IMSLP Forums

I know the question of rating scores according to difficulty has been raised elsewhere. As a novice searching for scores for his daughters string trio, it is quite daunting. Even a basic system, rating the required level of musicianship would be immensely useful. Something like a 1-5 scale, rating from novice to accomplished musician. The rating could be set by the uploader and/or
users (in the same way youtube users rate videos) who feel they are familiar enough with a score to rate its difficulty. That would eliminate the need for a few individuals to do that massive job.

It would be nice to have scores rated by difficulty and this added to the category walker. Users could then find a piece within their skill range. Though I am not sure how you would rate the scores.

Does anyone have any suggestions?

Merged the two topics - somehow that seems more efficient
I imagine that it wouldn't be difficult from a technical standpoint to implement a system of difficulty ratings. The problem, as pretty much everyone who has discussed this before has pointed out, is that 'difficulty' is extremely subjective; it varies between skill levels, of course, but also can even among those with similar prowess. In fact, it's almost hopelessly vague if you want to figure out something that will satisfy the majority of users.
What about a system where one could tag works for specific technical challenges (or the absence thereof)? For instance, one could tag solo violin music based on, say, whether it could be played entirely in first position, or entirely in first and third, etc. That would be much less subjective, since it would be based on technical fact rather than opinion; I guess the main problem would be finding people to implement it. (Yes, we've successfully undertaken large-scale team efforts before, but we still haven't even finished the categorization project started last year, and I guess most regular contributors already have things to occupy them. The issue with having the 'general public' implement it gradually - think file ratings - would be that one would need to tag it very specifically for what was in it, and a lot of time would doubtless be devoted to cleaning up the entries of the less technically-savvy members of our clientele.)
Thoughts?

Totally agree Kenny – the difficulty of a piece is entirely subjective unless you use some precise measure like the one you suggested related to the technicalities of violin method – and that only applies to that one instrument (or family of instruments). You need experts for each instrument (and combination of instruments) to make similar evaluations of difficulty of other works.

As for whether this can be implemented across 34,000 works: how many scores have file ratings? (i.e. quality rating from 0 to 10 which appears directly beside the copyright tag)

Enough said. If it can be implemented it probably should be – but we should be under no illusion that it could be made complete, or consistent across the whole site without going to the trouble of arranging for a project team just like the work tagging. (Neither am I convinced of the priority for this addition yet.)

Regards, PML

A good example of this is Debussy's Syrinx. Technically it is not that difficult -- in fact, one of the exam boards at one point put it on the Grade 6 list (out of 8 Grades), and later moved it to Grade 8, because it isn't really "complete" unless the interpretation is mature.

Not easy to tag, and this will not be the only piece with these issues.

A place to star difficulty rating would be:

First ,for Merton project items- to use their system which appears in their printed catalogue.

Second, for works listed in grade exam sylllabi- the grade with a link to the organisation's web site so that we can see other works of the same grade level.

I have been having this problem too. I started teaching a friend piano, and I'm driving myself crazy trying to find level 0 music, to the point that I'm just transcribing Weezer for her, with the occasional historical piece thrown in.

Thoughts:
1. Any single 1-value rating system is inherently flawed, as more than one value determines difficulty.
2. Any rating system should take into account, as you mentioned in the case of Debussy, both technical difficulty and, shall we say "opined" difficulty.
3. I certainly don't have infinite resources and time to do this, as I'm sure nobody has, so anything preexisting should be incorporated.

With those in mind, I think we should have a multi-faceted rating system. Breakdown:
A. Popular vote:
Unscientific as it may seem, I think it's the foundation of any rating system to see what the aggregate opinion is.
B. Pre-existing Rating:
If the piece is part of a certain theory level, or has been rated in some system already, all that should be listed. For further explanation, a link to a wikipedia article or website explaining that rating system could probably be had.
C. Comprehensive:
Here's the kicker, and this one would be the most work: a system for measuring the basic functions of difficulty, much like Pandora's Music Genome Project, except far less complex. All my ideas are based around the piano and the theremin, but I won't bore you with theremin method. Here's a few categories, as applied to piano:

Tonality: It's always harder to play a piece with accidentals, especially for beginners. So
1 - no accidentals
2 - few accidentals, preferably all key-related (fourth of the fourth, passing notes, etc.)
3 - many accidentals, but not more than one at a time and none consecutive.
4 - Consecutive accidentals
5 - simultaneous accidentals
Fingering / Chord Theory
1 - No Chords
2 - Intervals in one hand
3 - Intervals in both hands / Trichords in one hand
4 - Trichords - Both Hands / Tetrachords - One Hand
5 - Tetrachords - Both Hands
Reach
1 - 5th
2 - 6th - 7th
3 - 8th
4 - 9th - 10th
5 - 11th and above
Symmetry
1 - hands separate or repeated
2 - melody in one hand, backing chords in the other (including alberti/arpeggios)
3 - counterpoint
4 - atonal counterpoint
5 - I'm trying, but I don't know how to finish this one.
Key - I am lukewarm on this one, but we could just divide the circle of fifths into ranks something like
1 - C / F / G
2 - D / Bb
3 - A / Eb
4 - E / Ab
5 - B-Cb / C#-Db / Gb-F#
...Actually, maybe B should go further down. Either way, this category shouldn't be weighted as heavily.

Anyways, all categories would be weighted and averaged somehow to reach a single value, which could be broken down into component parts. If it seems like I'm neurotic, this is actually just an adaptation of a movie rating system I've been working on for a few months and... never mind, it still sounds like I'm a basket case.

Here's a few test runs:
Chopin - Prelude in E minor
T: 2 - counting those as passing notes
F: 3 - I don't know if half numbers should be counted, but there are so few tetrachords (and I wouldn't count the end) that I'll give it a three.
R: 2 - not counting the octaves at the beginning or the end; so few and are good opportunities for someone looking for a 2 to work up to a 3.
S: 3 - Three measures of counterpoint (?) This raises questions about half numbers again.
K: 1 - E minor, no questions here.
Average (unweighted): 2.2
Thoughts: This makes me realize that timing should probably be factored in here. That 2/3 split at measure 18 is a definite hurdle. In more complex pieces (eg. Charlie Brown Theme) that sort of rhythmic modulation is crippling to the unskilled.

Schubert - Ave Maria
T: 2 - This was the part of the rating system was most solid on twenty minutes ago. Now, I'm not so sure.
F: 2 - There are no more than two notes played at any given time.
R: 3 - Now I'm thinking this is too strongly weighted.
S: 1 - The definition of a simply bass.
K: 2 - Bb major
Average (unweighted): 2
Thoughts: I feel tempo should be a factor, because that is one of the things that makes this piece so easy. Also, somehow pedal work will need to be stressed.

I'm out of time for now, but obviously I'm not done. One problem will be to create different systems for different instruments. Obviously, I have no way of writing one up for violin, as the total experience I have in that area is teaching a viola player how to play Twist and Shout in summer camp when I was fourteen.

Feedback? Comments? Condemnations?

Ted Out.

I would like to question a couple of points:

1. "A. Popular vote"

I'll be frank. It's a good idea, but it's not one that I see getting done. The categorization project, as Kenny mentioned, has been ongoing for over a year and it's still not finished. To just do a few files using this system would be a lot of work, and not one that (I suspect) many people would do. In addition, this would have to be for a limited number of instruments according to what the IMSLP members actually play or have knowledge of. For example, I would never be able to tag flute music because I know nothing about the flute. This poses a problem for instruments that no member on IMSLP knows about. I guess this is a problem in general, and can probably be overlooked. My point remains, however - this is a cumbersome system.

2. I disagree with the key system to an extent. I really like the idea, but C major usually tends to be a nightmare for pianists - especially when playing fast passages. It doesn't mold well to the hand and leaves a lot of room for error. Keys like B major, on the other hand, are very easy for pianists - despite being further up the circle of fifths.

Overall it's a nice thought, but I think it's impractical and highly complicated.

Cheers!