$200 Linux PCs

Gekko and Webslacker were the first of many to tell us about the stir over at ZDNet, which is reporting on the arrival of sub $200 PCs due Q1 2000. These new desktops from Taiwan's Tatung come in eye-catching candy colors a la Apple's iMac. Tatung has opted for Rise and Cyrix K6 chips instead of Intel Pentiums, and a CD-ROM drive is an option. One wonders with the increase in the cost of DRAM how this will impact the price?

Comments

The Mystery of the Missing Killer App

[Clairvaux]

About two years ago, when PC prices began their long, unrelieved slide to levels today which would have amazed the PC buyers of the early 90s, I did some hard thinking about whether this was a whole new ballgame and what factors were at work here which I later turned into an unpublished article. Let me restate a few of the relevant points here.

NOTE: this turned into a GIGANTIC post, but I strike a blow for Penguin at the end, so hang in there! ^_^

* * *

First, pricing is a result of supply and demand forces. This is axiomatic. Thus when I observe some kind of systematic trend in prices, I infer that there are systematic changes taking place in demand, supply, or both.

On the supply side, we have Moore's law and it's cousins perpetually at work bringing down the cost-per-unit of processing power, storage, bandwidth, and every other dimension of computing. There are some irregularities to this which I'll address later but fundamentally, the past 20 years have been characterized by remarkably steady progress on the supply side.

If you accept this, then the relative stability of the average selling price of a PC between its advent in the early 80s up until roughly 1995 meant that there must have been concurrent, offsetting changes on the demand side stimulating demand for the constantly increasing supply of computing-units.

We even have a name for these demand-side stimulants. We call them Killer Apps. I my opinion, these Killer Apps are so significant to the history of the PC that one can even slice that history into phases characterized by the Killer App of the day.

I. Early DOS Era (1981 - 1984)

Spreadsheets and word processors! The first Killer Apps that drove initial sales of the PC, pushing it into the corporate mainstream. Spreadsheets, running in text mode under MS-DOS, supported several generations of hardware upgrades, because the performance improvement from each upgrade was visible and appreciable -- and contributed to productivity.

Typical CPUs: 6502, 8086, 8088 Typical RAM and storage: 256K RAM, 180K floppies, 10MB hard disks

II. Late DOS, early GUI and Graphics Era (1984 - 1989)

With growing sales of PCs came the development of lots of other uses for them, and the advent of rapidly improving graphics. The Mac GUI and page layout software like Aldus Pagemaker were acknowledged Killer Apps. And GAMES baby!

CPUs of this Era: 80286, 80386, 68000
Typical RAM and storage: 640K RAM, 100MB hard disks

III. Late GUI and LAN Era (1990 - 1993)

As the set of tasks for which one could employ a PC grew, users began needing to easily switch between applications. Several methods were developed that allowed the PC to run concurrent programs.

The GUI imposed a significant system overhead, once again supporting demand for several generations of CPU progress. Furthermore, all the software for PCs was eventually rewritten as graphical Windows applications to facilitate multitasking and incorporate new features that the GUI environment made possible. The new code was fatter, driving demand for more/faster RAM and disk space.

This era also saw the emergence of widespread local area networks (LAN) allowing shared storage resources and launching the client-server paradigm. This segmented the PC market into client desktops and network file servers. The PC-based server was a Killer App that definitely pushed demand for computing power and divided the market into today's enterprise/desktop segments.

CPUs of this Era: 80486, 80486DX2
Typical RAM and storage: 2MB - 8MB RAM, 400MB hard disks

IV. The 32-bit Era (1994 - Present)

The next Killer App that continued this saga of seemingly perpetual demand for more hardware performance was the advent of 32-bit GUIs in the form of Windows NT and Windows 95. Applications ONCE AGAIN being rewritten for the new OS consumed even more system overhead and incorporated yet more functionality, creating another cycle of appetite for better hardware.

CPUs of this Era: Pentium, Pentium Pro, AMD K5, K6
Typical RAM and storage: 16MB - 32MB RAM, 1GB hard disks

V. Today: The Internet Era (1996 - Present)

Today's Killer App which drives sales of PCs is the Internet.

Unfortunately for the companies that had planned on continued geometric demand growth for CPU speed, RAM, and drive space, this latest Killer App doesn't require those things. Applications for browsing the web, chatting, and communicating with e-mail, in their existing form are very "light" applications which don't require supercomputing horsepower.

Meanwhile, nothing new has emerged in other areas that offers appreciable functionality for mainstream users to carry the upgrade cycle further. Spreadsheets, word processors, presentation builders, and so on have apparently attained a "functionality saturation point." I'm sure we've all heard the comment, "What MORE can a word processor do?"

Clearly, the situation bodes ill for companies such as Intel, Micron Tech, and Western Digital, who have benefited from past generations of killer apps, but look to be out in the cold this time around. In the past, killer apps doubly pumped companies like these: by (a) shifting demand curves outward, and (b) simultaneously reducing price elasticity of demand, tilting the demand curve toward the vertical. A more detailed discussion of this effect is in a separate article; here, suffice it to say that these effects combined to heavily reward vendors of these products with each advent of a new killer app.

CPUs of this Era: Pentium MMX, Pentium II, AMD K6
Typical RAM and drive space: 32MB - 64MB RAM, 2GB - 4GB hard disks

So -- are you still with me? Amazing! ^_^

The way I concluded this article two years ago was to predict a recession in the PC industry unless a new Killer App emerged. Well, I was WRONG. No new killer app surfaced, but the PC industry has been booming.

Why is this? I believe it's because although the demand level for stuffing a ton of power and storage into a single box has plateaued, the demand for the number of boxes has continued to climb, as legions of new, first-time buyers, attracted by the "network effect" enter the market. Geoffrey Moore would describe them as "Late Majority."

Supposing that this continues for a while, and average unit selling prices of PCs continue to decline, I see a couple of interesting consequences of this. The first has to do with Milton Friedman's theory of component elasticity. This theory is very simple so don't worry if you didn't make it through Econ 180. The relevant part of it states that those parts of the whole product which are a big chunk of the cost of that product will be most sensitive to changes in the market for that product. Here's how it's relevant to the PC situation. Back when the average PC cost $2000, a $100 license for the OS was only 5% of the total. So nobody worried too much about the price of the OS.

HOWEVER -- for a PC that costs $400, the OS is now the BIGGEST COMPONENT COST. Vendors have a tremendous incentive to try and reduce that cost ... and guess what they're all thinking about right now?

LINUX.

Excellent...

[Xuff]

Ahh, finally, the bathroom computer comes home. No more printing /. to keep me entertained during #2, now it's all gonna be live!

Cheaper by the sixpack?

[A nonymous Coward]

Now that would be something. Maybe a hacker's dozen? Buy 16, get the 17th for free.

O rbuy as many as you need, and another for a spare, so when one of them dies, you just quickly configure its net address (or use DHCP in the first place) and you're back on line.

Heck, get two for each room, that way you have some hot standbys!

Yeh, I see some potential here.

--