> For research, the librarian does not need to be co-located with the library. Several librarians can work in a central location, like a research university or a national library, and serve the whole country.
Schemes to do this sort of thing repeatedly fall across time and industries.
The first problem is that people overestimate the amount of efficiency there is to be had. Oh, if we put them in a central location we can make 80% of them redundant. Ha! Librarians don't currently spend most of the day idle and somebody has to do the work they currently do regardless of where they're located. Meanwhile putting people in a central location generally makes things less efficient because now they're in a different location than the patrons and you've added communications overhead and, in the common case, a layer of middle management.
The second problem is that knowledge is local. When a patron comes in and wants to know about historical buildings in the library's city, or local regulations, or community events, the local librarian knows where to find the information because that's where they are. The librarian 100+ miles away may not, because they're most familiar with where to find their locality's information rather than some far away place.
The third problem is that diversity is advantageous. This is the sort of "laboratories of democracy" or "competitive markets are better than monopolies" effect. When everything is under one roof it becomes bureaucratic, resulting in ossification and resistance to change. Including changes that improve efficiency or outcomes. Whereas independent systems can each try a different thing and find out what works best in parallel, then look at how the others did and adopt their way of doing it if it seems better. Centralization efforts stamp that benefit out, and its long-term effect is cumulative because the benefits of the alternative accrete over time. This ultimately eats any nominal efficiency gains from the limited amount of legitimate redundancies that could be achieved through centralization.
The fourth problem is that central locations tend to become a single point of failure. Having multiple copies of information stored at multiple independently-operated facilities has high resilience. Centralized systems have systemic risk, and the best known way to mitigate that risk is to redecentralize the system to remove the monoculture.
In general if someone proposes to centralize a decentralized system to "improve efficiency" you should suspect them of malice, because the most steadfast proponents of centralization are generally adversaries who want to monopolize the system or adversaries who want to corral the system all into one place so they can more easily butcher it.