- by Paul Murphy -
About two months ago I looked at the cost of the Macintosh relative to Dell PCs and discovered that not only are Macs cheaper than PCs once you upgrade the PCs to rough comparability, but the PC line is narrower than Apple's with Dell offering nothing to compare to the 17 inch powerbook, the X-serve/RAID combination, or Apple's cinema displays.
What seems to happen on pricing is that Apple's inclusion of multi-media capabilities generally missing from the PC skews its price advantage toward the high end. Compare Apple's dual G5 to the top of the line Dell dual Xeon Precision workstation and the PC has most of the features, but costs a thousand bucks more than the Mac. The two lines are closer to even in the mid-range, where the Apple 15 inch PowerBook is only about $180 cheaper than the nearest comparable Dell 15 inch Inspiron. At the low end you get the opposite effect: the stripped Dell 2400 is $350 less than an eMac - but is a Windows/98 class machine that lacks the processor power and memory needed to run Windows/XP effectively and ends up significantly more expensive than the eMac if you upgrade it to match the Mac on a feature basis.
Last month I looked at the performance issue to find something almost equally surprising: if you strip away the effects of software and market differences by looking at "whole box" usage in GRID style super-computers, desktop Macs turn out to outperform Dell's best dual CPU servers by 30% while the X-serve blows them away by 50%.
Notice however, that raw processing power doesn't translate directly to perceived desktop performance. The hardware sets a performance limit -the application can't go faster than the hardware- but how close your system gets to that limit is determined by the quality of the software. Unfortunately, developers selling a PC product into the Macintosh market generally seem to put little or no effort into understanding the Macintosh hardware. In fact an easy majority apparently opts for little more than use of a Macintosh cross compiler on code that's been extensively optimised for Windows on x86.
Obviously that produces astonishingly inefficient Macintosh code and is itself a striking commentary on how those vendors see the Macintosh software market - and it's that market I want to look at in today's column.
At a gross level, software availability provides a measure of that market. So, remember Clinton's argument that his claim: "I did not sleep with that woman", wasn't a lie, because he was awake the whole time? Well, in that same vein you could claim that the Mac has considerably more software than Wintel simply because most Wintel software will run in emulation on the Mac, but hardly any Mac software will run on Wintel.
As truths go, that's a lie - but it masks an important point about the software market: Microsoft follows Apple in interface and operating systems development, but it's the Macintosh software developers and users who have been inclusive about Wintel applications ideas.
Look for real numbers and you draw a blank: you see the number 100,000 tossed around a lot for Wintel applications availability while Wikopedia.org lists only about 12,000 individual pieces of MacOS X software, but nobody really seems to know either how many separate applications exist for either platform or how they compare.
A theoretical approach to clarifying application availability would be count the unique function points in each piece of software and then value these points according to things like support availability or the extent to which they use native facilities like quartz. In practice something like that probably can't be done but even just thinking about applying a metric of that kind to a few key packages leads quickly to another discovery about the Mac/PC relationship that may seem counter-intuitive, but is actually a logical consequence of Apple's OS and graphics leadership. Specifically: where vendors selling to both markets make an effort to utilize unique Mac facilities properly, the Mac software tends to be more functional then the Windows product.
Consider, for example, Microsoft Office. Ever since Office 5.0, the Mac version has been about a generation ahead of the comparable PC product and Office 2004 for Mac Professional is no exception. Not only does it make good use of unique Macintosh features such as the PDF based display and underlying Unix engine, but it also leads Office/XP Professional in features. Here; for example, is what PC Magazine had to say in its review of the first release last spring:
The center piece of the suite is the new Project Center, an exclusive (for now) feature that helps you organise both large and small projects. The Project Center resides in Entourage (the suite's e-mail and calendaring app) but is accessible from Word, Excel, and PowerPoint. Simply click the Project Center button and then click to create a new project. You can then name the project, assign a deadline, create tasks within it, and create links to all related notes, documents, and contacts.
At the same time it's not unusual for the Mac package to either be missing something found in the PC version or to have it included but in such a way as to make it unusable.
To cite Microsoft Office again, the integrated video conferencing and some of the XML extensions used in Office/XP Professional are missing from the Mac version but you can access those facilities via the (included) PC emulation software.
In both cases, however, what appears to be going on is that Microsoft is selling the PC community an interim solution, or "hack," it knows Mac users would never accept.
"Live Meeting" on the PC, for example, is rightly considered very cool in that environment, but looks utterly hokey when put against iChat. Similarly the Mac community is sufficiently knowledgeable about SGML and related document processing issues that Microsoft's attempt to use XML as a web programming language would most likely arouse only derision and disbelief if marketed to Mac users.
Nevertheless, there are clear are "holes" in the Macintosh software market - areas where there is significantly more software variety and therefore functionality on the Wintel side of the ledger. That's true, for example, in both games and business intelligence processing on the commercial side and in the apparently endless variety of home grown "client" software used in financial and related business processing.
There are reasons these holes in the Mac market developed. Thus most of the PC games failed to develop on the Mac specifically because the demographics of the two markets are so different. Those developers, for example, who focused on replacing the player's imagination or intelligence with interactive graphics simply found no market among Mac buyers - most of whom are older and/or better educated than the people those products sell to.
The market gap in business inteligence or report processing has a more complex history. Until about 1995 most of the advanced commercial products in that field ran only on Unix. In 1994, for example, Statware offered advanced statistical processing and quality control software for Unix/CDE of the kind its successors now offer on Windows clients. What happened in between was that progress almost came to a stop because by late 95, "everybody knew" that Unix was dead and that the Mac had no future --meaning that venture capitalists, bankers, and corporate financial managers almost unanimously made Wintel focus a prerequisite for development funding. Thus even companies with sucessful Unix or Mac product lines faced enormous presure to focus on NT - Microsoft Great Plains, for example, started on the Mac but became a PC developer because the people it sold too all knew that Apple was history.
Today we're seeing that situation reverse, with a whole new crop of software funded on the promise of Linux as politically correct Unix and easily portable from there to the Mac and other Unix environments like Solaris. In fact, look at the really niche markets where tomorrow's software revolutions are being piloted now - markets with a few dozen specialist user/developers worldwide - and you see that the Mac stands to benefit tremendously from Apple's adoption of Unix. Take almost anything new done in academia since mid 2001, and it's likely to have been done first on Linux or BSD --meaning that porting it to the Mac (or Solaris) is trivial but porting it to Windows will be hard.
As a result the wave of academic NT software started in the mid and late ninties and from which so much of today's advanced business software for Windows came is, although largely still in use, now atrophying from its roots forward. It was that software, of course, which led to the dominance of home grown NT "clients" in business - a situation akin to that described to children in the rhyme starting " For want of a nail, the shoe was lost. For want of a shoe.." and which is now reversing with an explosion in the worldwide development of nails for Linux.
Cost can be treated as another "dimension" of availability. The myth here is that Macintosh software costs more than PC software, and like other availability myths it's both true and false. In general the PC focussed companies either charge more for Mac software licenses than they do for PC licenses or use differing marketing methods which mean that the Mac user pays more even if the list price is nominally the same.
The PC hardware companies, for example, usually bundle Microsoft Office at a discount to list price; Apple doesn't. As a result Office usually costs the Mac user more, even though Microsoft's list price is the same.
Notice, however, that this applies only to commercial software aimed primarily at the Wintel market. If it made sense to average cost across all applications, that average would be lower for the Mac because it can use just about all Unix freeware while software first designed specifically for the Mac tends to be priced at or below the typical PC level.
More importantly, Apple more than levels the playing field for business and educational use by managing its server licensing very differently from Microsoft. In Apple's world, the ten client restriction on the $499 MacOS X server license applies "only to simultaneous file sharing services for Mac and PC clients" and a lot of unrestricted standard applications either come bundled with the OS or can be freely downloaded for use.
In the Microsoft world client licenses apply to the interaction between specified server software and specified individuals or devices. Thus the basic Windows 2003/XP server costs $1,199 with 20 process client packs at $799 - but additional software costs extra. For example, adding fairly basic messaging and collaboration services requires Microsoft Exchange Server - at $1,299 plus $67 for each client after the first five.
Typically, therefore a company with fifty Mac users will pay around $80 more for each copy of Microsoft office, or about $4,000 more in total than its PC using competitor. On the other hand, that PC using company will pay Microsoft another $1,298 for the discounted five user Microsoft's Windows Small Business Server 2003 license, and then cough up $4,142 more for an additional 45 client access licenses --meaning that, in the end, the company using Macs spends about a thousand bucks less than its PC using competitor.
Perhaps the most interesting difference between the Windows and Macintosh software markets appears in their larger scale structures. Take any applications area where both play, and you'll usually find significantly more Wintel software "titles." But, take a closer look and you'll almost always see that one or two players own the Wintel market while all the little guys whose numbers give Wintel its claimed edge in having "more applications" are functionally irrelevant - something that's not true in the Macintosh market where market share tends to be far more evenly split among suppliers.
More subtly, the Wintel products all pretty much do the same things in the same ways - you can still buy WordPerfect for Windows, but it now uses icons and sets tabs pretty much the way Word does. In the Macintosh world that isn't true, not only are there a half dozen genuinely competitive word processors, but they do different things differently. Nisus and Mellel, for example, both do things Word doesn't.
To stretch a genetic analogy about species diversity: you can visualise the Mac market as a southern reef full of brightly colored varieties and the Wintel market as an Arctic deep populated almost entirely by nearly invisible plankton and a few whales.
So what's the bottom line? It's a mess with no simple answers. If you look just at current software availability, there are clear holes in the Mac market - particularly in action games and some business processing. Outside of those areas, however, the Mac software markets offer a greater range of choices and, including available Unix freeware, both more functionality and lower cost.