>Actually, for triangulation in 2d space, you need 3 observation
>points. One gives you a circle, two gives you two points
>(intersection of two circles), and the third tells you which of those
>two points is the right one. You can probably get away without the
>3rd, if there's any directionality to the signal at all...it's only if
>it's omni-directional that you'd need them all.
>
>Likewise, in 3d space, you'd need four. One gives a sphere, two gives
>a circle (intersection of the surfaces of two spheres), three gives
>two points (intersection of the surface of a sphere and a circle), and
>fourth tells you which of the two points.

That's what I thought at first too. I think it depends on whether you're triangulating based on distance or direction. You're working by distance, and analyzing the situation correctly based on that assumption. If we're triangulating based on direction, though, we probably need two observers at bare minimum: each establishes a ray, and these rays intersect in at most one point (unless the observers are collinear with Kei). To get any sort of accuracy you'd probably want more. (This isn't even getting into curved spacetime, which probably doesn't change the conclusions, but makes the analysis more difficult.)

Incidentally, to the authors, I'm enjoying this series as much as any of the Eyrie works. I haven't said anything before only because it seemed I didn't have anything to add.

- Chad