time travel - FTL combat in Newton's Wake

Some setup is necessary before I can get to asking my question.

In Ken MacLeod's Newton's Wake, FTL is possible. There is a form of chronology protection:

Say ye try tae send a signal intae the past, or your ship’s course mucks about too much wi the light-cone consistency conditions. Ye’ll find the transmitter disnae work, or your course takes longer or goes a different way than you plotted it, or – as the saying goes – she was never your grandmother in the first place.

Chronology protection is leveraged in a combat scene to win a battle:

The other two friendly ships, having diverged, were again converging. Another of the enemy ships fittled [went FTL]. The Stanley Blade’s and the Small Arrangement’s trajectories instantly halted. The enemy appeared, as it seemed on this scale, right beside them, and as rapidly was destroyed.
‘Ya beauty!’ yelled Lucinda.
‘What happened there?’ Armand asked.
‘Chronology Protection trap. It came out of the jump just too far away to hit them, and it couldnae fittle the remaining distance without going outside its own light-cone or back in time. They had a moment while it waited tae catch up wi itself, fired off a nuke, and—’ She clapped her hands.

So on to my questions: when she says "couldnae fittle ... without going outside its own light-cone or back in time," I'm confused because any time you go FTL, you leave your light cone, and there exists some frame in which you appear to travel backward in time. But this character is talking like sometimes that isn't true. So what am I missing here?

My other question is, if the author's treatment of FTL travel makes sense here, what does it mean for the ship to "catch up wi itself"? It has something to do with waiting until causality constraints pass, but how exactly does it work? I can't quite wrap my head around it.

Answer

To actually get causality violations in which an event is influenced by its own future, you need multiple FTL signals going in different directions, as in the tachyonic antitelephone; a single unidirectional signal may be received before it was sent in certain frames, but this alone does not qualify as a causality violation if there is no actual causal loop. So without having read the book, I presume the idea here is something like this. Suppose I send an FTL signal or make an FTL jump from event A to event B (with A and B being points in spacetime, not locations in space). And suppose there is another observer whose worldline passes through B, who immediately after that tries to send a signal or make a jump back in the direction of whatever spatial coordinate they would assign to A in their reference frame. Then they could find that FTL is strangely limited in that direction, in such a way that the return jump/signal always ends up in the future light cone of A. This would be the case even if they were in a reference frame whose definition of simultaneity assigns A a time-coordinate somewhat in the future of B, so that if FTL were completely unlimited in that direction, they would be able to arrive at the location of A before the event A actually happens. As for "catching up with itself", it could just mean a single ship making a pair of FTL jumps in opposite directions such that at the end of the second jump it ends up in the past light cone of the start of the first jump, which again would have to be forbidden if you want to preserve causality.

A similar idea, in which traversable wormholes allow for trips that look FTL from the perspective of space outside the wormhole, but where the wormholes blow up if you try to arrange them in a pattern that would allow for causality violations (inspired by real models of how chronology protection might work), is discussed on this page. Depending on how vague the book is about how "fiddling" works, perhaps one could imagine that it works by creating temporary wormholes, and that they obey this sort of chronology protection rule.