Appliance PCs

Here's the deal: I'm a technical guy. I have a degree in electrical engineering, and I've worked in the technology industry for my entire career. I've built my own computer (both on a breadboard and plugging parts into a mobo). I've built a web site and coded a few web applications. But, my dirty little secret is that I hate troubleshooting and maintaining computers, now more than ever.

I dislike keeping up with patches for my OS, browsers, media players, office apps, financial software (ad nauseum, you get the idea). I never make backups, even though my PC is full of irreplaceable photos and correspondence. It bugs me that it takes five minutes to boot my PC, even though it's relatively modern. I'm frustrated when I need a piece of data when I'm at work or on vacation, and I've forgotten to print it, burn it, or upload it.

It's about time the PC became an appliance. There have been a few noble attempts in the past. Remember the Iopener? Probably not. It was a PC/WebTV-like hybrid that was most famous for its easy hackability. Essentially, it was a LCD screen with a slow processor, OS-on-a-chip, and modem built into the back. Consumers could buy it from a regular consumer electronics store for around $100, take it home, plug it in, sign on to Netpliance's ISP (required, with a monthly fee), connect to Netpliance's servers, and have regular (PC-like) access to the internet, including the web and email. All of the applications were stored in an EEPROM, so it was near instant-on. The downside, of course, was that there was no local storage, so applications could not be added, data could not be saved locally, and there was a significant lag, since everything was client-server over a modem connection.

The Iopener was a great product. I think it had the potential to be very successful, but it sadly died before its time due to, in my opinion, a flawed pricing strategy: services over-subsidizing hardware. The hardware price ($100) was far below Netpliance's cost. They relied on an overpriced monthly ISP/service fee to recoup the subsidy. There were two problems with the pricing model:

1. Hackers: If you visited the site linked above, you would quickly see that the hardware designers vastly underestimated the ingenuity of the hacker community. A hard drive port was quickly discovered, as was a way to boot from it instead of the embedded OS. For $100, hackers had a flat screen PC unlocked to use with the ISP of their choice (cutting Netpliance out of the revenue stream).
2. Pricing Model: Customers were ecstatic to get a device that offered all the features they wanted at a bargain-basement price. But, the memory of the cheap hardware soon faded and was overshadowed by their awareness that they were overpaying for ISP charges (and were locked into them for as long as they wanted to use the Iopener). Since, at least initially, there was no minimum subscription period required, they abandoned the service and sold the hardware on eBay (ironically, to the hackers).

But pricing model aside, it was a great product! As long as your computer needs were limited to the internet, you had an device with the following attributes:

1. inexpensive
2. instant on
3. silent
4. patch, spyware, and virus free

But, if you had any advanced requirements, including non-internet applications and multimedia, the Iopener was not enough. The system had to be limited to minimize the network traffic. Netpliance also made it a pokey performer to save on hardware costs.

Now we have near-ubiquitous broadband. The thin clients has been an on-again-off-again fad in the corporate world, but never in the home due to the skinny data pipe. If you throw out Netpliance's subsidy pricing model and factor in the continued tumble of the cost of MIPS, a reasonably priced, yet well-performing, thin client system is within reach.

The model I dream about having consists of a few key parts:

Hardware:

• Ultra-thin client with no local HDD, a decent amount of RAM (say a 1G or so), boot from flash, no fans to make noise, low power and instant sleep so it's always on. Preferably an LCD with all the components hidden on the back (no CPU box). Beefy processor to handle video/mp3 decoding and complex web page rendering.
• Tablet-format PC with low computing power and extremely long battery life (say, 16 hours). Wi-Fi connection to my home LAN. IE/Zilla only + VNC to the desktop machine.
• Specialized mini-headends that connect to video and audio systems
• Storage Appliance that looks like a NAS or server and connects my network to the SSP (see the second bullet in Services, below) via SDSL. Stores my personal security certificates locally and triple-DES encrypts all data headed to the SSP on-the-fly.

Services:

• Dedicated gigabit SDSL
• SSP (Storage Service Provider) with monthly fees/GB. My dedicated SDSL pipe accesses this storage plus access to the internet. Fees should be comparable in price to my CoO to run my storage locally.
• ASP (Applications Service Provider) with yearly subscriptions to my applications. I don't want to have to install/patch/upgrade Office, Quicken, whatever. If my ASP doesn't have it, I can install it to the space I'm renting at my SSP. Again, it should not be a chance to gouge me-- I should pay no more than I would have spent if I'd bought the S/W bundled with my machine or bought the box.

I'm not a hard core gamer, but then again neither are 98% of computer users. I just don't need (or want) the power and flexibility of a traditional super-client. I'm willing to pay a reasonable premium to have an xSP take care of these IT headaches for me. They're going to make profits through economies of scale and automation (like virtually every other service industry).

Everyone wins in this model. Microsoft doesn't make any more per license, but they probably get more people to upgrade, since it's transparent. Plus, everyone is fully patched (or quickly patched) so the bad PR from these security breaches is minimized. The HDD industry sells more HDDs because SSPs will run RAID-1 or 5 and buy faster models than Joe Homeuser. Telcos sell higher speed connections and can use this pipe to sell other value added services (movies, etc). Computer manufacturers sell cheaper hardware into the home, but better servers into the xSPs (plus, smaller audience = cheaper marketing, cheaper service). They also can focus or truly unique applications for their machines instead of just reving their mobos to use the latest processor.

These are not unique thoughts-- thin clients have come in and out of fashion for decades. But, I think we are at an important inflection point in the PC industry right now. As a member of it, I am concerned that the purchase driver for the home user is changing-- instead of the MIPS treadmill, we will need to focus on new user models and features to continue the pace of the replacement cycle.

Those that can adapt will be wildly successful; those that cannot will go like so many before them. It's an exciting time-- I look forward to what the future will bring.

Update (1/3/2005): Two relevant articles on this topic, one over at The Inquirer and another blog, both of which are right in line with what I've been thinking.