CURATORIAL FILTERING (the next generation in algorithms)


Algorithmic recommendations are one of the rages of our times! But for better or for worse (and it's the latter) ALL the methodologies encompassed in these algorithms have ONE thing in common: They depend upon what you've heard before. They build upon similarities. They can't step out of the paradigms they're given.

This is true even of collaborative filtering, a jargony term for using the fact that if somebody else likes what you like, then you might like something they like but you don't know about. The bet is that your taste will be similar to that somebody who likes what you like. You are two peas in a pod.

But there's another, so far unexplored  way to approach this:

And this is our equally jargony (it won't make a move if it ain't got that groove!) "curatorial filtering".

Which is based on the fact that somebody who creates music (or any other kind of art) that sends us...might like things that we will also like, even if these things are totally outside of our knowledge/experience. Finally, relevant suggestions that step beyond the similarity horizon!!!

I love Herbie Hancock! Herbie knows WAAAAAAAAAAAAY more than I do about music and knows and knows of lots of musicians that I've never heard of. So there's a STUPENDOUS chance that musicians he curates will be interesting to me! Believe it YES my brothers and sisters!

And there's a very strong chance that curations of musicians Herbie curates will be interesting to me as well. And on down the various lines for some ways...and people and places... To new to me musical (or any kind of art) worlds.

Now, if Herbie wants he can curate Irish concertina players. Or English landscape painters. But perhaps we will give more weight in our algorithm to his curations of players of jazz, and so on through generations of recommendations, allowing us a well-calculated chance to surface for a seeker artists and artistic works they might not come across manually.

The everybody can hand-curate artistic creators system  is here. We're working on  automatic curatorial filtering for people (artists of all kinds) and music and other artistic works.




In order to handle this important facet in a non-intrusive way it will work like this:


You're on the page of somebody curating, and you click on a link taking you to the page/node of a recommended person.


That person will probably be recommended for several categories. Clicking on any of those category buttons will take one to the recommended person's page for that particular category, and it is here that commentary can be made.


The commentary of the person doing the recommending will appear at the top of the comments for all the categories in question, and this commentary will be the same irrespective of the category in which it was typed.