Because most or all of its staff live in different somewhere elses and mass transit is horrible at random route commutes. If you take a city and express it in polar coordinates defined by each spine of mass transit as being at a constant theta, putting the things that lots of people need to come to near the center makes the most sense.
Locating an office 3 miles out of the center at a random theta means that a lot of people need to commute into the center, change to the line serving the office [waiting], and continue their commute out, reversing the same on the way home.
If I'm competing for the best employees, I'm far better served to pay more to put my office near the transit hub and lower the inconvenience for my employees.
All the above assumes a significant commuting base on fixed route rail that tends to cluster around hubs. It might be possible construct (more expensive) transit that was not hub and spoke. If you do that, I predict you end up in the less centralized design you describe in the last paragraph (which still seems like it doesn't serve the needs of people who need to work together but are served by different of the numerous hubs).
Right, but the problem with this isn't the different somewhere elses, it's designing mass transit systems that are only useful on arterial routes into or out of the centre of a radially organised city.
This is often how things work today, but there's no reason it has to be, as long as you have a reasonable alternative layout and sufficient volume of journeys to make comprehensive mass transit viable at all.
For example, consider a big city like London. There are enough travellers to run both the Underground (metro) and bus services almost 24/7 now. A wide variety of routes, many of them not just arterial paths to or from some central hub, cover most of the Greater London area.
In that sort of situation, even if you aren't living within convenient walking/cycling distance of your normal place of work, it makes little difference whether you're travelling into the centre of a city or further around it.
The assumption of a single, central transit hub rather than a more distributed, uniform arrangement is the problem here. It has much the same inherent scalability problems as any of the other services that some of us were discussing further up the thread.
All the above assumes a significant commuting base on fixed route rail that tends to cluster around hubs. It might be possible construct (more expensive) transit that was not hub and spoke.
Exactly. The challenge is to match the scale of the city with the scale of the transit system so the "(more expensive)" becomes negligible. But since in this case the costs of inefficient transportation systems and unnecessary journeys are more than just financial, that doesn't seem like a crazy idea.
Coincidentally, my office is right next to the busiest tram stop in the city center, which is great because I don't have to change lines during my commute.