It amazes me that here we are in 2005, and we're still not exactly sure what an access point should be. Fat? Thin? Fit? And how come we don't have standard protocols on which we all can agree? CAPWAP? LWAPP? SLAPP? The answer is easy, and it can be found in any Marketing 101 textbook. There's no point in setting a standard when everyone is making lots of money without one.

But let's move on to another fat vs. thin issue, in this case the girth of clients in mobile devices. I recently retired my mostly beloved Danger/T-Mobile Sidekick (that was one too many outages, there, T-Mobile) in favor of a palmOne (soon to be, once again, just Palm) Treo 650. I consider the Treo 650 the best of the bunch when it comes to smart phones, though it's far from perfect. And the reason it's flawed-aside from the $400-plus price tag-is that it's a classic thick client. It runs an OS and applications; as such, it is not terribly intuitive and, yes, it does crash from time to time.

Most important for my purposes here, its browser also leaves a bit to be desired. For example, the browser does a poor job of rendering the Yahoo home page, which I use all the time. And it's not very good at caching pages, so there is much waiting as one moves back and forth. Given the amount of processor and memory devoted to this product, I had expected better.

But perhaps I'm asking too much. As a fat client, the Treo is designed to accommodate many applications; it's a platform, after all. It's also mobile, which means lots of compromises in functionality and design no matter what. But the big question is whether we need to carry computers, and all that they entail, with us.

For some time now, I have been convinced that the future is thin and that the client of the future is essentially a wireless browser, not something with an OS.

Assuming a Web services model, we thus could literally carry only a smart card for authentication and then pick a product from a variety of form factors-from cell phone to PDA to tablet to notebook to whatever-depending on the task and venue at hand. You could even borrow someone else's device: Plug in your card, up comes your desktop. No need to synchronize or worry about whether we've got the right files with us. No need to make sure the OS is up to date. No viruses. Low cost.

The only detail standing between us and this vision is ubiquitous wireless service. But with expanding 3G services (especially of the 1xEV-DO and UMTS varieties), metro-scale Wi-Fi based on low-cost mesh deployments and maybe even 802.16e mobile WiMAX or other all-Internet Protocol mobile broadband services, there will likely be no shortage of bandwidth in the future. Of course, some local caching of the address book and similar data will be required even if the client is thin. But, hey, we know how to make this work.

The value of the mobile device, compared with that of the data it accesses and manages, is today very low indeed. Let's keep this the case: Thin, mobile, wireless clients are the best hope for the future of mobile computing.

Craig J. Mathias (craig@farpointgroup.com), principal at Farpoint Group (Ashland, Mass.)