Interesting article, one important aspect it left out however is the matter of time needed to create a certain end result. By remixing (and to some extend also general collaboration) one can create a final outcome (regardless if it is more code or art heavy) in a much shorter time. This is usefull for prototyping, but is also a genuine quality of it's own.

If done all from scratch the final product might be ultimately superior, but if it is never finished (or not even started due to realistic time planning), I much rather have the remix that can be later polished.

Another thing it left out is entry barrier and similar issues. Lets have a look at SP mods for example, by remixing existing art assets one skilled story-teller can create an awesome experience which would have never even been possible to create by that person (who is not a great coder or artist), would it not be for the assets and code available for remixing.

So all in all I would say these two factors might in fact be the most important ones, and therefore the study presented in the linked article of rather little value ;)

I don't think that borrowing someone else's graphics counts as remixing.  That's just reusing.  I'm also skeptical about the relevance of the platform they chose to study.

A less then scientific study using less then scientific methods can come to any concolusion.

While I must say that "remixing" when it comes to code is more beneficial then it is for creative works, the benifit for creative works is still there just typically not in the same method. 

The highest quality creative works typically come from when someone see's piece of art x and wants to create piece y and z to go with it, not in a pure modification of the original.

As someone who "barrowed" code for my project, but intends to replace most of the artwork, I can understand the feeling.  After you look at a picture for a while, you get sick of it.  Then again, the project I barrowed code from did not have artwork of particularly good quality to begin with, that and the fact we choose to use a different perspective (45 degree isometric instead of the orginal slanted top down perspective) and its obvious why I had less incentive to directly reuse artwork.

Although I would welcome more artists on the project willing to draw in the proper style, I'm not sitting around on every message board trying to recruit them.  After alli, it still is much easer to have a project with place holder art rather than placeholder code.

Firstly I dismiss the concept that there are works which aren't remixes, no man is an island.

Secondly I'd agree with Julius,  the *average* quality is a poor indication, remixing existing art assets is easier than just remixing ideas and styles. On average there is less effort put in, so one would think on average the quality could be considered lower. I wonder what would happen if you factored in the time spent on a work? I'd imagine remixes would win on a time basis, since with a given hour of work, a given individual will be able to produce more work if they can remix existing works as their basis.

Finally, the community used as a basis for their study is very skewed towards remixes being of a low quality. I would suggest that if the authors did their study on OGA, they'd find remixes almost always improving upon the original in some way.  If you set your community up to allow people to spam low effort remixes, you'll see a lot of cruft, whereas if you encourage people to incrementally improve upon works, you'll find the quality steadily rising with each remix.