Two Non-Poem Thoughts...

Apparently the "geek" part of me was unusually active today. Aside from feeling alienated by my computer not doing what I want, I don't usually think a whole lot about how it works or why it's not simply the greatest thing since the macaroni noodle.

As it happened, I was lying down for a rest/nap (I've been recovering from a sinus infection lately), and listening to some music on my iPod. I had the first thought: Why do "modern" computer operating systems still force users to manage the hierarchical tree layout of files on our storage devices?

I think what got me started on this thought was something I was working on recently. I've been trying to organize a bunch of old computer files I had on 3.5" floppy disks from years back. I didn't know what was really on most of the disks until I put them into my old PowerMac 7200/90 computer. In many cases, the information on the disk labels was incomplete or completely inaccurate. To try to get a handle on the situation, I copied all the disks to a "folder" on my hard drive, and then copied them across an Ethernet connection to my newer Powerbook computer. I then burned the whole lot onto a CD-R, pitched the floppies, and proceeded to try reorganizing and sorting out all these crazy documents and programs from my past.

Well, today it dawned on me that I'd done all this reorganizing of files on my laptop, and then I was going to need to deal with the same exact slog of files on the Power Mac 7200, because much of the content required older software that may not even run on my newer computers. I was faced with the problem that I've sorted the files in one place, and may need to do much the same work all over again in the other place. Gads.

Wouldn't it be nice if I could just "synchronize" the two file systems somehow to match up. But it was painful enough just to do the copy maneuver one way. There's a good chance what I'll end up doing is sorting the files on the laptop and copying back whatever I decide I want to preserve on the older Mac. It will probably not be that simple since I can't open many of the files on my Powerbook, so I may still have a lot of sorting out to do once I get things back over to the 7200.

Anyway, hence my disillusionment with the hierarchical file structures of "modern" operating systems. All those nested folders with arbitrary names is a real bear to work with, especially considering I created all these files so many years ago and often have no idea what I'm looking at until I open everything up and take a look.

Well, what is the alternative? First off, I want to point out a deeper thought I had at the same time, as to why trying to manage files in a hierarchical tree structure is not a great situation.

Don't ask me why, but the other day I was reading up on a bit of Greek mythology on Wikipedia, and I came upon an article that pointed out that a labyrinth, in the classical sense, is not the same thing as a maze. (You can check out the article if you're curious what the distinction is.)

Then I read an article on mazes that points out that a common type of maze is a course of branching paths, topologically mappable from 2D (or higher dimensions) to a hierarchical tree graph. In other words, mathematically speaking, a hierarchical tree structure is the same things as a maze. A maze is a puzzle purposely invented to make things hard to find. So why in the world are we hiding our files in mazes of folders and subfolders? Start to see what is bothering me about this picture yet?

I did an Internet search for alternatives to the Mac OS X Finder, to see if anyone has addressed this question with a software solution. The closest I found was someone's graduate class project, a file system called "Birch" written by Casey Marshall. (See a PDF of his write-up.) Clearly, Casey gets what I'm thinking about, though I'm not about to try his solution because it's clearly not a mature "user-friendly" system. (Yet?)

But Casey Marshall's write-up acknowledges the situation thus: "The issue with hierarchical file storage is that it depends on the user to create and manage these hierarchies, and so the quality and usability of a particular hierarchy is only as good as the effort put into defining that hierarchy. Good classification schemes are difficult for the average computer user to define — we have historically relied on professionals to organize large collections of data."

In other words, file systems are like a maze of our own design. So they can be helpful and clear, or obscure and difficult, depending on how good a job done by the designer. Most likely somewhere in between, because I expect really confusing structures are probably more trouble to create than most people are willing to undergo.

Alternatives do exist, for special-purpose media browsing. Apple's iTunes and iPhoto each create self-contained islands of file organization based on attributes of the content (artist, album, last played, rating, ...) rather than exposing the underlying file system. Granted, they do allow limited hierarchical organization with "folders" of playlists, but the nesting level is limited. You can't bury your music or photos inside 17 layers of subdirectory madness. And even if you could, there would still be easy access to the files by their attributes in the search box or music/photo browser. The iPod itself has a scaled back version of the iTunes interface. It is even more purely non-hierarchical.

iPhoto has a cool feature that makes this all even more appealing: you can tag photos with your own keywords. Think "vacation photos."

My thought is, would it not be great if all the documents of any significance on your computer were accessible in much the same way as in iTunes and iPhoto? If you could add your own custom tags to any file, and bring them up, classified six ways from Sunday?

Well, it's a thought, anyway.

My second thought I want to share today occurred while reading the description of a file browser (still fairly hierarchical, so I won't mention which one because I don't remember the name). The web site was oozing about how cool its graphical interface was. It talked all about perspective rendering on-the-fly. I had the thought, well, wouldn't it be really cool if the objects drawn on your computer screen were presented as if lit by the lighting in the room where you're sitting?

These days a lot of computers have built-in or attached cameras. So what if the computer used the information from its camera to decide how to draw the "virtual lighting" falling on the objects on the screen? Think shadows behind windows and icons moving as the sun tracks the sky. Maybe tiny reflections of your office rendered on "shiny" things. It might really seem like you could almost reach out and touch whatever it was you're looking at, as if it were a solid, tangible object. Granted, maybe you're sitting in a dark room and couldn't see things on the screen. But in that case you could "turn on" a virtual lamp somewhere "inside" the computer, and it would light things up well enough to read your homework or play your games or find your lost files in their mythological labyrinth.

On second thought, maybe this is not such a cool idea. Why do we need computers to be so realistic? What's wrong with the reality we've already got?

Me so geeky. :)