Trolling4Dollars's Journal: File Systems and Users

I know I'm not the only one who's been thinking about the topic of file systems and computers. But recently I got a pretty good parallel in the real world to compare file systems with and why so many regular users have trouble with computers.

Knowing how to get around a file system on a computer is pretty much the equivalent of being able to remember how to get from your house to work in your car. You make a mental map of the route (based on a number of elements unique to each individual: more on that later). Without this ability, getting to work (or anywhere else for that matter) would be pretty much impossible. It took you a while to get to that point. When you first started driving, maybe it was only to your friend's house a few blocks away. Then as you got older, you started going farther and farther. You got familiar with more locations near your home, your city, your state, and then maybe even made cross-country trips in your car. You also probably needed a map at times.

When I am dealing with users who are new to computers, the one skill that seems to be the hardest to impart to them is getting around the file system. It doesn't matter what OS they are using, this particular skill seems to be very hard to acquire. To me, these all pretty much mean the same thing:

Mac OS - music:mp3s
*nix - /music/mp3s
DOS/Windows - C:\music\mp3s

But to the average user, there seems to be some confusion. They may "get" the three simple examples above when you explain it to them. But when they are at the critical moment by themselves, and need to go somewhere other than the examples you showed, panic of frustration ensue. A lot like the folks who had trouble with math when they would go home and the math problems were manipulations of the basic form.

If computers are ever going to get easier to deal with, one of two things needs to happen:

1. File systems need to become more intuitive
2.Users need to understand the hierarchical concept clearly

I think understanding where their files are is important to users otherwise a computer doesn't do them much good at all. Just like a car wouldn't be of much value if someone just had to drive around town randomly hoping to find theplace where they work.

There are projects in both the free and non-free software worlds that are hoping to solve this problem. Notably, Microsoft's WinFS and GNOME's Filesystem projects are the most visible. The concept being that a user need not know the name or type of file they want, they simply need to know some key words and facts about the file. "The e-mail I was composing this morning" "The pictures from the Australian vacation taken in 2002" "All movies with Kieffer Sutherland", etc... Hence the need for a database backend to abstract the underlying filesystem. A noble approach, and one likely to succeed in the future. But it doesn't help us today. In the interim, I firmly believe it is an essential and required skill to navigate a file system if a computer is to be a truly useful tool.

I believe that the barriers to this are many:

1.The Windows drive letter paradigm
2.Multiple file selector interfaces in all Oses (Compare WinDVD and WinAmp)
3.Inconsistent metaphors (File | Open in a standard application, Eject button or folder icon in a media player to call up a file seelector)
4.The average user's difficulty in memorizing a path to a file.
5.Kludges like Windows shortcuts and symbolic links to circumvent the navigation problem.

And doubtless, there are many more obstacles.

To get around these obstacles, I believe the user must have a clear grasp of the concept of directory hierarchy. Much like people at one time were perplexed by steering wheels in combustion automobiles, this new concept is something that must be layed out clearly for users who are stuggling with computers. Eventually they will "get it" if they are exposed to it enough in non-threatening situations.

I have much more to say on the subject, but I will stop here for now as I think a discussion among those who are intersted will bring much of what I have to say (as well as the thoughts of others) out.

Thank You for Reading,
T4D

The scariest thing is

[insanecarbonbasedlif]

when a user has been saving files to random locations on their hard drive, and then they want me to help.

Inevitably, I find three or four copies of the file that they want, and have to help them decide which one they want, and then get rid of the others. Sometimes I explain the filesystem, and consistent folder use, to them 4 or 5 times before they understand it.

I think you're square on the fact that file systems are inherently obfuscated to the non-technical user.

This is especially entertaining...

[Zontar The Mindless]

...when trying to help someone over the phone or the Net.

"Double-click on 'My Computer'... Now double-click on 'C'... Whaddaya mean, there's no 'C'? ... [*sigh*] ... Okay, tell me what all's there in the little window that opens up when you double-click on 'My Computer', then..."

I'm not really sold on the idea of embedding a DB server into the OS, tho.

At least, not SQL Server.

Moo

[Chacham]

Heirarchy is natural to Js, and not natural to Ps. Ps would rather have everything scattered, with probably piles of related items in some order that only they can understand. Considering the world is about split equally amongst Ps and Js, heirarchy is not the way to go.

So, the next though is, have a heirarchical system, and allow some non-heirarchical interface. That's a bad premise, since the file system is not heirarchical. Even on file systems where there is a superblock and directories are heirarchica

good observation

[goon]

intersting post. I'm actually working some tools that address this problem. It stems from the fact I download a lot of crap (files, documents), try to organise them, backup themup etc.

the most frustrating thimg about these steps is organisation, retrieval and backup. If you have no *system* to store data available where the protocols allow - (email, url links, html links, pages downloaded, applications, etc) the sheer numbers will kill you.

my approach is to build some tools that capture the structure of

info glut

[goon]

info glut [slashdot.org] - missed this one but pretty much sums up a few more problems. there are no tools to filter what is *good* from what is *bad* easily.