I've been using Finale since Windows 3.1 (maybe it was Windows 95 -- regardless, we're talking printing your scores on clay tablets here), and I've written many shelf-feet of music with it. But I find myself abandoning it. A specific issue was the initial impetus, but trying alternatives has made me realize that larger issues are at play. I don't think they apply to the vast majority of Finale users, but they apply to SOME fellow composers. (BTW, I've been a software developer for that long -- including a stint as a developer on the Microsoft C++ team. Currently I'm paying the bills doing C/C++ development on the LAMP stack and Windows).

The specific issue is that I just can't save files reliably (error dialog box, refuses to do it). My own bag of Windows tricks sometimes works, mostly doesn't. I've taken it up with Finale Tech support, and their suggestions worked sometimes, but not consistently. And if I can't confidently save files I'm in big trouble. I'm running on Windows 7. I've gone so far as to reformat the machine and reinstall Windows 7. Same problem. I'm sure the fundamental issue is Microsoft's, but the end result is the same. (I'm on top of every new Finale update in the hopes this problem has been fixed. Not so far.)

So I got a Mac (for various reasons). Saving files works fine, but -- no accelerator keys! Arg! Being reduced to using the mouse for EVERYTHING slows me to a glacial pace. Sure I could implement keyboard macros -- I looked into doing that and it looks like an enormous and fragile time sink. And then there's Windows 10 -- I can hardly wait.

The thing is, I seem to be having analogous experiences with other tools. Microsoft Word is a great example -- each release seems MORE problematic than the last.

Generally, I'm wondering if I/we are up against software life cycle issues. The program works great in its youth, then 'bit rot' starts to set in -- nothing that a major code refactoring wouldn't fix -- if one could squeeze it in given the pressure from management, the board and stockholders to pile on more features to generate more sales. Refactoring has no sex appeal whatsoever and can 'always be deferred'. Eventually bit rot gets so bad that it exceeds the available resources to do the necessary refactoring EVER.

I'm also finding that (binary) Finale files in general aren't as permanent a record as one might think. I have any number of Finale files now that are too old to open. Something of a kissing cousin to the 'bit rot' issue. (Fortunately I have paper copies. I'm realizing that ink on paper is highly underrated.)

And there are things that I've been wanting to do compositionally that just aren't realistic in Finale (polymeters is a big one). I've looked at Sibelius (which has its own issues --- ongoing existence being a big one) and Dorico (I'm not really interested in being Steinberg's free beta tester).

So I've ended up with Lilypond, which is an entirely different paradigm from all the GUI based music engraving programs (Finale, Sibelius & Dorico) -- you write what amounts to 'music code' that you COMPILE into a pdf. This is not for the faint of heart, and certainly not for the vast majority of Finale users. All software designs involve tradeoffs, and the 'compiler' approach has its own problems and merits. Since everyone reading this is likely acquainted with the GUI approach (i.e. Finale), I thought some general observations of my experience with Lilypond might expand the imagination of the music engraving software problem domain in general.

First, the bad:

1) It is definitely a beast of a learning curve. Their online documentation is pretty good as open source goes, but there are still lots of gaps.

2) Initial note entry is slower than Finale.

3) Futzing and nudging elements is a real pain.

However:

1) In my experience with Finale, initial note entry is pretty fast (I've gotten good at Speedy Entry with left hand on a numeric keypad and right on a MIDI keyboard) but the editing stage of entering and nudging staccati, slurs, etc. is really slow. And requires constant fiddling. For about half of my slurs the Finale default shape/placement is so embarrassingly bad that I HAVE to tweak them. Lilypond is surprisingly good, however, at good or better default slurs and placement of articulations without intervention on my part, even in use cases that twist Finale around the axle. End to end on a non-trivial engraving project, I'm wondering if total time to complete for Finale vs Lilypond isn't fairly comparable.

2) I'm not sure that Lilypond making it hard to futz with perfectly good slurs/element placement is a bad thing. What am I trying to accomplish? A  clean score that can be edited and parts extracted, with output that other musicians can read. With Lilypond I'm not wasting time with unnecessary tweaking "because it's fun" yet doesn't make my score anymore readable.

3) Lilypond files are plain text. Sure, a future version of Lilypond might have trouble reading today's file because of syntax changes, but there is the real likelihood of opening the file in a text editor and fixing the human readable contents. Or recovering MOST of the file. Something impossible with my old binary Finale files. All in all a text format has got to be much less fragile than a binary one. (Yes, I suppose I could save xml versions of Finale files. Ugh, those are fun to edit, and how fragile/human unreadable is THAT format?)

4) Lilypond's plain text format means that I can bring to bear the Unix command line tools like grep, bash, etc. I'm undertaking an exercise book where a bash script and a tool I've written (25 lines of C++) generate the remaining 11 keys. Want to insert an exercise in the middle of the list? Add the exercise fragment file and re-run the bash script to rebuild the entire book. Done. (And Lilypond can work with LaTeX.)

5) No licensing issues. Install it wherever you want as often as you want (on Windows, Mac or Linux). It's open source.

6) 'Tech support' on the forum is very good. And based on monitoring the forum, bugs seem to get fixed in a week plus/minus.

There is certainly a place for GUIs. And for most people for most purposes the Finale GUI approach works well enough. But there are alternatives that have their own advantages.

(P.S. Don't even think about using Lilypond without Frescobaldi.)