All the small things

Recently, I started extensively using my mp3 player from a well known company in Cupertino, which happens to love the “i” in almost all their products. I have to admit, I am actually pretty much of fan usually, but not this time. After so many years of experience and millions of sold products, I couldn’t believe my actual disappointment about the implementation of the shuffle mechanism.

But why? What can be so bad on a shuffle mode? Well first of all, what is the scenario in which you use shuffle? Usually you have your play lists with a subset of all your songs selected after certain criteria. In our current age, most of these criteria are captured in meta data. So iTunes for instance already supports some sort of intelligent play list, where you can sort for various meta data fields. Nothing sophisticated, pretty straight forward. So, playing shuffle in these lists is usually no big deal. You selected those to basically fit your mood. When you decide to play them they just fit or you select another one.

Last year, when I was driving a lot, I didn’t have the time and the mood to define play lists all the time, so (like many of my friends) I just used the shuffle on all of my songs. Here it really gets bad. The standard shuffle function just plays randomly songs (sounds odd, I know, but even in a shuffled play list, I’d like the have some sort of plot) and completely ignores the input I provide by skipping certain songs. More in contrast, it feels like the shuffle mode is trying to make up for the mist artists and places them more often in the list, although there are just 10 songs within 4000 for instance. I intentionally call it “feels”, because I don’t have hard facts, so it is most likely just my annoyance talking, but hey, that’s how it is – pure user experience.

What I am actually looking for is an “intelligent Shuffle” or “iShuffle” in short. Why can’t they use meta information all the web analysts are so eager to harvest for something useful. Why can’t they use information gathered by sites like Last.fm, seeqpod or even FaceBook and define some sort of listening graphs with close neighbors? You can easily figure out, which kind of music matches each other and find conceptual friends based on listening habits.

You don’t even need to go that far by looking into social networks. In many cases it is enough to just have a look at the existing play lists. Each play list represents some sort of similar taste. So don’t jump between play list after each song. Jump within a play list several times and find connectors like same songs, similar artists, same genre, etc. to make the switch between the list as smooth as possible. Create a, let’s call it mood protocol, which identifies the current “mood” of the listener by the songs recently played and find possible neighbors.

Another big hint, as mentioned before is the skip button. If I press skip, something isn’t right with that song. I have to take that into account. It is not necessarily that I don’t like the song at all, it might just not fit into the mood, so please don’t ban it right away. However, if I press skip on a second one in a short period of time, the relation between the two songs might be the key to know what NOT to play next in order to prevent a skip from happening very soon. Easy to accomplish such a similarity analysis are checks for similar bands, genre, play lists or even tags. More sophisticated checks assume some knowledge about social networks as mentioned above.

The bottom line for me is that in this area it feels like nothing has really happened so far. An algorithm as described, implemented as a simple AI, would immediately boost the user experience. I guess, the market just doesn’t pay well enough for happy customers. Perhaps people just don’t know what they can demand, that it is possible to have something way better than the current technology available. But hey, what are blogs for! Maybe this helps to push people thinking about what they are missing or address others to implement such an iShuffle mechanism in the next generation of consumer electronics. Preferably my favorite company with the “i” in their portfolio :0)

Best Regards,
Mirko
P.S.: If you get rich with it, please don’t forget about me

Name (required)

Mail (will not be published) (required)

Website

Powered by WP Hashcash