Last week’s Monday Note on the Macintosh endgame generated more comments and email than usual, which prompts me to pursue the topic, this time with regard to the history and future of the Macintosh operation systems.
“Operating systems are like the tax code,” goes the industry lore. “Each year, new lines of code are added—code that tells us how to allocate resources, how to deal with choices, what to do when an exception is encountered. And both are continually patched to deal with errors and new circumstances.”
A prime example: Apple’s macOS.
In 1984, the Mac’s software engine, which included an AppleTalk network stack and a LaserWriter driver, ran on a single Motorola 68000 CPU and needed just 32K of ROM and 128K of RAM.
Within a decade, the Mac had been ported to the ill-fated PowerPC (designed by the Apple-IBM-Motorola alliance), and the System 7 operating system, while small by today’s standards, had become bloated:
While System 6 could fit comfortably on an 800 KB floppy disk …System 7 needs to tighten its belt to fit on a 1.44 MB floppy. Users with a floppy-only pre-FDHD SE, Plus, or the dual-floppy education-only version of the LC needed to buy a hard drive or be left out in the cold.
When Steve Jobs and his band of NeXT computer scientists arrived in 1997 in a “reverse acquisition” of Apple, the first order of business was to modernize the Mac’s software engine. The OS team’s most important contribution was in skillfully and elegantly sliding a new Unix foundation under the existing apps, thus making the Mac both more reliable and more expandable. But it wasn’t quite powerful enough for Jobs, so the team then ported the Mac to the x86 architecture. (Porting a commercial operating system to a new CPU is almost unheard of. Apple did it twice.)
Today, macOS is a fully-grown computer operating system that is pleasant, fast, and flexible. But it’s also enormous—RAM and disk storage requirements are measured in gigabytes—and it isn’t exactly bug-free. An ex-Apple acquaintance recently told me there are something like 10,000 “open” bugs on an on-going basis. The number that are urgent is, of course, a fraction of the gamut, but like any mature operating system, macOS has become a battlefield of patch upon patch upon patch.
It’s the fate of all operating systems to grow more function-rich and consume more hardware and engineering resources. We always want more and the very generality and flexibility of operating systems makes it hard to resist the call for added features.
These growing pains occasionally spur software directors and VPs to fantasize about a “lite” version of their OS…but, look around, do you see any? Weight-loss discussions ultimately founder because there’s always some must-have feature, trussed for the guillotine, that’s saved in the eleventh hour…and then another, and another.
But then there’s the counterexample of the iPhone.
When the Apple smartphone project started, the key decision was the choice of software engine. Should Apple try to make a “lite” version of OS X (as it was then known)? Go in a completely new direction?
It appears that a new direction may have been tempting. At the time that Apple’s smartphone project began, an Apple employee and former Be engineer offered Palm Inc. $800,000 for a BeOS “code dump”—just the code, no support, no royalties. The engineer was highly respected for his skill in mating software to unfamiliar hardware; BeOS was a small, light operating system—draw your own conclusion. Palm, which had purchased Be a few years before that, turned him down. (I learned this when I was asked to become chairman of PalmSource, Palm’s software spinoff)
After what must have been animated discussions, Scott Forstall, part of the NeXT cohort, prevailed. He convinced Jobs to let him make a version of the Mac’s operating system for the iPhone. This was no small task. Not only was OS X huge, it was designed to run on powerful and power-hungry PowerPC processors (the iPhone decision was made before switching the Mac to Intel). The iPhone was going to use a tiny ARM 412 MHz processor with only 128 MB of memory.
No surprise that many of us were skeptical when Jobs announced, in Jan. 2007, that the iPhone would run on OS X. We thought his use of “OS X” was, to be polite, a rhetorical flourish, a slight of hand. But, no, when inquiring geeks got their hands on the device, they indeed recognized the OS X core services.
The engineering feat performed by Forstall and his team and its enormous economic consequences can’t be overstated. They changed Apple and an entire industry, inaugurating the Smartphone 2.0 era.
In order to run on less-than-muscular hardware, the team had to leave a lot on the editing room floor. There would be no cut-and-paste, no apps, no accented characters, no (user-accessible) multitasking or file system… But year after year, just as with any OS, functions were added and revealed. With a 2.34 GHz processor, up to 3 GB of RAM, and as much as 256 GB of “disk” storage, the OS X descendant now called iOS is both significantly smaller and simpler than macOS, but it’s robust enough to tackle many tasks that were once the province of a “traditional” PC engine.
iOS managed to succeed where “lite” versions of mature operating systems failed, and it succeeded on a much larger scale than existing personal computers. The ubiquity of iOS devices ensures the operating system’s future, but it’s not just the higher unit volume that’s attractive: iOS is younger and nimbler than its noble and worthy macOS forefather. As I stated last week, it will assume more and more of the duties of Apple’s historic Macintosh.
But this doesn’t mean Macs will disappear any time soon.
In the first place, the Mac is still needed as a development platform. You can’t run Xcode, the iOS app development and debugging environment, on an iPad. This is Steve Jobs’ trucks versus cars metaphor which he introduced at the 2010 D Conference: There will always be trucks; even if you make electric bikes, you still need trucks to feed your manufacturing plant. (Recode has published a set of interview highlights with Jobs; they’re worth the watch, and not just for the nostalgia).
Also, Apple has clearly stated its preference for cannibalizing its own products rather than letting others do the dastardly deed.
So what’s the future of personal computer shapes and sizes?
A couple of friends of mine, unknown to each other, posited a scenario in which we carry our smartphone around with us and then, when we reach our office or home desk, plunk it down next to a large screen and keyboard+trackpad combo. The mobile device connects wirelessly and, presto, the comforts of desktop computing. Surely, they both said, this is just a matter of software and time.
The notion jogged a memory: I recalled seeing just such a set up at the Stanford Shopping Center Microsoft Store last year. Microsoft’s Continuum connects a Windows Phone (albeit not wirelessly) to a large screen, keyboard, and mouse:
When I returned to the store this week, the display was gone. Still, the idea is in the air, and the two individuals mentioned didn’t know about the Microsoft product.
The Microsoft implementation might be too kludgy, or immature, or the concept itself could just be a doomed Rube Goldberg fantasy. Back to reality, we’re likely to pick up fresher clues on iOS’ direction when new iPads show up, probably next quarter.