Smart Folders, Views, Attention Gestures and SXSW Music

I've been an advocate for a while of moving away from a single structure to organize data. The typical folder heirarchy of a computer filesystem bugs me. It's the same with bookmarks, categories for nearly anything, etc.

Basically, any time I'm forced to put something in a limited number of places, I find myself wanting to put it in at least n + 1 places. That's why I like views, in whatever form they take.

On Thunderbird, they're actually called views. On the Mac, they're "smart folders". In iTunes, they're called "smart playlists". Regardless of name, I dig 'em.

What's helped them get moving is that we're finally starting to capture attention gestures better. The less recording of metadata that users have to do, the better the whole "view" thing works. A prime example is what I'm doing with iTunes this week.

This year, like last year, SXSW released a giant torrent of songs from the bands performing for the music side of the conference. And, like last year, Garrick is weeding through them to pick and share his favorites. I thought it might be interesting to do the same. It also offered an opportunity to mess with the iTunes smart playlist functionality that I've been dabbling with a bit lately.

I downloaded the torrent's 700+ songs and imported them into the iTunes library. I added a folder for the smart playlists for this and created the first one. I set it for all songs with "SXSW" in the album (all of the songs in the torrent are in the same "album"), that hadn't been listened to or skipped 1 or more times and made it random.

When I saw "skipped" in the available metadata, I was pretty happy. That's the kind of implicit, easy to capture gesture that has been ignored for a long time. First, I can just hit a button on my multimedia keyboard and skip the current song. It's the natural thing I'm going to do if I don't like it (to move on to the next song) and it means I wasn't willing to listen to it all the way through, even once.

That's entirely different from rating songs. I'm not very likely to take the time to rate a song as having one star if I am ready to bail on it 10-15 seconds into it. Rating a song requires bringing iTunes to the foreground, finding the song in the interface, and making a couple of mouse clicks. Given that listening to music on the computer is *always* a secondary activity, that's just not going to happen for songs I don't like.

All of that makes this first smart playlist a good way to get my first pass through the list. The next smart playlist is the songs that have been listened to at least once and never been skipped. After I've gone through the entire original list of songs, I'll move on to this playlist, which is, at the very least, the songs that didn't bug me. If any do, and I skip them, they drop out of the list.

Now, already, there are a few that I caught myself really digging and wanting to hear again. They rose above the background to be really likeable. In those cases, I *did* go and rate them. I tried to be really honest about the star scale and found myself only bothering to rate stuff that was at least 4-5 stars. That's telling.

There was recently a story about how useless Netflix ratings are because, say a fan of Jet Li's movies is going to rate Fearless and other movies highly, but is not likely to go into the romantic comedy section and rate those poorly. The net result is that *everything* is rated highly and the star system doesn't work. So far, this experiment indicates that if I use "skipping" as my "thumbs down", then the presence of a rating *at all* is the only really useful indicator of my "thumbs up".

Regardless, for the final playlist, I did make the criteria a rating of at least 3 stars, in case I do eventually rate something on the low end. That final list, listened to at least once, never skipped and highly rated, will be a list of favorites that I can easily share.

For the most part, it required no more thinking about building a list of favorites than just listening for the heck of it would. The majority of the process is handled by just capturing my implicit attention actions. The little bit of action I *am* taking is to just note down what's already being brought to the foreground of my mind by enjoying the song enough to consciously think about liking the song.

In iTunes, this kind of thing works really well, because much of the metadata comes from backend databases (like a 1980's playlist just reading the ID3 tags for release date) and much of the rest can come from just paying attention to your existing actions. At the operating system level, they're starting to pay attention as well.

If you're on Windows, you might be jealous of the Mac's smart folders. Well, things are finally getting going over there.

Windows Desktop Search brings rudimentary smart folders to Windows. You can save searches as links (not the same as being able to map folders, but a start). After doing a search, you drag the magnifying glass of the address bar to where you want the link and there you go.

You do have to tell Windows Desktop Search to index anything other than the "My Documents" tree, which may or may not matter to you. If you tell it to index everything (like I did), it will take a while, but once it's done, you finally have a quick search for the files, emails, appointments, etc. on your Windows computer.

And, if you're a bit geeky and looking for very specific kinds of searches, you might want to look at the search query reference document. It covers all of the nitty gritty like how to get all of the emails you sent for a given day, etc.

Every implementation I see that essentially puts your files and data into a big pile, letting you define views to pull out exactly what you want is a step in the right direction.

I *like* being able to search specifically through all of my PHP files as a subset of my files *and* being able to bring up PHP files by project. The fact that I could also bring up all of my PHP projects that use PEAR libraries or all of my C# code that uses the TortugaNET library without having to store things that way in folders is fantastic and, eventually, will be seen as the only sensible way to do things.

Is there some way to fast forward to that time and skip the crap we're stuck with now?