A synchronous interconnect provides not just a source of truth but also stabilizes your grid frequency. If you have an isolated grid you have to match generation to demand to keep the grid frequency stable. If you have a 1GW interconnect that means you can mismatch generation and demand by up to a gigawatt and still be fine. I imagine that makes for a much faster startup procedure
You can connect two running grids. Earlier this year the Baltic countries disconnected from the Russian grid, and synchronized and then connected to the European grid.
I imagine you can get close enough by syncing to a shared time source like GPS or the DCF77 signal, as long as you communicate how the phase is supposed to match up to the time source. Or at least you could get close enough that you can then quickly sync the islands the traditional way.
The question is if it's worth the effort and risk. Cold starting a power grid is a once in a lifetime event (at least in Europe, I imagine some grids are less stable) and Spain seems to plan to have everything back up again in 10 hours. Maybe if the entire European grid went down we would attempt something like that by having each country start up on their own, then synchronize and reconnect the European grid over the following week.
Nothing that complicated. You just carefully synchronize the state of the grid on both sides of the interconnect, and when they are perfectly matched, you throw the big switch.
It's a bit difficult in another way: Obviously 50/60hz is not such a high speed that it's difficult to synchronize.
The harder part is this: To pump power into the grid you lead the cycle ever so slightly, as if you were trying to push the cycle to go faster. If instead you lag the phase the grid would be pumping power into you.
That lead is very very small, and probably difficult to measure and synchronize on. I would imagine that when the two grids connect everything jumps just a little as power level equalize, it probably generates a lot of torque and some heat, I would assume it's hard on the generator.
From a physics point of view, by leading the cycle you introduce a tiny voltage difference (squared), divided by the tiny resistance of the entire grid. And that's how many watts (power) you are putting into the grid.
Yes, but that actually makes it harder to start up.
To synchronize the isolated grids, they all need to operate with an exact match of supply and demand. Any grid with an oversupply will run fast, any grid with an undersupply will run slow. When it comes to connecting, the technical source-of-truth doesn't matter: you just need to ensure that there will be a near-zero flow the moment the two are connected - which means both sides must individually be balanced.
And remember: if you are operating a tiny subgrid you have very little control over the load (even a single factory starting up can have a significant impact), and your control over the supply is extremely sluggish. Matching them up can take days, during which each individual subgrid has very little redundancy.
On the other hand, the interconnect essentially acts as a huge buffer. Compared to the small grid being connected, it essentially has infinite source and sink capacity. For practical purposes, it is operating at a fixed speed - any change is averaged out over the entire grid. This makes it way easier to connect an individual power plant (it just has to operate at near-zero load itself, move to meet a fixed frequency target [which is easy because there is no load to resist this change], and after connection take on load as desired) and to reconnect additional load (compared to the whole grid, a city being connected is a rounding error).
The UK keeping its own time just makes things easier for it IMO.