Columns

The laws of scaling are complex and variable. Early paintings of babies depicted them with proportionately small heads, despite common knowledge that their heads are disproportionately large. The physiological demands of shrews result in their extraordinary heart rate of 800 beats a minute compared to the different challenges facing elephants whose hearts lumber along at about 30 beats every minute.

Surface area, a major determinant of heat loss, is proportionate to the square of body length, while body volume - key in determining the amount of metabolic heat generated - is proportionate to the cube of body length.

If anyone at Apple thought that the transition from Tiger (body length 2.5m to 3m) to Leopard (1m to 2m long) would follow a simple scaling rule, they must now be wondering which size cat to choose for Mac OS X 10.6. With each new major release of Mac OS X generating ever-larger sales and increasing Mac OS X's market share, Apple would not appear to be drifting into danger.

But as Microsoft found out: the more you sell, the greater the outcry over problems and failures. Leopard's initial release bore all the hallmarks of a product shipped by date rather than maturity. This ensures that some users will be disappointed and a few suffer embarrassingly serious issues. If it was as mature as being 99% satisfying, then maybe as many as 20,000 of the early adopters would quickly find their enthusiasm turning sour.

Even Leopard's first update left overt

ADVERTISEMENT

bugs to irritate users day after day. Glitches rendering mundane interface elements on different standard display cards are obvious - and inexcusable. Fresh bugs in the Finder, such as problems ejecting disk images, show the emphasis on deadlines rather than quality. But compared to the pervasive problems when you try to connect to Active Directory servers, these are merely trivial nuisances.

If you were unfortunate enough to equip a new learning centre with Macs running Leopard and tried to hook up with your Windows servers, you would either have to downgrade those systems to Tiger or wait until the remedies in 10.5.2 or later.

I am not saying that Leopard is duff - even in pre-release use it was surprisingly robust. Features such as Time Machine make it a compelling upgrade provided that it does not pose new problems. When Macs were an acquired taste, purchased at huge cost from specialists, and we lived with the turmoil of major new system releases, launch by date may have been acceptable. But now that ordinary people are walking out of Apple's frenzied boutiques with box after box of commodity Macs, Apple needs to adapt to its new high-revenue world. We could also excuse periodic jolts in the early part of our journey into the brave new world of Mac OS X.

Part of the new scaling problem is that a small proportion of Mac users now constitutes a large body of people only too ready to bad-mouth, bitch and blogsult. As Apple's user base grows, so pre-release testing needs to grow both in terms of the number of testing sites and the duration of the prerelease phase. Given that Leopard could have enjoyed only about three months of such testing - predominantly among developers whose priority was to find and fix incompatibilities in their own products - it is not surprising that certain hardware set-ups perform rather poorly under Leopard.

So for 10.6, please let's not pre-announce the release date. Launch only when the bugs are thoroughly shaken out.

Then as the Mac's user base scales up, user satisfaction will increase disproportionately.