The Keyboard Philharmonic overview mentions that one focus of the project is to be “a vehicle for recruiting people and structuring their work for useful results”. There are reasons why this focus is important. 

Search the web for projects which aim to transcribe music scores into digital score form. Get beyond the first layer, of thriving projects like the Petrucci Music Library / Internet Music Score Library Project (IMSLP), and the Choral Public Domain Library (CPDL), which are concentrating on PDF files readable by humans only. Dig further, and there are a surprising number of projects to be found, which created digital scores, notation files readable by software as well as humans. Sadly, many of these projects are like ghost towns. They are evidence that bold pioneers were once hard at work, but also that the energy has gone. Websites are still running, but no few or no scores have been released recently; the forums have questions but not answers; and messages to the project leaders go unanswered. I respect the people who got these projects started. I admire the passion and hard work that got their projects to the state they now are in. But one person’s hard work is not enough. The project needs enough volunteers to sustain it.

What I hope the Keyboard Philharmonic will do better is to be people-centred. We need to welcome new arrivals. We need to find useful, simple, small tasks for them to do, and we need them to see the benefit their small effort has provided. We need to reward the small tasks, and have a gradual path to larger and larger tasks. We need new arrivals to replace those taking a break. We need a good mixture of deeply-invested participants making skilled, complex contributions, and peripheral helpers biting off small tasks.

Also, it matters what those volunteers do. Among the ghost towns of digital score transcription projects, there are a variety of workflows, a variety of formats for digital score, and a variety of editorial standards for their work. Try to use their scores, and it becomes clear that some work better than other. For example, I recently came across a  transcribed digital score for a Bach motet which I was singing. Reading the score, I found errors in how the words were placed on the notes. The errors made that score less helpful. A better proofreading cycle in the workflow that produced that score would have made the score better. There was no easy way to report the errors. So, no way to warn other readers, and no assurance that someone could correct the errors and update the score. This was frustrating. Furthermore, the errors in that score reduced my trust in other transcribed scores from that site. Poor proofreading led to a poor score; poor workflow gave me no easy way to make an incremental contribution to improving that score.

Successful proofreading requires a way of comparing the digital score to some reference.  The obvious initial reference is a scanned image of a published, public domain score.  This leads to an editorial rule, that the first edition of a transcribed digital score for a work should match a specific published source, to the extent required to make proofreading tractable. It should have the same system breaks and page breaks. It should reproduce the same layout, and word assignment to notes. Ideally, it should make a novice volunteer able to proofread just a page or a system of the score, not the whole thing. This rule says that, for this first digital score edition, good proofreading is more important than is improved layout. This is analogous to Wikipedia’s Verifiability policy, especially their zestier slogan, “Verifiability, not truth“.

There is a lot of scope for the Keyboard Philharmonic to facilitate creation of improved editions, with more useful layout, page break, and content than their old print predecessors. If software tools can compare, note-by-note and ignoring layout differences, the digital score for the improved edition with a digital score matching the original print edition, this provides confidence that the notes are right in the new edition. However, different musicians will disagree on what changes to a score constitute improvements. Requirements differ, and there is no accounting for taste. I suspect that the Keyboard Philharmonic would get bogged down in these disagreements.

So, the Keyboard Philharmonic can make a deeper contribution by focussing volunteer efforts on making a first edition of a digital score — objectively comparable to a reference, and carefully proofread — rather than making “better” editions. This is another example of “structuring [volunteer] work for useful results”.

I am not daunted by the ghost towns of music transcription. I am firmly convinced that the volunteers are out there. The large contributor base of the Wikipedias and MusicBrainz‘s of the world are proof of this. If the Keyboard Philharmonic can succeed in enticing those volunteers to turn their energies to transcribing music scores, rather than to other options they might have for their time and energy, then it will make a valuable contribution to the cause of bringing our music scores into the 21st century.

No Comments »