File Names and Evil Attachments: We must Shake off the File System Chains

Like most folks, i get a ton of email. I get a ton of attachments with email. And those attachments are evil. For the most parts they have only generic names like "my assignment" or "Job Application" or "invoice". The mail client will add a number to the name so that one file doesn't overwrite another, but that's for the file system's benefit, not mine, the human being trying to make sense of these files without having to come through associated emails.

So whose problem is this? people's for not using more descriptive name identifiers?

I blame the System.

But to be proactive, we suggest alternatives...

It's not right to ask people to who have some kind of file template to come up with nice rich file names - especially when some folks are still getting over the legacy of 8char file names plus three character extensions.

But those days of short file names are G O N E.

So we need file systems to step up to the plate and help name these suckers in meaningful ways. There's lots of simple stuff a file system could do: it could automatically prepend all files with a user id; it could develop a bunch of project codes/names that a person could use (like course numbers or business names) for specific files, and have defaults set up which a person only changes as they need - if the system can't determine the context from other cues in the data itself.

With more operating systems deploying technologies for rapid content indexing, such reading of documents for labeling cues isn't that unlikely.

Simple selections with well chosen defaults could take a load of effort of people for creating reasonable labels for files both for their own later retrieval and for sharing. These same file names could be decoded on the recipient end for multiple categorization of these resources, too. So i could say for instance "show me all the files associated with comp6012 from Alistair" without having to reef through file folders.

I know this kind of listing is just what Apple's Spotlight is aiming at in OSX Tiger due out April 15, but index retrievals alone are not enough. we still like to be able to look at where things are in relation to other things, so we need to make these labels apparent to us, not just derivable by the system.

In the myTea project, we're looking at this kind of approach of assistive file naming for bioinformaticians (see the short paper on this), whose biggest challenge it turns out is not coming up with new insights into genes, but is managing the hundreds of files they get on their desktop which are generated by the various Web processes they run.

We can do this. And more than just improve personal data, we can get that data into sharable forms which makes it feasible for people to share the parts of their work they wish to share with the world or subsets of the world, where this information is meaningful. Community. We can do this. It's time to be liberated from the file systems and provide interaction that frees us from naming files and lets us get on with what we want to do: have fun, create knowledge, share.