As we plunge headlong into the future of connected devices we can't make the same assumptions we have in the past about the nature of the ultimate killer app. A common misconception exists that if you can find the right mix of features and capabilities appropriate to the new post-PC appliance environment, the new killer app equivalent of the desktop PC will emerge. Then, it is believed, business as usual will prevail, along a predictable straight-line development track as consumers rush out to buy the new computing and entertainment platforms. People assume that we will see a repeat of the experience of the desktop PC in the '80s and '90s, but with ubiquitous wired and wireless connectivity thrown into the mix, and that many of the strategies that worked in the past will work in this new environment. Assumptions like this are valid only if the underlying foundations are the same. In the PC market we had a system of interrelated activities that operated within a mature marketplace with established norms and direct causal links. Given an established set of services, OEMs, and channels, then technologies and products along a particular line of development were assured of some degree of market success. However, we are going through a time much like that period in the '70s after the introduction of the microprocessor and before the desktop computer. Now, as then, the market is in a state of chaos. All previous causal relationships have been rent asunder due to the collision of computing and communications, with no industry wide set of services, OEMs, and channels to depend on. After the introduction of the microprocessor, it took ten years for the market to reach a steady state. I see no reason to assume that this new computing environment will settle down into a similar predictable framework much before the end of this first decade of the 21st century. In particular, I see this mistaken assumption at the heart of many of the new wireless multimodal computing and communications appliances that are hitting the market: PDAs with WLAN capability; cell phones with PDA capabilities; both such platforms with gaming and/or camera capabilities; as well as full-motion video and MP3 audio. Everyone seems to be adopting the Microsoft feature-rich strategy that worked so well for the company in the desktop era: build a low-cost box that is as powerful as economics will allow, and if you just continue to pile on the features the customers will buy and keep buying. Name any major player in this connected computing environment --- Intel, Microsoft, Nokia, Ericsson, Samsung, Texas Instruments, and Motorola among others -- and you see everyone assuming that these feature-rich multimodal devices are the end game, the platform from which years, and hopefully decades, of stability and profits will come. To support that strategy, fundamental changes are being made in the wireless connections that are becoming the common mechanism by which these various new embedded and computing appliances connect to the wider wired Internet. Three of the most important of these are: (1) the intensifying rush to achieve multimillion bps transmission rates over wireless connections; (2) improvements in quality of service to at least match that of the wired Internet; and, (3) achievement of an "always on" mode for wireless LANs, similar to that on the cable or DSL wired connections. Achieving those three goals, it is believed, will make the multimodal Internet appliance blossom into the wireless equivalent of the desktop PCs in terms of the unit volumes and profits generated. But as many companies-- IBM, Fairchild, and even Motorola, among others -- found out during the chaotic '70s, such assumptions cannot be made with any assurance of success. Ironically, all of the money and intellectual resources being committed to improving wireless connectivity are just as likely to have an impact quite the opposite of what every one hopes will occur. For example, let's look at the impact of just one of the above initiatives -- multi-mbps wireless transmission rates. An immediate result will be to make true the misleading television ads about wireless phones that allow you to quickly and easily transmit pictures of your pet dog or to "blog" video and audio on the Internet. If you have a chance to try out some of those devices, you'll find that unless you are willing to spend the equivalent of what you'd spend for a high-end desktop and create a carefully engineered environment surrounding it, what you get is a compromise design. That's the fundamental nature of such feature-rich, general-purpose platforms. It was true of the desktop computer and it is true here as well. Everything you do depends on the compromises the designer or the end user makes. You can certainly take a reasonably high quality photograph with a PDA-camera-cell phone combo, but only within the limits of the LCD imager that is built in. And you can save it as a relatively high quality photo image if you are willing to sacrifice processor and memory resources committed normally to a number of other functions. And you can probably send the image if you're willing to compromise even further on the quality of the image or the time it would take. Ditto for live video, Internet functionality, voice communications and any combination of features in such converged compromise devices. We want to believe that with high bandwidth wireless connectivity, the compromises found in present devices will be a thing of the past. I am not so sure. Certainly, with high wireless bandwidth you can send a high-resolution photograph or video image, but not without making compromises to conserve battery power or maintain the ability of the internal processor to perform other important operations. And it's just as likely that high bandwidth wireless connectivity could lead to a much more distributed computing future in which general purpose handheld wireless platforms will be a thing of the past. Think about it. With sufficiently high wireless bandwidth, what's the point in having everything converged into one multipurpose unit? Why not a confederation of separate dedicated devices with integrated high bandwidth receivers or transmitters: a high quality, high pixel density CCD for image collection; a high quality, high pixel density LCD integrated into a pair of glasses; a high quality wireless earphone and a high quality wireless microphone or connection to the Internet and so on. One objection to such future would be the size and number of such devices. Given even our present fabrication technology, each such dedicated device would be no larger than a few coins or a pair of dice. The audio and video input and retransmitters certainly could be small enough to attach to a shirt collar or coat lapel, or to put into a finger ring. The computers that must perform the post and preprocessing almost certainly would be no larger than a credit card. When I look inside a converged PDA/cellphone/camera and subtract out the sensors and analog-to-digital conversion circuitry, the core computing devices that are left would fit on a card not too much larger than that. Isn't carrying half a dozen or so personal computing devices, no matter how small, rather inconvenient, you ask? I don't know about you, but these wireless computing and communications devices we're presently burdened with are much more inconvenient: too big to fit in a shirt pocket and too large to comfortably slip in a pants pocket. And when you hang one from your belt, it's likely to fall off if you sit down. Besides those you need to connect to your collar or to your ear, a half dozen or so such quarter-sized personal electronics devices could be easily deposited in all those pockets we no longer use for coin or paper money in this increasingly cashless society. The power draw on the batteries of each of these dedicated would be much smaller, I suspect, than in a converged multipurpose unit. And it would certainly give you more choices. If you were going to be traveling by car, you wouldn't use the audio/MP3 player in your pocket. You would use the better one in the car. What I am saying is that we don't have the same ubiquitous services/products/channels infrastructure that came into being during the desktop era to direct the way in which new technologies will evolve. In an environment in which we are still establishing such an infrastructure we can't make assumptions about what the ultimate killer app will be. Reader Feedback This is an excellent output from Bernard, but I wonder about point below: If we compare the above to software between having small individual utilities performing their own functionality in a UNIXified OS to one giant which has everything embedded into it. Now, the strange thing about pragmatics is the point about how the Giant Bloated OS has succeeded in the consumer market in sharp contrast to all our principles! WHY? Is not because of user convenience? Isn't Making a jukebox certainly one thing that is convenient to the layman than using an assorted set of utilities for the user to mix and match? Saravanan T S Senior Engineer TEL Nice article. I share your vision about the distant future. For the near future it seems that multimodal is the way to go. I think it is marketing thing. Hansdpring makes mobile phone/PDA combo because it has a head start in PDAs and Nokia makes such combo because it has third of the mobile phones market. I welcome it because I don't really want carrying both palm AND phone. On a different note, when I think of the future of devices, there are invariable factors, the human nature. The distance between human mouth and ear. It limits the shrinkage of mobile phone. Being a user of bluetooth headset I cant see wireless earbuds as a solution. They must be light as air if I am to wear them all the time. Anyway, I liked your article and I too think we'll all have a lot of fun watching Microsoft battling Nokia and Nikon choosing Linux for its next OS. Michael Kariv Bernard Cole is site leader and editor in chief of iApplianceweb as well as site editor for Embedded.com and an independent editorial services consultant working with high technology companies. He welcomes your feedback. Call him at 602-288-7257 or send an email to bccole@acm.org.