labor for the construction process, further inflating costs.
In stark contrast, the development of large language models (LLMs) encounters minimal city regulations.
While the toilet scenario may appear trivial, it mirrors a broader issue. Similar regulations extend to the housing sector—an integral industry in America, synonymous with influencing living standards significantly.
This microcosm encapsulates the essence of America. Unshackled from intrusive regulations, we emerge as the most industrious society in history. Yet, in heavily regulated domains, we often fall short of matching the efficiency of developing nations like Tanzania. It’s imperative to address such disparities, ensuring basic amenities are readily accessible as we navigate towards an AI-driven future.
