According to the third step of the rendering rules, the direction parameter is determined based on the cue text contents by applying the Unicode Bidirectional Algorithm.

Depending on the text direction, the cue is positioned differently within the video element with the same cue settings.

An use case of not changing the final positioned cue with respect to the determined language are translated captions.

For example, if the cue was positioned in a specific part of the video as to not occlude visual elements of the video, this would allow the translated caption to preserve this property if the final position is not changed.

Adding a setting that would override the default language dependent behavior would allow caption translators and automatic services to have no knowledge about the different positioning schemes of the cue (and not require to reverse engineer the original position).