Last month I focused on the negative aspects of Apple?s new products to make a point. And based on the feedback I got, the point I appeared to have made is that there are an awful lot of Apple folks who really don?t like anyone who disparages their beloved products.
While there are a number of trade offs that Apple has made for the Mini that I don?t think were necessary, the product is actually closer to what I think the consumer market is actually looking for in a desktop offering. No, Steve Jobs didn?t just arrive at my house with explosives, or worse, attorneys. I just think it is time I acknowledged that there are a couple things I really wish the other vendors would learn from what Apple has done.
The Windows/Intel PC
The traditional desktop PC running Windows is designed to be modular in that you can change out, relatively easily, major components. Graphics cards, drives, memory, input/output cards, cooling fans, and even power supplies can be replaced and upgraded relatively easily. Even the processor can be upgraded and replaced. Buyers can configure to order boxes with components from a variety of secondary vendors and have someone else build the perfect computer.
The problem is the average buyer, consumer or corporate, doesn?t use most of this flexibility. Most buyers not only pick a standard configuration, from what I can tell, better than 95% never open the product up themselves, ever, after the product has been purchased and installed. And the majority of people who do open their product up do so in order to facilitate a repair; and even that is done by someone else most of the time.
This flexibility has several costs. The products have to be larger, they are more difficult to cool, they generally aren?t particularly attractive (particularly in home or desktop settings) and they are more expensive than they need to be. And this cost hits several areas, it causes an increase in finished goods inventories, increases manufacturing cost, and an increase in costs associated with prepping the product for sale.
The resulting lack of standardization also results in higher failure rates and a generally less reliable offering than what you see from Apple or any embedded systems vendor we have ever seen.
The ?Why? behind the Unneeded and Unwanted Complexity
I blame the reviewers. In our industry those of us who review hardware do not represent the average user. We like to take things apart, we like the idea of upgrades and, many of us including me, like to build our own machines. As a result, when we are reviewing products and making recommendations, we favor the configurations which have slots and traditional cases over more appliance-like products. In fact one of the first things we will downgrade a PC for is not having enough slots even though the average user will never, ever, use that extra slot.
This isn?t reality and it has resulted in a generic hardware platform that most of our friends and relatives openly hate. To Apple?s credit they don?t listen to these reviewers and build what they think is right, unfortunately they have dropped so much in market share that their example has no where near the power of us misguided reviewers and the end result are products that few, outside of Apple customers, some power users, and gamers, get very excited about.
In my opinion, when we are reviewing general use PCs, we should focus on the out of box experience, adequate performance, ease of use, and reliability. We should stop setting the bar at the hottest new product and, instead, set it at the best user experience. If we did we would drive the market to a better future.
The Appliance PC
For years I have surveyed buyers and they have consistently said what they want in a PC is something that is drop dead easy to install and use. Basically they want something much closer to a typewriter than a mainframe. When asked they would rather not have it under their desk because they need to access the drive(s) but most desktop products take up too much desktop real-estate, and, particularly in the home market, most would like the product to be at least somewhat attractive.
If you look at how people actually work, they often live within a few set applications: a word processor, a browser, a few utilities and sometimes one or two vertical applications. In addition, the hot platform right now is a laptop computer where virtually all of the market growth exists today. Laptops are purpose built machines. They have PCMCIA slots that are seldom used and are accessorized, generally, through external ports.
We actually have a number of examples of Appliance PCs in the market today. TiVo and Replay TV boxes are appliance PCs, POS Cash Registers are Appliance PCs, thin clients machines like those sold by Wyse and HP are Appliance PCs, the XBox/PS2/Nintendo Cube are incredibly powerful and inexpensive appliance PCs, and the joint AMD/Microsoft Personal Internet Communicator is probably the most advanced product in this class targeted at general use currently on the market.
The two product classes closest to the consumer today, for general use, are Apple?s consumer lines (iMac and Mac Mini) and the HP Digital Entertainment Center version of Microsoft?s Media Center offering. Of the two platforms the Media Center still has too much general OS in it and that extra complexity works against the user experience and inflates cost which is why HP announced a similar Linux based Media Hub product at CES.
It does seem to me that we are drifting closer and closer to a PC appliance and maybe we should stop fighting this trend and get behind it.
The Operating System
I am convinced it is time we also rethink the operating system. Windows was largely based on OS/2, which came out in the mid 80s, and both Linux and the MacOS X are based on UNIX which is even older. All three platforms are based on the way the world was in the 80s and both hardware technology and users have changed dramatically since then. When these operating system’s were new, users were still relatively technical while now the platform has to deal with a much lower understanding of the core technology as users increasingly abstract these old cores behind applications.
Particularly when you look at the security exposures that exist today and the inability of traditional virus companies to get signature files out quickly enough and the massive problems all three platforms are having with patches I believe it is time to take a step back, lose the religion, and see if we can?t do better. In fact, I think we should probably do this about once every 20 or so years.
The closest thing I see to what might make sense in today?s world is a combination of embedded systems with connected applications on top of virtualized hardware. Virtualized hardware is where you separate the actual hardware with another layer which emulates something else. The most popular products today are software offerings from VMware and Microsoft (as part of their Connectix acquisition). However both Intel and IBM are showcasing hardware based virtualization and, in Intel?s case, the product?s code name is Vanderpool and hardware virtualization has the potential to dramatically change our PC experience. At the very least it would make driver problems, for most of us, a thing of past history.
One of the ways virtualization could be implemented is so that each application would have its own operating system wedded to it and lie on top of virtualized hardware optimized for that operating system/application bundle. This would allow each application to work securely and not unduly compromise another application, it would allow a much higher degree of application portability, and it would remove much of the legacy problem that made the move to the current MacOS so painful and is both causing Longhorn, Microsoft?s next OS, to slip and keeping it from becoming all it could be. Effective viruses would be a relative nightmare to write in such an environment and making it harder for virus writers is always a good thing.
Intel?s Vanderpool technology, as it matures, will help make a new platform like this possible and hardware virtualization has a fraction of the performance cost that software virtualization has. It also promises an era of greater flexibility because one of the things Intel is showcasing is that they can emulate the PowerPC, just as one of the things IBM Microelectronics is showing with the PowerPC is they can emulate the x86.
Having looked at the various development efforts under way it is clear to me we are actually on this path, how fast we get there will depend greatly on the decision makers ability to see the goal clearly. Right now, we seem to have an eyesight problem that has resulted from silly arguments over which platform is better even though we should conclude that all of them have serious problems. While, given the clear threats to both revenue and security, we should work together for a better future. We seem unable to do so, but maybe it is time that changed.
But back to the core point, the market wants an appliance experience with their PC technology. Both the iMac and the Mac Mini do the best job of this today, however there are a series of Linux based solutions that are coming that look at least as good, and Microsoft?s embedded division actually has the closest thing to this in an OS solution today though it is generally targeted at industrial use. Maybe, once again, Apple is showing the way, it will be interesting to see if they, or others, get the most benefit from this direction this time.
