Is there a specific (theoretical) reason for using ties or different voices for notes of different length?
For instance this:
against this:
I understand that if it's literally meant to be sang by two different vocals you'd definitely use the latter. But I'm wondering if you can use it when scoring for a single instrument too? To me it seems more clear/readable than the tied notes. Well, maybe not in this example, but in pretty complex harmonies or motions it does.
But I'm wondering if you can use it when scoring for a single instrument too?
Yes, you can use it for a single instrument as well! It's pretty common in polyphonic instruments, like piano, guitar etc.
I've seen both of them, and if I remember correctly, the first one is usually used in more amateur sheet music.
The only time I use the first one is when I'm composing something and I'm not sure of how long I want to hold the note for; you quite often change the duration of the notes and it's far easier to edit the first one. But when I'm writing the final 'clean' sheet music, I will always use the second version, because like you said, it's far easier to read.
While the second version is far more common, and often does actually relate to different voices, I see no reason for using the first, even if both 'voices' are played on the same instrument. In fact, even then, they can be considered to be two voices. Can't remember coming across such as the first version.
Tieing some notes of single-stemmed chords tends to obfuscate the visible structure. It partly depends on the instrument and its ways of attack whether you do that: for a percussive instrument, the ties may serve as a reminder to "retrigger as necessary". It also depends on whether your notes are intended as constituents of a harmonic progression or actually as voicing. When doing a piano reduction, munging orchestral voices in that manner is pretty customary. For polyphonic original piano compositions, it would be quite more unusual.
Basically, if you consider this multiply voiced music which you'd split to different singers/instruments in a consistent manner if you only had single-voiced instruments at your disposal, it tends to make more sense to actually notate stuff in a multi-voiced manner. Your example is a bit extreme in that regard: one would hardly ever write it in the first manner you show.
When the notes are more haphazardly distributed and a consistent voice distribution is not feasible (are you doing your listeners a favor, though?), the first notation frees you from having to think up voice distributions making notational sense.
