You might be surprised at how boring it is to explain data trees over and over and over… I’m tired of it.
The surface fragment edge you have chosen with List Item is divided by 3 (4 points) so your random numbers are all between 0 and 3. The end points are duplicated by the adjacent edges’ end points. So what are you trying to accomplish with these random points?
Because this would make more sense if you first partition the surface fragments into rows and columns (Flip Matrix will swap them). Then when you get all done you can use the random points to make interpolated curves.
Here is one possibility. Nurbs curves work better than Interpolate for this because they don’t overlap and intersect. Note that curves don’t reach one end because of the edge method, very similar to the problem I mentioned in the other thread.
RandomPoint_2022_Nov3a.gh (10.7 KB)
It takes a lot of time to answer questions like this, and there are many similar questions on this forum.
P.S. This version gets the curves to reach the “top” edge. The yellow group is a little nasty (or clumsy?).
RandomPoint_2022_Nov3b.gh (12.5 KB)
RandomPoint_2022_Nov3bb.gh (18.8 KB)
P.P.S. Is this yellow group more elegant? 
RandomPoint_2022_Nov3c.gh (21.6 KB)