Pandora's pitfalls
David Hearst
Recently, while digging through the business news, I ran into this article about Pandora's earnings shortfall: Pandora's earnings miss. Those of you that caught my previous articles about digital streaming's unfulfilled promise and digital vs. analog search know that I have been disappointed in the past by Pandora and other streaming services. The article about Pandora's earnings announcement inspired me to dig deeper into Pandora and whether my criticism is justified.
First, a disclaimer: I submitted Scattershock's album, Wrong Train, to Pandora shortly after we completed it. Within a short period of time we received a rejection letter, with the message, "Thank you for your submission to Pandora's Music Genome Project. We wish we could say otherwise, but we have decided that this submission does not fit our collection needs at this time."
Having attended UCSF for my graduate studies and received a Ph.D. in Pharmaceutical Chemistry, Pandora's use of the term "Music Genome Project" has often nagged at me. The concept of mapping musical forms by a defined set of criteria and using that mapping to establish relationships between artists is intriguing. What bugged me, though, was that unlike the Human Genome Project, Pandora's project is mostly subjective, based on qualitative assessment of each artist's songs. Here I have excerpted the goals of the Human Genome Project (courtesy of the HGP website):
(The Human Genome Project's) goals were to
- identify all the approximately 20,000-25,000 genes in human DNA
- determine the sequences of the 3 billion chemical base pairs that make up human DNA
- store this information in databases
- improve tools for data analysis
- transfer related technologies to the private sector
- address the ethical, legal, and social issues (ELSI) that may arise from the project
Those goals are clear and are all-encompassing. The first bullet point clearly states the goal includes ALL of the genes in human DNA. Imagine if the HGP had instead opted to catalog only some of the genes? Would the resulting data be anywhere near as scientifically useful? Absolutely not. Partial data in this case would lead to skewed conclusions and lack of confidence.
So then, what does that mean for Pandora? Well, they've opted to call their efforts the "Music Genome Project", with an obvious tip of the hat to the HGP. But, we know that they chose to leave Scattershock out of their database. And that's where the article about Pandora's earnings left me with this nagging question. Is there a reason that Pandora isn't simply blowing everyone's mind and making money hand over foot? There are clearly some fundamental business reasons why the streaming business may not prove successful for anyone. I'm not here to dig deeply into that question. However, I am interested in the question, "Why are some digital music services more valuable to me than others."
As a DJ (at KGLT) and a guy with a huge appetite for new, heavy music, I'm looking for ways to discover the wide variety of heavy bands all around the world. Pandora seems like it'd be a powerful tool for that but I wanted to understand how helpful it really is. I decided to assess a number of the music content and streaming services to see which ones fared best.
I decided to ask a fairly simple question, then collect appropriate data and analyze the results. I took the playlists from two of my recent KGLT shows and identified all the bands that I'd played. Then I asked, "How many of these bands could I have discovered using each of the music services I surveyed?". I have included the raw data below, but here's the executive summary:
| Total Bands | Pandora | iTunes | Last.fm | Spotify | MOG |
|---|---|---|---|---|---|
| 74 | 56 | 74 | 74 | 64 | 72 |
| 76% | 100% | 100% | 97% | 97% |
And that clearly demonstrates why Pandora has never worked well for me and Last.fm has. If I used Pandora, I wouldn't be able to play one quarter of the bands I play on my show. This data is for only two of my shows, meaning that over the last 8 months that could amount to over 100 bands simply left out of my show. Pandora would sweep them under the rug and pretend they just don't exist. You say, "David, but those bands are all obscure ones like Scattershock, no one knows about them and no one cares." If you look at the detailed data, you will see that Pandora's oversight includes Amaranthe, Bloodbound, Down From Up and The Slot. Sorry folks, these are serious bands with multiple album releases and videos. You might think that these bands somehow haven't submitted their works to Pandora, and that's a valid point. Maybe all these bands are boycotting Pandora for some reason. That's absolutely possible but, that just goes back to my point.
Pandora's collection is biased. It omits major name bands, many of whom have won music awards around the world. It also omits small, unestablished bands, some of whom will eventually become household words. And if you want to be an early adopter of such acts, you are much better off with any of the other services I surveyed, and especially iTunes and Last.fm.
So, tell me, what's your favorite music service? And, if any of you have invested in Pandora, what are your thoughts about your investment and the bands that aren't available via the service?