NavList:

As a rule of thumb the smaller the space (often a triangle, though can have four or more sides) that encompasses the presumed fix the better. This makes intuitive sense; since as you note perfect LOPS would coincide at a single point, the fix.

Note that ‘presumed fix’ lying at the centre. That’s all it is, not necessarily the actual position. It would be if the LOPs have identical error; this is the presumption. But they may have differing error. Two LOPs could be fairly good, the other wildly wrong, shifting the fix away from what the ‘good’ LOPs indicate. The actual position may well lie outside the LOPs. With a triangle made up of sights taken from less than any 180 degrees of azimuth (from less than half the visible celestial sphere) the actual position WILL lie outside the LOPs.

So the larger the area encompassed by the LOPs the more likely it is to include the actual position. This is logical, but not very helpful. What we need is a single fix position, which can only lie at the centre of our calculated LOPs.

If your sights lead to a nice tight bunch of LOPs then that is PROBABLY an encouraging sign, just as the opposite probably calls for caution. One way of proceeding from LOPs with ‘much spread’ is to take their indicated fix as a DR and repeat the sight reduction. This technique, known as reiteration, may give LOPs with much less spread. Another is to continue taking sights as possible and reducing them – nav is an ongoing process, ideally with fresh data arriving regularly to check and test and move forward the results of old.

As to how long that bit of string is – a lot depends on the sight taking conditions.