IMSLP Forums

Hi. As a (former) administrator of IMSLP, I'm full of questions after a week of discussions, proposals, and emotions. (Please excuse me in advance for the bad english).

Many new names are building their utopic IMSLP on the forum. I wonder we will have people enough to implement the dreams of many users. Before the closure, we were alltogether, after two years being on-line and 60000 downloads per day, only 10 trustable admins and copyright reviewers.
We had to work daily to keep up with the never stopping train of submissions. In the last weeks of IMSLP, we couldn't cope: there was constantly a pool of 100 to 200 unreviewed submissions. I doubt of many people on this forum if they are aware how much work is needed every day of the year. One submisson took, with experience, up to 10 min to find out the simple copyright premissions we had, and to inform the new or the stubborn submitter of his errors. There were circa 100 submissions per day. Sure, after this event of world wide web attention, there will be more people volunteering to help. Will they be there six months later?

After being a week on the forum, people start discovering trivial things about IMSLP and copyright. For example, copyrightship of the editor of a score is something what was explained on the first lines of our "public domain" page, one of the pages most linked to on the website, but only discovered today by some users. Why?

What I also wondered during the last two years is why there were, in contrast to the immense popularity of the website, so few people motivated to make their submissions conform to our guidelines, and why there were so few people motivated to make real valuable contributions, instead of dumping their score archive on IMSLP. There were daily duplicate submissions, copyrighted submissions, submissions without any relevant copyright information,and a gigantic load of formatting errors.

To make sure a better quality control in the future, I'm interested in the reasons behind this lack of interest. Also, the more work is done by submitters, the less is needed by administrators and reviewers.

The most obvious reason would be that we did not provide enough information. Or was there too much information? Or wasn't it properly translated? Or you just didn't care about anything else than downloading or uploading your score? We tried to inform most users on their talk page of their errors, in directed them to the right pages, in some cases with success, in other cases without any reaction.

For reference, you'll find the public domain page of the former IMSLP here.

You need another option - "I found it to be a helpful and useful guide, and printed it out to refer to while scanning and uploading."

Or "I tried to follow it as best I could, but sometimes needed a bit of extra help."

Aldona

Maybe, when a user tries to submit a piece, when clicking the upload button, it directs first to the article, with a link right at the bottom to continue with the upload. much like the Eula on software installations.

What I also wondered during the last two years is why there were, in contrast to the immense popularity of the website, so few people motivated to make their submissions conform to our guidelines, and why there were so few people motivated to make real valuable contributions, instead of dumping their score archive on IMSLP. There were daily duplicate submissions, copyrighted submissions, submissions without any relevant copyright information,and a gigantic load of formatting errors.

I tried to make sure my submissions always met the guidelines (and I hope I was making valuable contributions), but I agree that there were tons of submissions which obviously did not meet the guidelines and I always saw the admins' deletes in the recent changes. I understand the pain that the admins went through daily. I've thought of a few possible reasons I'll throw out there:
* Anonymity hides responsibility: 'Someone else will do it.' Things just magically happen on the internet.
* Some people saw this site as something similar in purpose to p2p networks, but for sharing scores legal or not.
* Quantity over quality also seems pervasive on the internet.

I hate to say this because it would cause legitimate submitters more work, but perhaps it needs to be more difficult to submit a score and it needs to be easier to remove a score for the admins.

A list of questions which must be answered could deter some dubious submissions (rough idea):

* Have you read the submission guidelines? (yes/no)
* Why do you believe this score/book is legal to copy?
* Why do you believe this score conforms with the submission guidelines?
* On a scale of 1-10 what is the quality of this score? (not quality of the music itself)
* etc... (The questions can be worked out later)

If a submitter fails to answer any of the questions satisfactorily an admin/copyright reviewer can delete it all with one button push. If the questions are answered well, then a normal copyright review can happen. This will be more work up front to implement but it might work.

I can certainly relate to Peter's salient points here. We deleted Lucian Garban's reduction of the Ravel Piano Concerto (still copyright in Canada) no less than four times! BTW, Aldona and Horndude, your submissions weren't the problem he's referring to here - not for me at any rate.

When IMSLP goes up again, we really need to establish a "buffer" area that submitted items can go into while they are reviewed an brought into conformity with IMSLP's guidelines - not to mention copyright compliance - before they actually appear on the site. I suppose that making publisher a required field for scanned scores would help weed out the dubious cases also, but with reprint houses, CDSM and others stripping out all of the original publication info, it may well be too limiting.

Horndude raises another interesting point here - the issue of scan quality. Should IMSLP set some standards for this? The quality of items ranged from borderline illegible low-resolution scans to wonderful high-resolution color scans. If a given item is truly rare - like a scan of a composer's manuscript or of rare early printed scores - one could make the case that having a low-res file is better than having nothing. On the other hand, a low-res file of something as common as Breitkopf score of a Beethoven Symphony isn't of much use to anyone as far as I can tell.

BTW, Aldona and Horndude, your submissions weren't the problem he's referring to here - not for me at any rate.

Thank you - I give full credit to you good people at IMSLP for helping me get up to speed from completely clueless newbie to hopelessly addicted scanner and uploader.

A list of questions which must be answered could deter some dubious submissions (rough idea):

* Have you read the submission guidelines? (yes/no)
* Why do you believe this score/book is legal to copy?
* Why do you believe this score conforms with the submission guidelines?
* On a scale of 1-10 what is the quality of this score? (not quality of the music itself)
* etc... (The questions can be worked out later)

I'm only about half way through the Schubert-Lieder, easily 300 or so to go...so I can just imagine going throught this whole procedure with each one of them...

Maybe after a user has a proven record of legal and acceptable submissions (e.g., after he/she has successfully submitted 10 or 20 or whatever number of scores of sufficient quality and conforming to legal guidelines), an exemption can be granted from some or all of the above questions.

* On a scale of 1-10 what is the quality of this score? (not quality of the music itself)

Hahaha...If you ask the one who is submitting the score, the quality will always be overestimated. A bit like asking them to look in the mirror and tell you how beautiful they are on a scale of 1-10.

Aldona

A bit like asking them to look in the mirror and tell you how beautiful they are on a scale of 1-10.

I'm ugly.

ArcticWind7 wrote:

A bit like asking them to look in the mirror and tell you how beautiful they are on a scale of 1-10.

I'm ugly.

Or maybe a better analogy would be...asking a mother how beautiful her baby is, on a scale of 1-10.

Aldona

Much better.

I think what Aldona means is that the rating of a music scan from 1-10 is very subjective, and the submitter will almost always biased towards a better quality rather than a lower one. This is way I think it is better for a scan to be rated by as many people as possible. A fair score or rating could be achieved if rated by a number of people rather than the submitter alone.

Or maybe a better analogy would be...asking a mother how beautiful her baby is, on a scale of 1-10.

True. I just added that because there was already an under-used rating system in place. It would be nice to get some value out of it. Perhaps there are a couple objective measures of scan quality that could be used instead (i.e. dots-per-inch, monochrome/color-depth, # crooked pages, # black edges, etc.)

Also, true that it would get annoying to have to answer questions every time. It was just an idea I was throwing out there. Maybe every submitter should undergo the questioning until they are known to be trusted. Or to go further have every submitter pass something similar to the copyright test .

It's a fine line which wikis walk between encouraging and accepting all submissions and ensuring quality and avoiding vandalism. In this case we also have to take into account the work levied on the admins/copyright reviewers. Most of the work should be on the submitter and the admins should just make sure standards are being followed.

other criteria I always used for rating were:
- overall: readability on screen (<6/10) / printability (>7/10)
- horizontal alignment, page centering
- appearance of compression artifacts (in case of jpeg)

In addition to the poll options, I would say, putting to detailed copyright information on your site for jurisdictions outside your own may be construed to the effect that you are actually directing at those jurisdictions, and thus be an implicit acceptance of the laws of those jurisdictions. For that reason, I would, apart from a general remark that copyright laws in other jurisdictions differ, refrain from saying much more. Note that this doesn't stop you from providing information on publication and death dates, which may be helpful for others to establish the copyright status in their jurisdiction.

Furthermore, I suggest bifurcating the IMSLP into a Canada and a US based organization, both independent legal entities (or individuals), and handle all uploads via the US side. This way, the US IMSLP can accept user submitted materials under the safe harbor provisions of the DMCA (similar to the way YouTube and Archive.org are). The US organization can load everything PD in the US, and the Canada organization those scores that are not in the PD in the US, but are PD in Canada.

Project Gutenberg follows a different approach: all copyright checks (called copyright clearances) are handled through just two people, and having a clearance code is required to upload a text -- which is handled, by a team of just five or six dedicated posters (called white washers, based on an inside joke). You could also set-up such a system, where people upload materials, with a range of fields filled in to indicate that they believe the work is PD. (list of involved people and their death dates, plus date of publication of the material at hand, all being more than 50 years ago for the Canadian situation)

jhellingman wrote:Furthermore, I suggest bifurcating the IMSLP into a Canada and a US based organization, both independent legal entities (or individuals), and handle all uploads via the US side. This way, the US IMSLP can accept user submitted materials under the safe harbor provisions of the DMCA (similar to the way YouTube and Archive.org are). The US organization can load everything PD in the US, and the Canada organization those scores that are not in the PD in the US, but are PD in Canada.

I disagree, somewhat. I think bifurcating is an excellent idea, but I think that both entities should have as much material as each can, and share what they can between them, mostly because that way, if one goes down, we don't lose half (or more?) of the archive.

emeraldimp wrote:I disagree, somewhat. I think bifurcating is an excellent idea, but I think that both entities should have as much material as each can, and share what they can between them, mostly because that way, if one goes down, we don't lose half (or more?) of the archive.

This could be solved by having a liberal mirroring policy, and allowing a set of selected entities to mirror the entire collection on a daily base, similar to the many Project Gutenberg mirrors on the web, a reborn IMSLP should not have single points of failure.