Meaning if you want the best connection you must point the Bloomsky (in relation to your router) in this direction.
It will probably turn out that when close (as you note), rotation of the ball is less important. However, as the ball moves away from the router, then it might be that the difference between a signal maximum and the signal minimum becomes important or even critical.
As xcom said, a typical dipole has max sensitivity perpendicular the line of the two dipole elements (broad side), falling off to a minimum if the dipole is end on (looking down the line defined by the elements). Here, there are at least two complicating factors. The dipole (if that is what is under the tape) is a bit rounded by the ball. Probably more important, there are the metal parts on the board, and in the board as copper traces (possibly a ground plane layer too) which are within a wavelength of the antenna. To give a feel for the dimensions, each side of the dipole is usually about a 1/4 wavelength, the dipole being 1/2 wavelength long in total.
If it is a dipole, or dipole like, it looks to be horizontally polarized. Should WiFi antennas be horizontal or vertical?
There might be relatively simple ways to improve the impedance matching from the black coax into the existing dipole, or ways to replace the internal antenna with something better.
There is plenty of room in the ball on the other wall. There could be other types of antennas, or possibly an array, such as an array of two or four dipoles. The hams on the forum might have some ideas.
