Why does it lie!

Interesting development last week, changing the order of some of the code and fixing some minor problems gave a slightly different output. Observe below;

The graph 3elt drawn in 3dimensions using the Barnes Hut approximation. As can be seen, the graph is drawn very well (much like outputs seen previously) when seen from the front (0,0,1 > 0,0,0), however, when rotated (1, 0, 0 > 0,0,0) the view shows the same z-expansion shown previously.

This makes this work, the most awkward and infuriating peice I have come across thus far. The code is now clean and most errors identified and fixed, but this output is awful. From one direction, the graph is fine, but any other, it isnt? What kind of sorcery is this? Why does my algorithm hate the z-axis? What is the significance of edges changing direction half way down? is that 0,0,0?

As always, I will continue to search for my mistake and will update accordingly. For now, here is my favourite graph, data, showing the same problem.