Two minor footnotes: First, this was framed as a "centralised versus decentralised" debate. There are of course many variations on the decentralisation theme. (Do you really mean distributed generation, rather than transmission? Does this include microgrids? Etc.) Given the way the questions were asked, I simply took it to mean the absence of a centralised electricity grid. Second, when I talk about first-best and second-best alternatives, I don't quite mean in the strict economic sense of optimality conditions. Rather, I am trying to convey the idea that one solution is only really better when the other is unavailable due to outside factors.
Why are grids still vital? Why is a functioning electricity grid necessary for economic growth?
These two questions are more or less the same, so I‘ll take them together. Large, centralised grids constitute the most efficient and cost-effective way of delivering (and consuming) electricity in modern economies. Not only are decentralised options substantially more expensive and (generally) less reliable, there’s no intrinsic reason to believe that they will be better at delivering a clean energy future.
Is the centralised argument being lost in places like SA where the grid is so poor and not being improved?
I wouldn't say that South Africa’s present electricity woes are the result of grid failure. Rather, the problem is primarily one of generation capacity and government mismanagement. On that note, the grid is the one component of the electricity system that is best thought of as a “natural monopoly”. (The other components of the electricity value chain – i.e. generation and distribution – should then be left to competitive forces.) Your question highlights an irony. Eskom’s mismanagement on the generation side (huge overspends and delays on the Medupi and Kusile power stations, etc.) are undermining confidence in its ability to manage a centralised grid, the one aspect that government can legitimately claim needs to be operated as a regulated monopoly.
That all being said, Eskom is falling behind the required investment goals for maintaining an adequate grid infrastructure into the future. A deficient grid network has also constrained economic growth in many other developing countries, from Nigeria to India. And, yet, this is not to say that the decentralised alternative offers an intrinsically superior solution. A grid system remains the first-best option. Decentralised solutions are really a second-best option in the absence of the former. The distinction is crucial.
What role for de-centralised solutions?
I think that decentralised solutions will remain a second-best, niche alternative for the next few decades. There are several things that cause me to take this position, of which intermittency and local storage are probably the most pronounced. Now, there do happen to be a number of exciting developments on the storage issue, but nothing that I would expect to fundamentally change the equation. More to the point, I believe that the resilience of a decentralised generation system will fundamentally require a functioning grid. The increased intermittency and smaller scale of decentralised power production will necessitate excellent access to similar, small-scale generation in other regions. This can only be achieved through a robust grid network. (An example may help to make my point: Germany’s much-fêted Energiewende was supposed to involve a fundamental shift towards the decentralised paradigm. What we've seen in practice, however, is that the Germans are investing hugely in extending their inter-regional grid capacities to places like Norway, whose hydropower resources offer the most cost-effective means of accommodating the intermittency of wind and solar.) Similarly, the parallels that people inevitably draw between a dentralised electricity system and the communication sector (i.e. where fixed-line telephones were leap-frogged by cell phones) are misplaced. Beyond various other differences, cell phones networks are fundamentally centralised in nature: Cell phone towers are the grid equivalent of the modern-day communications sector.