Nice to see confirmation that at least the end result of "fuckloads of energy being released" is accurate.
It isn't, really. The energy isn't
released, and there isn't a large amount of it either. In a sense, there isn't any at all.
By creating a causal loop, you create a zero-length four-dimensional path from A to A, but which passes through your entire FTL path instead of being just... the single point, A. This allows an infinite amount of virtual particles to stack on top of each other, which one way or another
will ruin your FTL drive, so you can't keep going and make a
negative-length path.
(That is, a path from A to previously-at-A. Not really negative length.)
However, virtual particles are... well, they're not real. They're a mathematical trick used to simplify calculations, and they don't behave like ordinary particles. There are other formalisms (which I prefer) in which they don't exist at all, such as lattice QCD, and I was admittedly being lazy in naming them.
In particular: You can't divert any amount of them from this loop. It might seem like you should be able to, but these things only exist in the first place
because of the loop.
Virtual particles are a (spontaneous) fluctuation in the underlying quantum field, and have... well, not exactly zero energy, but from a distance they don't exist, so effectively zero. In a sense, they're always counterbalanced by a dip in the same field -- though the formalism in which they exists seems to punt on calculating that. Conservation of energy also still holds. (
Conservation of energy does not hold for the universe as a whole, but that's besides the point here.)
(They're a mathematical trick, and the energy they can carry is inversely proportional to how long they exist, aka. how long a 4-dimensional path they travel. See the problem there?)
It's not that their apparent existence here has no consequences, however. I'm not sure what might happen
precisely at 'just barely have a closed loop' point, since by definition any virtual particles there won't be able to take any 'detours', but certainly once you get even a
tiny, electron-sized negative delay in there, they'll be able to take 'detours' such as from bumping off the particles in your FTL drive. And there's an infinite number of them (ish; from one view it's the same virtual particle overlaid on itself an infinite number of times, but from another quantum fluctuations (or particles) don't have identity), so they'd interact with every real particle in the loop at the same time...
I don't know exactly what would happen, but I do know you wouldn't survive that. In effect, this is taking random vacuum fluctuations and boosting them up to infinite magnitude.
From a different perspective: Their magnitude is inversely proportional to their size, and you just made a macroscopic section of space (i.e, the part containing your ship, the spaceport you launched from, all the space in between,
you, etc.) have zero 4-size, by making its 4-length zero. Well done.
When talking about FTL options, S here is better thought of as the mass-acceleration invariant, eg no matter how much you accelerate S remains the same, and likewise if you could change your mass (changing it to being imaginary essentially forcing you to FTL - this is the turn yourself to tachyons answer). You have a similar situation with regular euclidean distance, which is the rotation invariance.
The way I like to see it, our universe is four-dimensional; S is invariant because it measures something real, i.e. the four-dimensional length of a path between two events, whereas the distance in space or time varies depending on your perspective for pretty much the exact same reason that flagpoles have a different apparent length on a photograph depending on whether you're standing beneath it or far away from it, seeing it at different angles.
This intuition has trouble with FTL paths, being as they'd have imaginary lengths, but FTL paths don't seem to exist so that's not really a problem. The set of events which can affect each other becomes exactly that set for which at least one path
exists between the events.