This definitely happens in IT departments of large orgs. Centralization happens under the guise of efficiency / role specialization. That works for a while until clients at the edge can no longer get the service they used to get when their IT was a direct report in their org chart, and a decentralization effort ensues.

It also seems to be a function of the exec level change over. The new C level wants to make their mark by doing things differently from the last person.

Could be seen as an expression of dialectic development. The overall tendency in many (most?) living systems is towards integration at the next scale, which needs a certain amount of centralization. But whenever centralization is imposed it is usually overkill (the pendulum swings too far), which causes a decentralization effort (back to the autonomy of the parts), that however will retain some remnant of the previous centralization drive. At a certain point the decentralization swings too far too, and so centralization takes place again, in a slightly different way this time.

The combined effect of these oscillations is an equilibrium between centralization and decentralization of the organism which is optimally adapted to the environment at that scale of integration. Which is why neither centralization nor decentralization are ideals, they are both needed in an evolutionary process, albeit at different times.

I definitely see some pendulum effect scattered through the field.

Maybe there's a sensory about system pressure (need for more bandwidth, or whatever) that when above a certain threshold trigger that swing change.

There is a tension between complexity and order. A decentralized system like the internet is extraordinarily complex, and with that complexity comes robustness and resiliency. Sometimes people see complex systems and feel they’re messy or “inefficient,” and want to replace the complexity with something simple and orderly. Such systems tend to be centralized, easier to control and understand, at the cost of being less robust and resilient.

I also see two things:

- Investment barriers for innovation: you need something large to motivate funding and overcome research/innovation => centralized. When costs amortized, you start selling small and cheap.

- Maybe also a kind of inertia or fatigue. People will try to find ways to improve. When it's too hard to do so in centralized they'll try to distribute, or the other way around.

I remember a long time ago reading a comment about business schools basically teaching the same: you have a large company which is inefficient, so new management comes in and decides to split it into multiple companies that do different things; then at some point, each minor company is inefficient, so they merge with other companies to achieve scale benefits. Repeat forever.