Which way you bridge is a choice, not a fixed property of the part
A rectangular opening or gap can often be bridged in more than one direction depending on how the part is oriented, and the span length in each direction is rarely the same. A gap that is 20 mm one way and 35 mm the other has a meaningfully different bridging risk depending on which direction the printed lines actually run, since a shorter span sags less and cools with more support from its own tension before it has to hold itself up over the full distance.
How the comparison works
The two directions you describe are compared directly by span length, with the shorter span presented as the generally lower-risk choice. If your printer's bridging ability on this material has not already been tested, the result explicitly flags that as a reason to treat even the lower-risk direction as unconfirmed rather than a guaranteed pass, and recommends a small test bridge first.
Worked example
A 20 mm short-axis span versus a 35 mm long-axis span, bridging ability not yet tested: the tool recommends bridging along the short axis as the lower-risk direction, but explicitly notes that since bridging ability is unconfirmed on this setup, a small test bridge is worth running before committing the full part.
A common mistake
A frequent error is comparing bridge directions only by their span length while ignoring which direction the cooling fan actually reaches best. A shorter span aimed away from the strongest airflow can underperform a slightly longer span that cools efficiently. If your printer's fan ducting favors one side, factor that into the direction choice alongside the raw span numbers.
Limitations
This compares bridge-span direction and recommends a bridge test when your printer's bridging ability is not yet known; it does not simulate cooling performance or guarantee a specific span will print cleanly. Run a small bridging test print before committing an untested span on a part that matters.