Music interest software, twitter as a protocol and #myownlastfm

13 Jan. 2010

I need 300 cc's of stats STAT!

I am a sucker for statistics. I've completely fallen for their ability to describe what has happened. And it's interesting trying to use them to predict what will happen. For this reason, I love tools like Last.fm. And this is why I'm bummed when tools like Last.fm aren't as popular as tools like Pandora.

I am grouping them in two separate ways because Last.fm group attempts to use statistical correlations in listening habits and the Pandora attempts to use algorithms and analysis done by musicologists. Care to guess which one is winning actually winning the market over? I would bet money it's Pandora. Just be staying out of the topic and waiting for people to bring these things up, I have heard more Pandora than anything else. On top of that, there is reason to believe CBS (owners of Last.fm) have been lazy with it since they acquired it in 2007.

Inquirer's dilemma

This approach unfortunately also demonstrates an inquirer's dilemma: do your promote the tool you like or do you stay quiet and wait for the honest reactions of people who've just discovered one of the two services? You should also know that I made that term up. There might be a better term. Perhaps incompleteness?

With this particular topic, I have largely chosen to stay quiet and wait to hear what people say. We live in a time where tools that are neat find their way to the top of people's minds in an artificial, yet organic, way. We distribute the data and then rank them compared to the days data. We ask, "what do people care about?" Because of this, I don't feel as though I'm unhelping my personal favorite by not promoting them. As an example, how many people do you now that haven't heard of Pandora or Last.fm? Do they own an iPhone? (To be fair, those two might be strongly correlated...) A recent poll done on mashable shows that they appear to be tied. It's a small poll with 2465 votes.

How to choose

I wanted Last.fm to be better. I love the idea that it's based on how people actually behave instead of trained opinions. Unfortunately, I have experienced problems with it that motivated me to remove the software twice. After removing it once I fell for it again, installed it, and shortly after I got frustrated enough to remove it yet again. It might be due to my setup, so I'll describe that. I have all of my music on an external hard drive and have iTunes configured to know that. It works fine but sometimes loading is slow. With Last.fm installed I experienced extreme slowness when hooking up my iPhone to report what I listened to. Often enough it would completely fail to report anything. Argh!

Often enough my friends have suggested I use Pandora to listen to stations by band name. If I choose a Mastodon station, it knows what kind of other music I'd like. Last.fm tries to do this too. The difference in their approaches reveals interesting ways of thinking about music. Pandora's musicologists will consider the shape of the song as a single work and describe it to a machine. The machine can those uses those stats to look at songs and determine if they're similar. Last.fm purely listens to what you're listening to and then maps it to other people who listen to the same stuff. It will suggest anything in that pool of similarity that you haven't heard.

And so the benefits of using Last.fm or Pandora's algorithm become the deciding factors. Last.fm has proven useful to me when I'm in dire need of new music and am exploring a new genre. For example, I really like Amon Tobin and some other electronic music, but I'm hardly an expert. I listened to Amon Tobin, Alarm Will Sound and Aphex Twin and Audio Scrobbler sent that data to Last.fm. Last.fm suggested quickly that I check out Xploding Plastix. I had never heard of them, but they were awesome. So Last.fm found the gaps in my library.

Pandora, on the other hand, seems to take into account whether or not a band is up and coming. It will place bands gaining popularity in the stations and help spread the word. Last.fm doesn't seem to do this because it converges on averages in the statistics. It's as though it tells you what's popular that you haven't heard and Pandora tells you what will be popular. Last.fm seems to get redundant awfully fast though...

#myownlastfm

It's hard to beat an approach that will find the new things before they're big. That's the buzz behind Twitter and it's connections to real-time search. But I'm convinced Twitter is actually more of a protocol than a status service. It sends small message fragments around and people group the fragments with free-form tags, like #iranelection or #tigerwoods. So what if I submit my listening habits to Twitter with #myownlastfm as a tag? I think I'd still need to filter by username too, but that combination might be sufficient for tracking my listening habits.

I'm going to keep doing this for a while. Twitter, I'm convinced you're more useful as a message exchange protocol and other sites, start-ups in particular, can provide the context for the messaging.