I've been listening to Pandora a lot lately. This is a service, based on the Music Genome Project, that recommends music you might like based on the artists you choose. So far, it has worked pretty well for me - most of what it has recommended has been appealing, if not mind-blowing. However, a friend of mine insists it does not work for her. She says it recommends stuff she hates and she has to spend her time "responding to its requests for up/down votes on the recommended stuff. Who wants to do that, I just want to listen to music."
I've also been to more movies in the last week than I usually get to see. Here's what I saw over the course of 7 evenings - "Sex in the City," "Mamma Mia!," and "Roman de Gare." The first two are simply plotless. The last had more plot twists than I could count or follow even with the subtitles. I enjoyed all three. I've linked to Anthony Lane's New Yorker reviews of each of the three movies, even though I steadfastly refused to read his insights into either "Sex" or "Mamma Mia!" before seeing the movies - do you really need to consult a film critic about a TV show or Meryl Streep singing ABBA? Each night's selection was chosen by me and some close friends. I'm not a marketing expert, but why we chose what we chose and why we enjoyed what we saw would be tough to plot against phenotypes.
All of which leads to me to wonder about philanthropy, recommendations/advisors, and where we get the information that guides our choices. I know, even as I write this, that someone out there is either building or thinking about building an "Amazon-style" recommendation service that draws from the Guidestar database - "People who gave to the Red Cross also gave to WritersCorps." How do I know this? First, it is a patently (joke intended) obvious idea. Second, two different organizations have told me they are thinking about it.
As my made-up example is intended to imply, giving to the Red Cross and giving to WritersCorp could easily emerge from any algorithmic approach to recommendations - but what would such a recommendation mean? Do you base such algorithms on where money goes (the Amazon approach)? Or would you, as Pandora does, attempt to deconstruct nonprofit organizations into types and characteristics and then make recommendations on patterns and similarities? Or is the whole idea absurd, given that giving is both passion-driven and rational in ways that music, movies, and pop culture are simply not? Can you tell something about someone by where they give their money? Can you predict where someone will give by mining data on where they live, what they read, what kind of ice cream they buy, and whether or not they vote? (This kind of thinking seems to be what drives much of the direct mail marketing that charities do.)
What do you need to know about someone in order to help her/him make better giving decisions? What do you need to know about the organizations to which s/he might give? Here's an observation - it is now more important than ever before that donors understand how these recommendations or advisors work. Why? Because the marketplace of organizations offering a 'view' into that database of nonprofits just keeps growing. From one massive list of organizations (1.5 million plus in the USA) a donor can find recommendations and advice and giving opportunities sliced up several ways - the GiveWell way, the NetworkForGood way, the community foundation way, the Federation way, the Charity Navigator way, the GuideStar way, and, of course, the United Way....and so on.* How do those choices get made? What criteria get used? What do the recommenders know about you, the donor? What do they know about the organizations? How do they generate their recommendations - is it algorithmic? Phenotopic? Personal? And what should they know, about the donors, the organizations, the goals and the resources? The philanthropic marketplace, as evidenced by the continuing growth of sites like GlobalGiving and Kiva, the interest in conversations about market data, and the creation of new discussion venues and conference sites, continues to grow. Very soon, I think, we will be rating the raters and recommending the recommenders.
*Disclosure: I used to be on staff at a community foundation and still work with as many of them as I can. I am on the Board of GiveWell. I know the founders of GlobalGiving and I am one of two moderators of the discussion on data and social capital markets. I have subscribed to The New Yorker since 1985 (and read my parents' copies before that), but I am always at least three months behind in reading it.
Pandora is also doing a poster design contest with Global Giving. Sounds pretty cool, if I could draw!
I've found that the UK-based last.fm delivers a better selection of music and gives you more control of the music. Once you "train" the software by inputting the music you like, there are only a few occasions where you need to upgrade/downgrade songs.
Artificial Intelligence and algorithms will soon micro-manage most of the world in unimaginable ways, impossible for the human brain. Have you heard about the $1,000,000 X-prize to create an algorithm for Netflix that is only about 20% better than the one the use already? There's huge money in this industry.
As to giving, I don't trust that my money will get to the people intended. How many times have we heard about misappropriated relief money?
Isn't this the question that is driving all these social media sites?
Music sites can be very effective for some people and not nearly as much for others. While Last.fm and Pandora work great for me, I too know others who don't like either services' recommendations.
I think that different people (or personalities) seek out different types of recommendations.
I'd say there are about three types of recommendations that people look for, Expert (a professional reviewer), Social ("hey, my buddy really liked this"), and Managed (radio plays and automated recommendations). The best resources will be the ones that offer all three.
Post a Comment