Continuing work on Active Archives, and getting back to the YouTube API, I noticed a change. I’m working on code that maps data from the YouTube API into RDF, and noticed that part of my code wasn’t working anymore; the code that looked for “tags” was now coming up empty. A while back I posted this entry about uploading a clip to YouTube, and detailing the process. When you enter a video on YouTube, the input forms include a place for “Tags”.

However, it would seem this information is no longer shown publically (All the links below are ones I entered in the Video description after I learned that links could only be entered as raw URLs).

As it turns out, the tags have been shifted somewhere else in the “data feed” (which is why my code that used to retrieve “tags” was coming up empty). They are now represented in the same way as the “Category”. Given that this information is no longer shown publically, one can wonder if it will remain in the API (in any case, it’s lack of visibility certainly discourages it’s use). At the same time, I was looking at the data feed of another clip, which I didn’t upload myself:

Here, a relatively new feature of YouTube is visible. Based on the music in the clip, the public is offered links to the name of the “Artist”, and are shown the name of the song with a link to purchase it online. This is part of YouTube’s automatic content based indexing that auto-detects “known” music and is part of YouTube’s strategy to allow the use of copyrighted material by allowing copyright owners to share in any advertising profits, and to benefit from the “buy” links, that frame user provided content.

Tellingly, this information (the artist and song names) is not provided in the public data API. Nor is any interface provided to correct / alter / remove this labelling. (A question on stackoverflow about this limitation.)

One interesting bit of info that is in the data feed (but not shown publically) is the following:

I would imagine the restriction of this work from being viewed in Germany might be related to copyright as well, though this reasoning is not given.

What I think is telling in all of this is it shows a movement away from user-curated / user-linked data, and towards “automatic” tagging that in this case relies on proprietary algorithms and business agreements made behind closed doors. YouTube’s editorial-lopsidedness is here visible as users are encouraged to provide (creative) content, but not asked to participate in how the content is potentially linked to other content, and thus contextualized, as one falls back to YouTube’s own rather obscure notion of “related” clips.

(the AA video wiki is broken as a result of the same change)
So who’s integrity is in error?!