As PDAs become ever more powerful, their specs more and more resemble what used to be regarded as the domain of notebooks. PDA memory today is often measured in tens of megabytes and supplemented by removable storage up to a few gigabytes; processors run at hundreds of megahertz; and even VGA resolution screens are appearing on the highest-end devices. Meanwhile, the smallest notebooks get smaller and lighter, and the gap between the two markets starts to blur. With such capability in a handheld, it's inevitable that some observers are advocating that PDAs should adopt PC concepts, eventually turning into simply very small PCs (or, in some circles, very small Linux desktops). In my opinion, that would be a huge backward step for the evolution of the PDA.

Don't get me wrong, I am very much in favor of PDA/PC convergence where it makes sense. For example, I love the fact that I can now edit Microsoft Office documents on my PalmOne Tungsten PDA using the latest version of Documents To Go without having to convert between desktop and PDA formats of a document. (Why is this so critical? Because when documents arrive on my PDA as E-mail attachments, they are invariably in the native Office format). I also enjoyed being able to project PowerPoint presentations directly from my Dell Axim handheld using Margi Systems Inc.'s Presenter-to-Go, allowing me to forget about dragging my laptop through tedious airport security checks.

These examples of convergence around content and documents make sense because they treat the PDA and the PC as complementary tools: I create presentations at my desktop, tweak them on the road (perhaps in response to the latest figures), and present them from my PDA at the destination. What doesn't make sense is the idea that a PDA increasingly should come to resemble a PC in other ways. In particular, there are three aspects of today's PCs that I would hate to see adopted in PDAs: the operating system, the file system, and the user interface. Before I go any further, let me emphasize that I'm not saying that these elements are necessarily bad, simply that they're not what's best in a small, personal tool like a PDA. I also should add that Microsoft isn't the villain of this piece: Desktop Linux is just as unsuitable and in some ways even worse, for PDAs. (And yes, I know I'm going to get mail from Mac users .)

Unnecessary Exposure
My major gripe against the PC operating system is quite simply that so much of it is unnecessarily exposed to the user, and that consequently the user is forced to think and act in ways that make sense only from the system's perspective. For example, why does the user ever need to explicitly close an application? Of course, from the system's perspective, closing an application tells the operating system it can free up resources, but a modern personal operating system shouldn't be resource-constrained in normal use just because certain applications have been used since the last time the device was booted up. Memory can be paged, code can be swapped out, and internal resource tables (such as memory segments and graphics handles) can be sized, or dynamically resized, to meet most requirements. A model that makes sense in the data center and even on a heavily loaded desktop, namely launching and closing applications so the system knows whether they're in use, shouldn't apply to a personal device.

And in fact that's exactly the case on my Palm: I rarely, if ever, exit an application, I just switch from one task to another as necessary, and I have a single, consistent model for bringing a task to the fore. When I want to look at my PDA's calendar, it makes no difference whether I just looked at it a minute ago, whether I used several applications since I last looked, or even whether my PDA has been switched off for days since I last looked.

In the same way, the desktop file system exposes far too much of the system to the user. Consider how easy it is for the naive user to accidentally erase great chunks of essential operating system files, which are just sitting out there in the file tree. Special folders such as My Documents are a valiant attempt to paper over the reality of the file system, but opening up My Computer exposes the "man behind the curtain." In fact, the whole concept of an hierarchical file system is one that has little place in a user's view of the world. Useful though it may be to an administrator to subdivide the massive number of files that comprise a modern operating system, and a handy way for the system to separate files belonging to different users, it's cumbersome at best for an ordinary user.

For example, consider this article: should I file it under the "InformationWeek" subfolder of my "Articles" folder or the "PDAs" subfolder of "Devices"? Or both? Neither one will help when I come to look for articles I've written that mention Microsoft. The simple fact is, most people don't find hierarchies, whether of folders or anything else, very easy to understand. What I really need is a document "catalog" that I can dynamically view by any given criteria on demand. Criteria include both information about the document and information in the document--for example, InformationWeek articles that mention both PDAs and PCs written in the last year.

Mixed Metaphors
Finally, there's the issue of the user interface. I'm far from the first to wonder whether the current model has had its day. One only has to look at the clutter of the current PC desktop, which is covered in mixed metaphors--things on the desktop can be links to folders, shortcuts to applications, URLs, actual files or even, in the case of Outlook, an actual application (not a shortcut) that cannot be removed from the desktop without losing functionality. That's a far cry from the original metaphor of a place to keep frequently used documents readily to hand. Contrast that with the directness and relative simplicity of the Pocket PC "Today" screen, which is designed around a user's key tasks and essential information.

This system-centricity of the PC comes to a head in the concept of saving a file, something that we've all been trained to accept as normal, although it's something that a user should simply never have to do. If I make a change, whether to a document, to my calendar, or to a task in a project, it should be done--I shouldn't have to reconfirm it when I close a document. If I change my mind, I should be able to undo the change, even after I save, close, and re-open the file, a set of system-level acts that in most applications today divide your work into arbitrary sessions, carve your changes in stone, and lose your undo history.

It's interesting to note that Microsoft Word, in particular, already carries enough information to overcome this limitation, at least if you have Track Changes turned on. The concepts of undo and revision tracking could readily be merged into one consistent user-centric model that allows the user to undo individual changes regardless of sequence (as rejecting a revision does today), to redo those changes (by accepting a previously rejected revision), or to back out of changes in sequence (the conventional undo/redo concept). Saving a file should vanish from the user's mindset, and closing a file should be an organizing convenience, not a milestone in the finalization of changes.

The combination of these three elements of the PC model leads to a needlessly complex user experience. In all three of these respects, I firmly believe that the best PDA designers have got it right.

In fairness to the PC, it should be acknowledged that some PC applications do implement user-centric models. Some applications, such as FileMaker and Quicken, don't require an explicit save to retain user changes. Some applications can remember where you left off when you last closed them rather than requiring you to find your own way back, thus hiding somewhat the "close" concept. Word can potentially retain a change history and undo revisions for as long as you want them--at least until you accept and save.

Media applications that manage collections of songs by artist, genre, or other criteria demonstrate how a catalog-centric view of content can replace a hierarchical file system. Unfortunately, all of these improvements are limited by having to sit on top of the visible system model, and it's rare to find an application that fully embraces a user-centric view of the world. The next major revision to Windows may well embed some of these ideas and more closely resemble the catalog approach to content management than the familiar hierarchical file system. All of these changes may be hopeful signs that the PDA's philosophy of simplicity may be diffusing upward, even while some PC complexity is seeping downward.

Fundamentally, the PC still mostly reflects a conceptual model founded primarily on running applications or, at best, editing documents in an environment directly descended from the data center. By contrast, the well-designed PDA reflects a conceptual model founded on carrying out tasks using a tool, and concepts like "application" and "file" are secondary. It's my profound hope that it's the PDA conceptual model that prevails as these two world views converge in the next generation of powerful personal hardware.

Carl Zetie is an analyst with Forrester Research.

To discuss this column with other readers, please visit the Talk Shop.