The three major compositional functions of Finale are MIDI piano keyboard input, logical functions (such as cut and paste, transpose, or undo), and playback. By comparison to paper and pencil, Finale makes certain processes such as cutting and pasting phenomenally easy. The insertion of new measures in the middle of a passage is now painless  and the transposition of an entire piece can take just a second when it would have formerly taken hours. Composers no longer need to imagine all the parts in their heads or to play them on an instrument (such as the piano), as Finale enables them to hear texturally complicated music and to quickly complete operations that formerly required a great deal more time to perform. For instance, with Finale, it is now easy to move an entire line over by a quarter note. Composers are no longer required to be able to play the music; Finale will play back a passage using a timbre that is close to that of the desired instrument, and Finale will play it at performance speed. Most of the notation work is done during the composition process (no need to make a “fair copy”). My interviewees all contend that these pros by far outweigh the cons we are about to discuss. Indeed, by removing many of the cumbersome aspects involved in composing music by using paper and pencil, composers can devote more of their time and energy to perfecting their music, and to operate edits that composers of previous generations might have shied away from undertaking. Yet, the list of cons is far from negligible, and I will not be talking here about program glitches, although unfortunately, new problems appear with each new version.
When Finale users input music through a MIDI piano keyboard, instead of using other available input methods (such as mouse point-and-click or computer keyboard), the setup encourages the composer to improvise at the keyboard while composing, and can lead to pianistic scorings and gestures, even when composers are writing for instruments that are unrelated to the piano. By incorporating the piano for MIDI input, Finale acquires all of the baggage of criticism that had been thrust at the piano (i.e., the composer favors pianistic chord spacings or pianistic gestures when composing for other instruments than the piano). Moreover, the keyboards used for MIDI input are often smaller than the full 88-key piano keyboard; the smaller the keyboard, the more the user will need to go through extra steps in order to notate pitches outside of the little keyboard’s range.
Computer screens tend to be too small or to lack the resolution of a sheet of paper. For instance, it is impossible on current screens to see an entire orchestral score at once, legibly. Physically handling paper remains clearer and more intuitive than clicking on a zoom button. Moreover, MEPS programs still do not allow one to simultaneously display pages that are not adjacent. Display problems also make it difficult for the composer to visually keep track of what is going on in an instrument which does not have a staff nearby. Because of the isolation of the “zoomed in” mode and the impracticality of temporarily migrating to a different part of the score, it is easier for the user to rely on the playback and to devise or alter counterpoint from hearing the context, rather than seeing it. While printing the score can be a useful remedy, the score only represents a snapshot of the piece at a given point in the process. Since the computer file contains the latest modifications and additions, composers may need to print new versions very frequently to ensure they are working off of current information, a process which could become overly cumbersome.
Even though Finale is technically capable of printing complex scores that involve a great deal of graphics and 20th century extended notation, this is so “non-idiomatic” and time-consuming that a composer may save time by handwriting the score. Ultimately, some scores are completely beyond Finale’s capabilities. The difficulties even extend to some exceedingly basic functions such as the notation of tuplets or the changing of meters. Notating either of these requires an amount of time and effort disproportionate with their currency, and is frustratingly complex when compared to how easy it is to indicate them by handwriting. Such a simple rhythmic figure as the one in Figure 1 requires an odd, unintuitive additional step.
Figure 1 : A simple rhythmic figure.
After telling the program that one wants to input a triplet of some kind, one has to input the first pitch as an eighth note, and then change itto a quarter note.
When faced with having to input a large number of these figures, we shouldn’t be surprised if a composer consciously or subconsciously avoids the use of this rhythm. In fact, precisely because of this difficulty, if a piece is to include roughly the same amount of triplet rhythms as duplet rhythms, and alternate between them frequently, a composer would save time by choosing 6/8 as the meter. Thus, instead of the 2/4 passage shown in Figure 2.a, a Finale-using composer might prefer to write the same passage in 6/8, and use the dot (very convenient to input in Finale – just press the period key) when a duplet is needed (Figure 2.b). Performers, however, find this sort of notation to be more confusing, and, if 6/8 is necessary, would prefer to see a duplet (Figure 2.c), which is by far the most cumbersome to notate using Finale.
Figure 2 : Three ways of notating the same melody.
MEPS is notoriously inaccurate (and thus inadequate) in closely imitating a live performance by human players. Among the issues with playback, until recently, Finale had a limited array of timbres and involved MIDI sound banks that sought to imitate the sounds of a selection of instruments, through synthesis (as opposed to sampling). More recently, Finale comes bundled with a far more realistic sound bank (Garritan Personal Orchestra). Very realistic sound banks (e.g. Vienna Symphonic Library) are also available separately, for several thousands of dollars. Even if a composer can afford to purchase these, the sounds remain limited to traditional orchestral instruments, playing in traditional ways. Composers interested in writing for an instrument that doesn’t belong to this bank would have to either record samples of their own, or find, among the patches that are available, the one that sounds most similar to the real instrument. In addition, the sound banks cannot be comprehensive in their coverage of extended techniques (e.g. multiphonics), as performers are constantly on the lookout for different ways of coaxing sounds from their instrument. Finally, sound banks are still utterly unable to simulate a human voice or a choir singing text.
Percussionists are in charge of an ever-growing arsenal of instruments, including for instance custom-made instruments such as Xenakis’s sixxen, built from other objects bought in hardware stores or collected in junkyards. A composer such as Kathryn Alexander (b. 1955) was so eager to hear these sounds as she was composing that she actually recorded a patch for each of the non-standard instruments she was using. Such a time-consuming, uncompromising, and patient approach is characteristic of the veteran composer, constantly questioning the process and technology, and finding ways of extracting what is needed. Younger composers are less likely to undertake such extreme remedies. (It is not so much a matter of laziness as much as a difference in approach, as explained above.) Generally speaking, even if Finale does provide a patch for the instrument, it does not provide the variety of sound that can be obtained on one given pitch (variations due to changes in embouchure on a wind instrument, for instance, or variations due to alternate fingerings). Even with relatively good sound banks, the computer does not faithfully reproduce a performance with human players, who will for instance adjust their intonation to a given harmonic context. Composers who use the less optimal sound banks may become convinced that certain instruments (e.g. strings) or combinations are undesirable, based solely on the computer playback, when an actual performance would prove perfectly acceptable. Since slower tempos focus the composer’s attention on the sound (which tends to be ugly), composers are wont to avoid slow and texturally thin music. As Chris Watson explains, the dry nature of early MIDI sounds caused composers to want to fill the void with a lot of activity. This resulted in busy textures and excessively fast tempos. This problem is greatly reduced now that sounds are closer to those of real instruments.
Finale may also mislead the composer by inaccurately depicting the relative balance of each instrument. Finale will willingly play back superhuman gestures that performers have a great difficulty playing, or cannot achieve whatsoever. Beginning users, for instance, may not realize that instrumental ranges are limited, or that they do not correspond exactly to the range of the keyboard they happen to be using. In the less optimal sound banks, instruments are programmed to produce a sound throughout the entire range of the piano. This could be partly due to the fact that a particular patch might be intended to simulate a whole family of instruments (from bass flute all the way up to piccolo, for instance). But an inexperienced user may be misled into believing that the flute can in fact reach several octaves below middle C. More recent programs address this issue automatically in two ways. The patch can be designed such that notes are only available on pitches that a real instrument can play. Sibelius warns users when they are about to input a note that lies beyond the range of an instrument. While Finale does include a plug-in that allows users to check the range of each line, this implies inaccurately that instruments have a finite and constant range, whereas this varies from player to player and across different models of the instrument (mouthpiece, reed, etc.)
Range is an easy problem for a computer programmer to remedy when compared to the following, blurrier issues. Finale is not, as of yet, aware of what constitutes a difficult or impossible trill on any given instrument. Finale will not detect music that exceeds reasonable performance speeds. Finale cannot keep track of how much air is left in a performer’s lungs and happily plays back notes indefinitely, even in wind parts. The lips of Finale’s brass patches never tire. The pianist in Finale’s playback can play 30-key chords if desired.Lastly, Finale will ungrudgingly calculate and perform any rhythm, even if its complexity would baffle the most seasoned human performer.
Inability to correctly interpret notation for playback. Percussion notation is non-standardized (by comparison to, say, the violin), and essentially needs to be redefined or at least clarified at the beginning (or in the course) of each new piece. Getting Finale to simultaneously notate and perform percussion music accurately can prove a challenge. Users often forgo trying to get Finale to play back the percussion part, or create two different files: one that looks right and one that imitates sounds accurately.
Finale’s difficulties with simultaneously notating and playing back percussion apply to all 20th century unconventional notations, for which, once again, users are well advised to forgo hearing playback. While MIDI can perform pitch alterations relatively easily, Finale cannot easily be used for the writing of pieces outside of the standard equally tempered, 12-note tuning system. Finale is currently incapable of playing back scores that call for even the most basic and controlled forms of improvisation (such as, for instance, figured bass, or jazz charts).
Stereo bias. Because the majority of computers are equipped with standard stereo output devices, Finale is designed for stereo output. Recent versions of Finale now allow the user to assign instruments a given position on the stereo spectrum, but beyond stereo, it is unable to convey spatial placement (which would also, incidentally, require the user to have a studio equipped with at least a “quad” loudspeaker setup). Composers unaware of this bias may overlook the potential of spatial arrangements easily implemented in live situations. Loudspeakers and headphones do not correspond very closely to a live performance anyway, rendering sounds in an altogether different way.
Expressiveness. Finale is particularly clumsy at making its playback sound musical. Composers, who want the program to perform a convincing accelerando, ritardando, or sudden shift in tempo, should be prepared to program every minute tempo variation into the music. Otherwise, the program is left to guess, for instance, the extent of the ritard. While recent versions of Finale include a “human playback” function, the effect still remains, unfortunately, mostly inhuman (that is, rather mechanical).
Figure 3 : “Mary had a little lamb.”
a. Desired notation.
b. Notation resulting from an excessively fine quantization setting.
c. Notation resulting from an excessively coarse quantization setting.
This can be misleading or confusing to students unfamiliar with notation as well as those who trust the tool and do not double check the computer’s notation. With other sequencing programs (such as Digital Performer) already transcribing with better results than Finale, one can expect that Finale will follow suit, and that composers will use the hyperscribe function increasingly.
Beyond these more technical issues, we can list five over-arching risks. Compromises: Educators worry that, consciously or not, the students will gravitate toward writing music that the program notates and plays back with ease, and away from what the program has difficulty handling. All of my interviewees acknowledge that they are often tempted to compromise, and that, occasionally, they surrender in the interest of time or out of sheer frustration.
Short-sightedness, sheltering, and uniformization. Composition teachers worry about something more insidious than compromises. They worry about the options, the possibilities that students will never become aware of because of the program’s assumptions. The program’s tools and menus attest to a myriad of options that inquisitive students will be keen to experiment. It is what the program doesn’t offer, which is perhaps the most dangerous.
A default Finale document implies a number of assumptions. For instance, Finale automatically assumes that each new measure is in the same key and same meter as the one which preceded it. Were the program set up to ask users to confirm the meter every time a new measure is added, users might experiment with this aspect, instead of composing a whole piece in the default 4/4 meter.
Thanks to the cut-and-paste function, a piece can be built of exactly identical building blocks (from bar to bar, and/or from section to section). (Many of my interviewees brought up the fact that Finale’s cut and paste function was ideally suited to the repetitive textures of minimalism, although they also acknowledge that the style was in full swing long before Finale’s heyday). In other words, Finale encourages a sort of uniformity that is rather antithetical to the explosion in possibilities witnessed in Western classical music just a few decades ago.
Shift of goals. For the composer, while the first live reading of a piece, the rehearsal process, and the performance can be enjoyable, these phases are often short, nerve-wracking, and sometimes quite discouraging. Performances often occur months after the composition process is completed. As these phases become less significant and less rewarding to the composer, the importance of MEPS playback grows in inverse proportion, becoming the more meaningful and rewarding event. It can be difficult for the composer to favor a distantly scheduled live performance, where vaguely-imagined and hopefully well-calculated effects will finally become tangible, over the immediate reward of the computer’s feedback. However, if a composer produces a score that, while it lends itself to compelling computer playback, overlooks legibility and playability, rehearsals are likely to reinforce the composer’s notion that human performers are inadequate. They might even encourage the composer to switch over to producing computer music exclusively.
Resistance to migration. Because Finale is a complex program that requires a lot of time to master and use fluently, veteran users may become increasingly unwilling to migrate to different programs. They might also loose sight of the fact that there are entirely different ways of composing music. 
Risk that the computer will supplant the composer. Finale has recently added new tools that auto-harmonize music, auto-arrange, and auto-orchestrate. A Finale manual describes them as follows:
"[The] Composer’s Assistant plug-ins … are specifically designed to help generate musical ideas. These tools won’t write a piece for you, but they will analyze existing music and offer suggestions for chords, rhythms and voicing. If you are looking to generate harmony for a melodic passage, Finale can offer a large number of suggestions for up to six parts with the Band-in-a-Box Auto Harmonizer plug-in. You can even create an entire jazz accompaniment for any melody with the MiBAC Jazz Rhythm Section Generator plug-in. Finally, if you are composing serial music, or a related genre, you can easily rearrange notes by inversion or retrograde with the Canonic Utilities."
Meanwhile Finale's competitor, Sibelius, proposes something very similar:
"The Arrange feature is an extremely powerful tool to create music in the ensemble of your choice. …
The Sibelius Arrange feature is able to reproduce ensemble voicings made famous by Sammy Nestico, the Count Basie Orchestra, and others."
The writer concedes that the program “will not harmonize your music, realize chord symbols, add counter lines, substitute chords, or remove the need to study those subjects further if you need to use them.” Evidently, we can only imagine that Sibelius software programmers are hard at work developing those particular plug-ins.
The effectiveness of these tools is currently limited, and they result in unimaginative arrangements. With time, however, there is no reason to believe that programmers cannot find a way of improving their product. When can we expect Finale to have an auto-compose function? As the computer makes more and more decisions, users are at risk of letting their skills atrophy. Literally speaking, however, the computer isn’t truly making creative decisions either, as those are dictated by software engineers. As such, the future for composers might be, precisely, as the masterminds behind composition software.