Where’s My Stuff? Beyond the Nested Folder Metaphor

By Paul J. Sherman

Published: March 12, 2008

“Search tools help us to find our stuff after we’ve already lost it. They don’t help us organize our stuff.”

As I mentioned in my inaugural column, “Envisioning the Future of User Experience,” I often have difficulty remembering where I put my digital documents and files. It takes considerable cognitive effort to maintain a mental map of my computer’s nested folder structure, and inconsistencies creep into my folder structure as a result of the on-the-fly taxonomic decisions I make when filing things away.

There’s got to be a better way of keeping track of our digital stuff than this decades-old organization scheme.

Google, with the laudable mission of organizing the world’s information, has created desktop tools for content retrieval. Microsoft and Apple, too, have added desktop search capabilities to their latest operating systems. But let’s face it: Keyword search happens after the fact. Search tools help us to find our stuff after we’ve already lost it. They don’t help us organize our stuff.

And with some of us dragging around twenty or more years of digital stuff from one personal computer to another—documents, pictures, what have you—only the most determined and organized of us have their digital files organized in a way that facilitates rapid findability—in a file structure that is easily remembered and traversed.

I blame the file/folder metaphor for this rampant problem. It’s no longer up to the tasks we’re attempting to accomplish using it. I think there are better ways for us to organize our digital stuff.

The File/Folder Metaphor

“The file/folder metaphor just doesn’t cut it anymore.”

The file/folder metaphor for organizing content on our computing devices has been around since the advent of Unix-like command-line interfaces and shells. It became more visually oriented with the releases of the Canon Cat, Xerox Star, and of course, the Macintosh.

Along with the desktop and paper-based computing metaphors, the file/folder metaphor has attained near-monolithic dominance in the most popular consumer and business-oriented operating systems. It reflects our paper-based methods of organizing information and hence makes it easier for people who are new to computing to adapt to the digital world.

But here’s the thing: The folder structures on our computing devices now organize more than just our documents. We store an incredible array of stuff on our computers—for example, here’s just a small sample of the stuff I keep on my computer:

  • a picture of my signature that I occasionally use to sign documents that I fax directly from my computer
  • my music MP3s and audio books
  • my family pictures and digital movies
  • my “art collection”—about 300 hi-res images of landscapes and Hubble space art I’ve been collecting over the past ten years and use to decorate my desktop (Don’t even get me started on the stupidity of affixing wallpaper to my desktop.)
  • templates for user experience work
  • my presentations, papers, and so so

The file/folder metaphor just doesn’t cut it anymore. Sure, it makes sense for the digital objects the device itself maintains and relies upon—that is, operating system files, program files, scripts, and databases. And to a certain extent it’s still appropriate for organizing document-centric information. But it’s not up to the task of organizing the variety of stuff we keep on our computing devices these days. It’s as if your office mate decided to use her file cabinet to store her kid’s inflatable pool, her husband’s chainsaw, and her scrapbooks, and also kept some paintings and potted plants in the top drawer. It’s just not the right kind of container for all those things.

Is there a better, more natural way of organizing information on a multi-purpose computing device such as a PC?

People Organize For Findability

Findability is, according to Peter Morville, author of “Ambient Findability”:

  • the quality of being locatable or navigable
  • the degree to which a particular object is easy to discover or locate
  • the degree to which a system or environment supports navigation and retrieval

In the distributed world of the Web, findability usually focuses on getting people to the stuff they’re looking for, but didn’t previously have. This could be information; an asset like a program, image, sound file, or some other entity; and so on.

For stuff we already have, findability takes on a slightly different meaning. The problems we need to solve on our personal computing devices include

  • memorability and findability—How can I organize my stuff so I can easily remember where it is?
  • navigability and findability—Can I easily get to something once I remember where it is?
  • flexibility, maintainability, and extensibility—Can I easily modify how I organize my stuff?
“In the physical world, people organize their stuff using many criteria simultaneously.”

I’d like to put forth this proposition: In the physical world, people organize their stuff using many criteria simultaneously. Their organizational schemes are, of course, limited by physical factors such as available space, but people’s organization schemes are heuristically driven—by convenience, frequency of use, and other factors relating to the context of need. In the physical world, we also create spatial relationships between things.

But in our computing devices, space restrictions don’t apply in the same way they do in the physical world . And, we have the added advantage of being able to reorganize our digital stuff at a single command.

All of this begs the question: Why don’t our computing devices let us organize our stuff in a way that leverages some of the advantages of the way we organize stuff in the physical world, while avoiding its restrictions?

Can We Leverage Existing Behaviors and Predilections?

“We can actually can improve on how people organize their stuff in the real world, because the spatial limitations the real world imposes are not factors in the virtual world.”

Here’s what I’m proposing: We can actually can improve on how people organize their stuff in the real world, because the spatial limitations the real world imposes are not factors in the virtual world our computers’ programming and storage capacity make possible.

Let me get concrete here: Why can’t we organize our stuff in virtual spaces—rooms, really, not the “one damn desktop after another” spaces of Linux or OS X—that we can then configure to our whims?

And wouldn’t it be nice if we had the option of instantly having our digital stuff rearrange itself according to any attribute we chose—for example, creation date, date last saved, type, or various metadata tags—in a three-dimensional representation that we could browse by walking or flying through it.

And what if you could approach one of your digital objects and have it resolve into a preview of itself as you got closer. Then, at a certain point, you wouldn’t be previewing it anymore, but editing. (Yes! One step closer to modeless computing.)

Here’s another scenario: How valuable would it be if you could create a virtual project room that was as useful as some of the best physical-world project rooms, complete with lots of table space, whiteboard space, a long wall on which you could hang and rearrange storyboards or architecture diagrams, or whatever noodlings and doodlings you wanted.

Here’s the kicker: Let’s say you’ve finished the project. In the real world, when you finish the conceptual design phase, you almost always have to break down the project room. After all, another team is going to need that space. What if at the flick of a bit you could recreate the project space from a completed project, displaying it exactly as it was before you shut down the project?

If this sounds like virtual reality, well, I guess it is. The key difference to me is that, to date, we’ve used VR technology to solve a different class of problems—mostly those related to entertainment, training, and social interaction. I have not seen many instances of this concept being applied to the problems of findability, storage and archival, or the organization and retrieval of personal content. But several recent items in the news suggest we may start to see VR technology being applied to this class of problems.


The blog nooface.net—“In search of the post-PC user interface”—reports on a number of products that present search results and the contents of a user’s desktop in visual—sometimes zoomable—user interfaces. One interesting program that’s now available is SpaceTime, a PC desktop program that lets users display search results in an unlimited 3D environment, shown in Figures 1 and 2.

Figure 1—SpaceTime user interface

SpaceTime user interface

Figure 2—SpaceTime video


A “lookable user interface” application, according to its creators MATT Services, Spectasia presents objects in a 3D perspective plane, as shown in Figures 3 and 4. This allows users to select near or far objects, rotating the portion of the plane containing the object to the foreground.

Figure 3—Spectasia user interface

Spectasia user interface

Figure 4—Spectasia video


A three-dimensional representation of a desktop surface, Bumptop lets you toss, stack, move, and gather up icons representing your files or folders, as shown in Figures 5 and 6.

Figure 5—Bumptop user interface

Bumptop user interface

Figure 6—Bumptop video

Cooliris Piclens

Cooliris Piclens is a tool that enables users to browse images on the Web as though they were in a gallery. It arrays images on a vertical surface that stretches off into the distance, giving the impression of strolling along a gallery wall, as shown in Figures 7 and 8.

Figure 7—Cooliris user interface

Cooliris user interface

Figure 8—Cooliris video



From Microsoft Live Labs, Photosynth is a technology that actually stitches together multiple pictures, drawings, or schematics of a particular location, enabling users to traverse the virtual representation of the physical space and zoom in to examine details at very fine levels of resolution. A video demo shows Photosynth in action.

In Summary

None of these applications gets to what I envision: organizing our stuff in virtual spaces that are more like a well-organized closet or kitchen drawer than a desktop. But all of them take steps in that ultimate direction, and I, for one, can’t wait to store my kids’ pictures on a virtual wall. That would be so much better than putting them in a folder.


Some interesting ideas here, but I’m not sure I see how these approaches will actually help us find our information more easily.

For example, in the picture you posted the other day of the massively overloaded desktop, how would the software you’ve presented here help fix the problem?

Aw, come on. Have you seen my closet?

Thanks for summarizing these different organizational metaphors! It’s great to keep abreast of these designs—sooner or later, the desktop metaphor will get replaced, right?

I disagree though, with this point:

“Keyword search happens after the fact. Search tools help us to find our stuff after we’ve already lost it. They don’t help us organize our stuff.”

I find tagging, labeling, and organizing content to be time consuming, and often, still not adequate to find everything. It’s also frustrating that different applications organize things differently by default, requiring more effort to keep everything consistent.

I’d rather have my computer organize my stuff for me, using any available metadata, and make it keyword searchable. To build on your kitchen drawer comparison, wouldn’t you like to just ask your kitchen where the ice cream scoop is rather than check the drawer, then the sink, then the dishwasher, and then ask your spouse where she / he left it?

You started out discussing methods of storing files and ended up discussing 3-D interfaces. The two are not necessarily the same.

Is a folder not a virtual space? How will storing files in a 3-D virtual space fix the problem any better than folders?

I think the idea of rearranging by any attribute has some merit, and creating tags and searching by keywords do, too.

I think you need to throw out the concept of hierarchical file systems altogether and find something else. Then you can put whatever interface you want on top of that.

I think this is the direction we are going to be heading. Aza Raskin gave a talk related to this subject; I found it quite interesting.

In order for this to work, I think you need intuitive software backed by a large developer base, plus innovative hardware to navigate the interface efficiently. It may take a company like Apple, who controls both the hardware and software and who have a large developer base, to make this thing work. Apple has shown their willingness to try out new technologies and interfaces with the iPhone, Spaces, and multi-touch trackpads; the trick is for them to go all the way to something like what Aza Raskin described in his talk. There are probably many months—and possibly years—before we will see big companies embracing this approach, because the desktop metaphor is so ingrained into users’ mindsets, but I think we are slowly but surely going that direction.

Wow, some nice applications. I see everyone making a big deal out of the CoolIris interface; however,I always seem to get an error when I run it. It does seem like something promising though.

I hope Firefox will come up with something similar.

The information (meta-data) you are going to search for in the future has to be associated with the file. There’s no way around this. At a certain point, thumbnails do not help. We need the computer to actually index and find the file/meta-data. This means that users need to create files with meta-data—good file names and so on. It’s like trying to find an email with a poor subject.

Yeah, too much Everquest, methinks.

I can easily imagine my folders as rooms with closets and cabinets with shelves and drawers—which of course ultimately contain nothing but folders and files—if I wanted to. Would the overhead of a kiddie-game style UI enhance that abstraction? Not much.

Let’s look at the Google search though—the after-the-fact organization. It does a pretty darned good job of organizing information that was not originally stored in a structured manner—nor was it intelligibly tagged with metadata. And I think it can only improve over time.

Flip that coin over, and ask yourself this: What if the intelligence that goes into a Google retrieval were applied to the initial storage of a file? Granted, any file will already contain a wealth of metadata from which much can be inferred on retrieval, but what can that metadata trigger for us on the initial save?

I’m playfully envisioning nine holes in my desktop, where I can drop any file of any type. And considering the file type, the context that it came from, and the choice of hole that I dropped it into, the system pops up and asks me one very specific question—something a little more subtle than “please enter a file name”—that suddenly increases the metadata for that file by a huge amount, but without me having to hassle with tagging it.

The next time I need it, I can probably search directly to it, with just a few clicks or keystrokes.

No silly icons or draggable walls required. Just a thought….

I agree with Joshua on keyword search. I find that I am less concerned with thoughtful organization as I become more reliant on tools like Google desktop to find what I want. I can often remember enough about a document to find it in a single search. I don’t even file email anymore now that Outlook allows me to sort and group by so many attributes.

Of course, these technologies work well for text documents, but media files are still a challenge.

I think the real key is to put things in places where you can find them. There are some ideas that may be better utilized from cognitive science—such as mind mapping, mnemonics, and what I have already seen somewhat, placing items in a 3-D space like a room—an ancient Greek memory trick.

The mind mapping is thinking about how our mind links tags, meta-data, or basically just thought ideas—cat leads to dog leads to bark leads to tree. It’s the relationships that make you find the object.

Mnemonics is making something memorable by how you place it in your mind. I am sure you know all these tricks like memorizing shopping lists and so on. Maybe we can use these ideas to see how we can rethink metaphors for objects, data, and so on in our computers and networks.

Thanks again, Paul, for turning me on to this blog. It’s great to see how you’ve presented this problem in your usual smart yet down to earth manner. We all have trouble finding things. I can’t even find words at the tip of my tongue. Maybe it’s time to fold the file/folder metaphor.

I disagree that putting things on your computer is like filing an inflatable toy in a filing cabinet. There you are talking about the physical world. Here we are talking about a virtual/digital representation of many different things. What ties them all together is that they exist in a digital medium—so one perspective to file them is fine.

I do agree that content management on a personal, partner, team—pick whatever level of group—is important and that Google seems to do a pretty good job.

For me, I want a happy medium. The file, forget, and search approach is one end of the spectrum. The metadata nazi is the other end of the spectrum—that is, you have to fill in all the non-technical, tribal, contextual metadata that you associate before your virtual system will let you land a file in its purview. Some of that non-technical information and data might be: How were you feeling when you created the document or picture? Happy, sad, mad. (I suspect any amount of AI would not be able to handle that.) Where were you at physically? Beach, home, work—with internal GPS you could get that. What was your intent for this piece or group of digital information—future plans, document status, refer it to a friend, put it on a digital wall, use it in a future column. By the time the Nazi got through, you would have spent a considerable amount of time and effort in documenting that virtual file.

One project I worked on was documenting generic project documents for a consulting firm. The content had a database that enabled the user to pick a perspective and then list the documents that were available—type of document. (Typically, I remember, if a document was created in Word, Excel, PowerPoint, Microsoft Project, and so on.) This is easily done programmatically. Another perspective was use—template, guideline, presentation, requirements gathering, summary. Another perspective was audience—internal to consulting firm; user oriented with breakdowns for executive, manager, user, DBA, and so on. Another perspective was methodology—two types. One specific to the consulting firm—how they approach a project—and the more generic PMI methodology, which had a matrix, comprising Initiation, Planning, Executing, Controlling, and Closing, and nine knowledge areas—Human Resources, Integration, Scope, Time, and so on.

This allowed many people to use the room analogy you like to find the content they needed, because I had filed the data in a house they were familiar with or that they could visualize the floor plan of once they were exposed to it. The problem is that each of us has created our own house. Each of the specialties, families, or cultures we have been exposed to either have created multiple houses we use to order our world or multiple galleries in the museum of our mind.

There is no universal solution—unless you take the route of the content nazi. There are, however, some fairly common methodologies that you can apply that will get closer to what you want. I thought the virtual world in Johnny Mnemonic was great. I think some of the graphical node/line relationships are very useful. All of those require someone to create the structure and then file according to that structure and, if more than one person is involved, you need consensus.

How am I going to file this? It is named content management, it is a Word doc, it is in my documents / projects / content management, I did not add key words nor any additional property information.

Agreed—the file folder needs to evolve. I am working with a company that is setting out to do just that. Think metadata at the folder level and folder types. This approach greatly reduces the need for user metadata on documents, images, and so on, while enhancing findability.

David, you make some really interesting points. Your comments on perspective strike me as similar to facets in the faceted classification model. (See Peter Morville’s article on information architecture at the speed of the Web. I’d look for the link, but you should be able to search for it, he says ironically… ;-)

Join the Discussion

Asterisks (*) indicate required information.