The decision was probably somewhat pre-made by Ben Gurly, the ingenious designer of the PDP-1. The displays coordinates are simply a memory word with a precisio

The decision was probably somewhat pre-made by Ben Gurly, the ingenious designer of the PDP-1. The displays coordinates are simply a memory word with a precision of 10 significant bits. So it’s managed quite "naturally" by under-/overflow wraparound. The art is really in withstanding the urge to introduce any boundary checks, like, "why should it bounce, space is infinite". (However, there were clones that introduced bouncing, including the Atari 2600 cartridge version.) Another decision probably dictated by the display hardware was the speed of the spaceships. Because of the long persistence of the phosphor, you get these iconic trails for moving objects. If you make the ships go too fast, you'll see distinctive outlines "stamped" onto the display. So you automatically want to make this as smooth as possible (so that the individual plots merge into a solid trail). This is on the other hand facilitated by the fact that the display coordinates provide 8 fractional bits in a standard PDP-1 18-bit word. So you can move and turn things by really small ("subpixel") increments.