Do you still need a Mac?

Fifteen years ago, the Mac had exciting features you couldn't get on other products.

Nowadays, even one of its former fans, David Cartwright CEng FBCS CITP, thinks the Mac no longer stands out in terms of desktop computers due partly to Apple not opting for the most suitable operating system.

In my youth (in the early 1990s) I was the Mac specialist in the computing faculty of a university. Although we were one of only two departments that had a decent amount of Apple kit, the list of advantages a Mac had over anything else on the market was unbeatable.

Why the Mac was fab

They had proper sound, not just the ability to beep in a monotone. They had built-in networking in the form of LocalTalk, so you could network them to printers and to each other.

You could bolt on anything from a 12-inch monochrome monitor to a huge A3 'two-page display' – and if you wanted a second monitor, then assuming you had a spare slot to put it in, you simply dropped in a second video adaptor.

Oh, and all you had to do to make a Mac a member of an IP network was drop on a copy of MacTCP – none of this faffing with semi-stable packet drivers that my PC-based colleagues had to do.

And because of all these advantages, one found that software availability – particularly for packages that played on the Mac's benefits, such as graphic design, image processing and the like – was excellent, albeit pretty expensive.

Apple pushed the boundaries, did a lot of things first, and quite frankly they ought to have had a higher market share than they did, given the pretty lousy competition afforded by PCs at the time.

Where they are today

Now let's wind forward 15 years or so and see where we are today. First of all, let's look at where they've changed for the better. Step one was to move from SCSI disks to IDE (i.e. PC-type) internal disks in the 1990s.

SCSI was, and always will be, more expensive than IDE or its derivatives, so given that we have the option to connect external devices via FireWire or USB2 (one of the neat tricks the SCSI port on a Mac used to give us), that was a sensible move.

They also realized that the underlying architecture of the Mac operating system simply wasn't workable going forward, as it really didn't lend itself well to running loads of stuff at once; the decision to move to a Unix-style operating system was a brave one, but on paper a very good move.

Not only this, but as part of the Mac OS these days, you get a shedload of immensely powerful built-in tools for manipulating photos, making your own DVD movies, and so on.

Finally, Apple has stuck to its traditional approach of making proprietary hardware against which its operating systems are written instead of doing what Windows (and Linux, and Solaris x86, …) do and allowing pretty well any hardware to be used via a layer of third-party-supplied drivers.

This is dangerous in principle for the consumer – a vendor that takes this approach could simply hike the price of add-ons in the knowledge that the customer had no alternative but to buy the vendor's own bits.

The benefit is that reliability is potentially improved because they only have to be concerned with a small amount of different hardware when making their software run. In practice, this approach worked for Apple because they made their final change - they brought their prices down to a sensible level.

What they did wrong

This all sounds pretty good so far, but Apple has a problem, or, more accurately, four problems.

Problem number one is that they've got their pricing policy about right - with one simple but massively significant exception: upgrades. Allow me to give an example. One of my clients has a Mac server running MacOS 10.3 Server. The DHCP service on 10.3 works great, except you can't allocate IP addresses based on hardware addresses.

That is, you can't make sure with the DHCP server that (say) your printer always has the same address. Hence, you have to set all your peripherals' addresses by hand. In 10.4, such address reservations are supported - and there's a pile of other nice features which the client would quite like to go with.

Except the client was quoted several hundred pounds for the upgrade. Why? It’s a minor version release. We don't mind paying for major version jumps (e.g. 9.x to 10.x) but paying for basic features, which should have been in there already, is galling.

Problem number two is that Apple's innovation has moved away from the Mac platform. By this I mean that they've brought out the most amazing piece of kit ever in the form of the iPod, a device that has quite rightly been stunningly successful for them since the day it was launched.

On the desktop computer side, though, it's getting harder and harder to see what the Mac does that you can't do with, say, a Linux or Windows PC. All of my 'why the Mac was fab' comments earlier now apply to numerous other computing platforms, but there don't seem to be any new clever things to differentiate the Mac from the competition.

Question marks

Not only this, but problem number three is that the operating system itself has question marks over it. My client just moved from his Mac OS X server to a Windows-based PC because the database software he uses runs an order of magnitude faster on the latter.

Some research has been done www.anandtech.com/mac/showdoc.aspx?i=2436&p=6 that suggests Mac OS deals abysmally with concurrent queries, yet Linux running on similar hardware has no such issues, thus pointing the finger squarely at the OS rather than the
equipment.

Other research has been done www.geekpatrol.ca/blog/106/ that suggests there's a problem with the speed of memory allocation. Rumour Control has it that the kernel in the 10.5 release will have significant modifications to try to fix these problems but in the case of one customer at least, they've missed the boat.

Finally, we have problem four: the one that was thrust on Apple. In order to stay competitive, speed-wise, they really had to move from the PowerPC processor to the Intel family. So they now make a range of computers which (a) have a similar architecture to the competition, and (b) can only run Mac OS.

There are, of course, people who've had Windows running on an Intel-based Mac for fun, and people who've had Mac OS X running on a bog-standard PC, but these are both things Apple is far from keen on: they want you to buy a Mac and run Mac OS on it.

Conclusions

So what is a Mac? It's a well-made, competitively priced box based on an Intel processor, whose manufacturer has an excellent reputation for making quality equipment. But it runs an operating system which, despite being in its fourth minor version, still causes concern and let's face it, even Microsoft tends to get things fixed by the time Service Pack 4 comes out.

So, to answer the question: why bother with Macs? Well, one of the advantages of Macs is, as we've said, that the hardware base is finite and modest; if you're writing an OS, you only have to deal with a small variety of hardware. It seems, sadly, that Apple may have picked the wrong open source operating system kernel to use.

What I'd love to see, then, is a Mac that ships with a popular Linux distribution (Red Hat, say, or SuSE) pre-installed. Good computer, plus good operating system, plus good price equals a no-brainer purchase, in my book.

What about a Mac with Windows pre-installed? Well, perhaps, but only in a niche where people care about simplicity and size (the Mac Mini is perfect for high-density offices) and let's face it, there are plenty of people who will sell you a small, quiet PC. And as for a Mac running Mac OS? Well, I don't think I'll be bothering until they show me that the OS works properly.

September 2006