The Coming Resurgence of Internet Appliances

Dennis W. Forbes - December 17th, 2001

Dennis Forbes


Internet Appliance (IA)
An inexpensive, minimally featured terminal for web access

One oft misquoted observation of the software world is that "80% of users use just 20% of the features of a given software product". The natural (flawed) extrapolation being that if a product implements 20% of the features, it therefore satisfies 80% of the market. In the software world such a design formula has been frequently proven wrong. Instead, 80% of users use a common 15% of the features (in a word processor that would be the common features like word justify, spell checking, font selection, etc), but each of them draws upon a disparate 5% to complete their tasks. In other words, whichever 20% of the features that you choose to implement, you'll quickly find that many users will be frustrated that their pet feature isn't to be found. Because of this fact "light" versions of anything in the software/computer marketplace have usually been met with very limited market success. I say this having been one of those misguided software developers that oft had the epiphany that what the world really needs is a nice lightweight, non-bloated, core-featureset word processor! (Usually right after wrestling with misbehaving cascading styles or numbered lists in Microsoft Word), but thankfully I abandoned such notions early on, accepting that though Word has a lot of features I don't use, it also has some features that I consider critical but others may consider unnecessary.

This paradox of requirements has resulted in the general market failure of Internet Appliances: They achieve 75% of the features that the average IA buyer wants, but that missing 25% is critical for users who want to use an IA in place of a "real" computer, rather than as a supplement to it. With some refocusing by IA vendors this can change quickly: The prophetic words of Marc Andreesons many moons ago (I believe in 1995) are coming true, and the reality is that the web browser is becoming your "operating system", at least as far as the user is concerned. The back-end that's doing memory management, managing the heap, moving bytes around, etc, is becoming a mere commodity that a large number of operating systems do equally well (as far as the user is concerned). The real action that people see is in the web browser that they interact with for the majority of their work. The Opera web browser is the same whether it's on the Mac, Linux, BeOS, QNX or Windows: It renders the various web standards in the same way, leading to transparent transitions between the different platforms. This paradigm holds true even one layer of abstraction higher though, all the ways to HTML itself: If a user knows how to use Hotmail or Slashdot on one browser, then they know how to use Hotmail regardless of which web browser they use to interact with it. As such, the more services that sites such as MSN (which now has full contact and schedule management, etc) offer users, the less the user needs the features locally, rendering traditional applications such as Outlook unnecessary for moment to moment basic information appliance use.

So we know that the web is becoming the application for many users, and that many users use their PCs for two simple things: web browsing and sending and receiving email. Instantly those people are catered to by an Internet Appliance because of sites such as Hotmail or Yahoo Mail. So why haven't Internet Appliances taken off? Well there are a lot of reasons: HTML wasn't complete leading to interim widely used plug-in requirements, usually facilitating a heavyweight browser with a plethora of additional software modules. Additionally IAs were pushed as all encompassing replacements for PCs, and in a marketplace that is still evolving it simply isn't possible to hit a moving target like that with any measure of success.

These are the things that I'm looking for in an Internet Appliance:

Non-gimmicky sales pitch
The IA needs a straight-forward price with no tie-ins, cross-promotions, or mandatory services.
Instant On
The Microsofts of the world would likely try putting Windows XP on it with a hard drive and a "short" 7 second boot-up : That cannot be for an IA product to succeed. The whole power of an Internet Appliance is that it is a tool, and tools need to be available to be used on a whim: If I've decided on a whim to go see a movie, I want to be able to pull up Tribute.ca in seconds to get the listing. If it's 7:58pm and 4 movies are starting at 8pm, then I will want to get to IMDB as quickly as possible to get the ratings on them to help me make my choice. If I'm considering shovelling the driveway I may want to jump to The Weather Network to double check that another foot of snow isn't incoming. All of these are spur of the moment, instantaneous information requests and I want the information ASAP.
Reliability
If a tool doesn't work reliably, it quickly frustrates and gets dumped in the trash heap. This is the general reason why game platforms are purchased by people who already have powerful computer systems: You know that you can stick in the DVD and the game will work instantly without complaining that a system dll has been overwritten or you need to update your DirectX, etc.
Web Browsing Is Key
If it can properly render HTML 4 with CSS 2 with Java support, then it'll satisfy most people. The addition of media playback should be on deluxe models and should not be forced on consumers who don't want them (again no-gimmicks: Don't force RealPlayer down our throats because Real gave you some money for co-branding). If I had the previously mentioned capabilities then I could successfully read online newspapers, participate in community message boards. I could also send emails, do my banking, update my schedule, or any of countless other web enabled applications available for users.
Zero Mechanical Noise
This means no spinning hard drives, no floppy disks, and no fans. Use low power chips and passive heat dissipation. I want this thing unobtrusively sitting in the corner of my living room, and I don't want it purring away with a fan that's bound to fail into a grinding chunk of metal irritation a year down the road. Solid state equals a better MTBF generally anyways so it's a win/win situation.
Networking Options
A standard Ethernet connection (10Mbps is sufficient), with the option of 802.11 wireless Ethernet, both of which should offer DHCP or manual configuration. Indeed, I see no reason why it can't have a "network expansion box" that accepts a special modem, Ethernet card, or wireless transceiver alternately, allowing users to switch in the future and not to have to buy extra hardware with their box that they don't need (ex. I don't need a modem and I don't want to pay for one just because it was built in an all in one box. Additionally I don't want modem circuitry consuming electricity/emitting heat just because 1 out of 4 consumers need a modem).
Integration with Existing Systems
For deluxe (media integrated) models I'd like to be able to set up `server' software on one of my full PCs (sharing a volume of MP3s for my home network, for instance), allowing the IA to browse and play back MP3s, etc. It'd also be nice, though not necessary, if the IA could print to said server machine.
Zero Maintenance
Nothing should be configurable on the machine apart from perhaps a user associated list of favourites, and the settings for the network connection. Again, it's a tool not a hobby. To be a tool it needs to provide a consistent platform for users. This also means that there is very limited security risk because there is very little to be compromised.
LCD Display
Two great features of an IA are space efficiency, and low power consumption, and neither of these is true if you bundle it with a traditional CRT. In any case, the reality is that LCDs cost less than CRTs when produced in volume, so let's beat the chicken/egg syndrome and get the LCD production ramped up.
Easy Setup/Moving
Perhaps the CPU should be built into the LCD box: I want to be able to move it from place to place with one trip and requiring one power plug, and if necessary the Ethernet cable (though secure radio networking would be preferred).
Make It A Computer Accessory Rather Than a Computer Replacement
Being a software developer I'll need a real PC for many years to come, and indeed I'd say that this holds true for most people: a real PC will be necessary for a slew of things from hunting down MP3s to doing their taxes to writing resumes. This brings us back to the first point, which is that limited functionality won't serve most people's needs completely, which is something that IA makers should acknowledge and accept rather than pathetically trying to fight with half-products full of half-features (basically IA devices that are becoming full PCs): Market these things as an addition to home networks rather than a replacement for the PC. Let's face it: The whole IA debate that has occurred up until now has largely been political fighting between the Microsoft's, Suns, Oracles, and Netscapes of the world, rather than a legitimate evaluation of what customers actually want.

What would my perfect Internet Appliance be? I'd say a 16-bit colour SVGA or XGA LCD screen, Java 1.3 capabilities, HTML 4 with CSS2 rendering, compatibility with the major graphics formats (GIF, PNG, JPG), maybe even with support for SVG for vector graphics, and 128-bit SSL/TLS: Everything else is just fluff (i.e. while nice, Flash, RealPlayer, etc, are not necessary). Whatever software is behind the scenes is totally irrelevant to me, whether it's a custom Linux kernel running Mozilla, QNX Neutrino running Opera, or Windows XP Embedded running IE 5: The net results to me will be the same presuming it fulfills my stated needs, and that is the beauty of the web platform. Hardware wise it'd be ideal if I could plug a keyed (i.e. secure) wireless transceiver into an port on my 10/100 Ethernet switch and instantly the box will have secure communications with my home network.

Of course I've considered the option of setting up my own IA device with a micro main box, perhaps with the QNX realtime platform, the Opera web browser, etc, but unfortunately without some serious engineering work that solution would require me to install a hard drive, which is something that I really would rather not do, and to find a fanless solution usually requires delving into the industrial hardware arena which usually means much higher prices than traditional for the consumer computer market place. If anyone has any ideas for a board (fanless, preferrably with onboard Ethernet), software, solid-state HD, etc, please drop me a line.


Other Articles By Dennis Forbes