I generally use the voice colors to keep track of voices since some day down the road, when one will be able to assign different MIDI channels and sounds to individual voices I want to have my ducks in order to make use of the added functionality.
It depends what kind of memory you’re worried about: you don’t need to worry about your computer memory as far as the number of voices you use on a given staff in Dorico is concerned, but you might want to consider your own memory!
There’s currently no way to know which voice a note is in, is there? Like, I can tell that two notes are in the same voice because they have the same color. But there’s no way for me to definitively know that those notes are in Downstem Voice 2… right?
Select the note, hit Enter or Shift-N to invoke the carat, and look at the carat design. It’ll show you whether it’s up- or down-stem, and which number it is.
Ah! Cool, thank you 
This really is a marvelously powerful (and simple) system. Start your score, hit enter, start adding notes. Tada, you’ve added voice one and didn’t even know it. Don’t need a second voice? Don’t add one. Need a second voice? Add one (choose your stem direction). When you enter up notes, make sure the carat has an up stem; want down notes? Make sure the carat is down (and that there are/will be notes in the upstem voice otherwise natural stem directions will follow).
I do organ music a lot; I say this because I think there are few examples (apart from harpsichord and some piano music) that can trump my multi-voice use cases. This system is so flexible. I’d utterly loathe to go back to the old way! I’ll never forget the first time, back in the prerelease development days when Daniel shared an example of some complicated Bach. It blew me away!
To the OP: if you’re looking for a hard-n-fast rule, just consider upstem voice one the same as “voice 1” in Fin./Sib. Voice 2 in the old programs was downstem which is the equivalent of “downstem 1”.
In other words, as far as F/S were concerned:
Voice 1 = upstem 1
Voice 2 = downstem 1
Voice 3 = upstem 2
Voice 4 = downstem 2
This does gross injustice to Dorico to think in these terms, but it might make it easier for you to get started. The main thing is don’t add voices Willy-Nilly. Just use the enter key to activate note entry and “v” to cycle between voices already added. Try and stay in voice 1 as much as possible and only go into additional voices if there is a need.
The main reason you had to keep track of voice numbers in Sibelius is because it didn’t attempt to work out how to arrange simultaneous notes to avoid collisions between voices. So there was a fixed and small number of voices (4) on a staff, with a set of arbitrary rules which you had to learn about which voices were vertically aligned and which were not.
Since Dorico sorts out the note spacing for itself, the entire reason for having to know that using voices 1 and 2 for up and down stems in Sibelius would give different note spacing from using voices 1 and 4 simply doesn’t exist any more. The only thing that matters is the stem direction - and that doesn’t matter much, because you can flip the stems on individual notes without wrecking the note spacing.
so to summarise, there are actually two schools of thought in this post - the “carefree” school which says “don’t keep track of voices and just make sure they’re pointing in the correct direction”, vs. the school of thought that says “stick to one or two voices so that rests don’t start randomly appearing” due to (quoting @dankreider) “voice mis-management”, and also to eventually use it for better MIDI functions later on…
What I take away from this post is that it’s probably good to stick to one or two upstem/down-stem voices if possible, but it’s also OK if you don’t - as long as you don’t run into trouble - in which case you can fix it.
The lack of consistency in approach is initially alarming but the lack of hard and fast rules about voices is also rather liberating.
See? You’re already starting to get used to the new paradigm. Rob’s and Romanos’ explanations above are excellent, especially as concerns the difference between how Finale and Sibelius handle polyphony and how Dorico does it. The latter is more intelligent and flexible and, in the final analysis, more problem-free, regardless of whether you’re ‘carefree’ or fastidious about voice management.
For non-Sibelius users, this is what it does by default with different combinations of voices (at least what it did, in a fairly old version!) compared with Dorico, where people who never used anything else may be wondering what all the fuss is about
Nice example! The results in Finale are about the same as in Sibelius.
hmm the difference between the four Dorico examples isn’t as apparent to me, aside from the colours.
There isn’t any difference. That was the point of the example - the only thing that matters is the stem direction of the voices, not the voice “numbers”.
A friend of mine, whom I persuaded to switch to Dorico, was still struggling a bit in the beginning with the voices concept and the available shortcuts, and got into the habit of creating new voices (shift-V) for every bar with more than one voice in a staff. He ended up with dozens of voices per staff. With voice colours on, his score looked like a psychedelic christmas tree. Yet, while it may not be best practice, Dorico didn’t blink, it still looked great in print… 
I thought someone would at least mildly disagree, Dan… I’ll probably hit a situation where it bites me, but I’m in Dorico everyday, all kinds of projects, and haven’t found a problem yet. I used to be anal about this in Finale, but it all just seems to work in Dorico.
Dorico isn’t the first notation program to work like that. Creating new voices for every small section of music (not necessarily aligned with bar lines) on every staff is one fairly “standard” way to use Lilypond to engrave complicated notation - though you do sometimes hit problems when you have thousands (not dozens) of voices.
I have found a few situations in Dorico when it’s worth doing a bit of forward planning over voices. For example if you have a slurs starting and ending on notes in one voice, you don’t always get automatic collision detection and reshaping the slur for notes and beams in other voices if Dorico can’t find a “logical” way to position the slur. Putting at least some of the intermediate notes in the same voice as the end notes (even if you have to flip a few stems to do it) can be nicer option than manually tweaking the shape of the slur.
Is there a rule of thumb for best how to arrange them? Like say you have four voices, top line is up-stem 1, next is up-stem 2, followed by down-stem 1 and bottom is down-stem 2?
It’s per-staff. The first voice on all staves is Upstem Voice 1. If you have soprano-alto and tenor-bass on 2 staves, that’ll be just two voices: upstem voice 1 and downstem voice 1.
If it’s more than two voices on one staff, it’s probably best to just alternate them, but it entirely depends on the music. And of course you can easily flip stems any time.
Okay WOW I’m glad I read this thread, because I just found something really really cool.
“Select more” is scoped to voices. What this means, for me, is that I can place each phrase in its own voice. Then when I want to explode a particular phrase to other instruments, I can select one note from each voice in the phrase, select more to get the whole thing, copy and explode.
Previously I had been writing long lines in a single voice, and then copying the first and last notes of each phrase. That works, but this is better, especially when there are multiple lines on a single staff.