50 Years In Tech. Part 11: Getting The Mac Out Of The Ditch
by Jean-Louis Gassée
When I landed in Cupertino in May 1985, I had a few simple ideas for short and medium improvements that would help sales, and was happily surprised by the support that most of these ideas got inside the engineering community.
I count 30 faces around the conference table. That’s one too many. As Apple’s newly-appointed VP of Product Development, according to a temporary org chart, I start with 29 direct reports… who’s the interloper?
“Ah, I’m your Human Resources representative. Given the potentially…tendentious nature of this meeting, HR felt it would be prudent to attend as an aid to communication.”
Indeed, the meeting looked to be… contentious. I have to harmonize the Mac and Apple ][ engineering groups into a single euphonious choir. The two factions are openly disdainful of each other; some of the attendees aren’t going to be thrilled with the new org chart.
As it turned out, HR’s presence was helpful, although not as intended. As I looked around the room, the barely-hidden smirks and eyerolls (on both side of the table) were reassuring. However much these engineers despised each other and feared for their positions, we had a common target. Engineers liked to call HR the Thought Police, KGB, Gestapo… and insisted these were meliorative nicknames. To be fair, Apple HR would prove a boon when it came to helping my family and me get settled. Unfortunately, they also tended to create political turmoil as they overplayed their power-behind-the-throne role at the highest levels of the company. Even more regrettably, I allowed myself to get sucked into the turbulence.
Having formed a (somewhat) unified engineering team, it’s time to get down to business. May, 1985: Apple ][ sales are falling; the Mac has yet to take off. We need to make some changes, pronto, that will attract new customers and keep the old ones coming back.
First, I felt that the Apple ][ family was becoming needlessly complicated. As a multimedia machine, the proposed Apple IIGS (“G” for graphics, “S” for sound) was undeniably superior to earlier Apple ][s and the current Mac, but it used a virtually unknown processor with a doubtful future and offered so-so compatibility with earlier Apple ][ models. Why are we going down this path?
I paid a political price for my desire to deemphasize the product. The Apple IIGS was the favorite child of a senior exec who took my view as a personal attack (admittedly, I could have expressed my opinion a bit more diplomatically). That Bill Gates refused to let Microsoft adapt its apps for the GS should have helped my position, but the ego damage was done.
On the Mac side, I suggested we needed to do three things: Implement a few quick improvements that will make the Mac feel more muscular, design an “open” Mac with interface card slots and color display capabilities, and slide a robust operating system kernel under the existing Mac OS.
The “muscular improvements” made their way into the next Mac offering, the Macintosh Plus, announced in January, 1986. The most important new feature was the incorporation of a SCSI connector that let you plug in an external hard drive. Less conspicuous were double-sided floppies, a larger ROM, and a doubling of internal memory (RAM) to 1MB — an excess that Bill Gates dissed at the time, arguing that nobody needed more than the DOS PC’s 640K upper limit. The reaction to the Mac Plus was positive, a welcome stopgap. Inside Apple, we liked to think that Gates’ reproof made the machine more popular.
I expected stiff resistance to “opening” the Mac, but found none. To the contrary, I found myself offering an argument against my own position: “I realize that Steve Jobs strenuously opposed such features…”. “He sure did, but that was him and then.”
As a cri de guerre, I got a suitable license plate for my car…
…and handed out the obligatory t-shirts:
The team quickly settled on a bus standard and went off to create the Macintosh II. Partly based on the design of the ill-fated Macintosh Office File Server, the Mac II shipped less than two years later.
Unfortunately, the suggestion that we introduce a kernel into the Mac OS was completely unsuccessful. Here, a brief definition may be helpful — with my apologies to genuine computer scientists for the oversimplifications.
Today, all general-purpose computer operating systems are regulated by a kernel, a base layer of software that sits right above the processors. The kernel is a combination of traffic cop and nanny, making sure that applications play nice with one another; that hardware resources, such as memory, are properly protected and shared; and that traffic inside the box and to the outside world is coordinated and prioritized.
The original Mac didn’t have such protection, partly because Jobs insisted that the Mac have the smallest possible system software footprint and RAM size (visiting from France, I was present when team members pleaded with Jobs to let the Mac have 128K RAM as opposed to the 64K he had originally dictated). I’ll venture that Jobs’ decision to “go light” with the Mac was a result of the Lisa. The Lisa had a proper, home-grown multitasking OS that was far ahead of its time. Unfortunately it was also ahead of the day’s hardware and was both sluggish and buggy
A lack of a kernel does make the footprint smaller and can make some processes feel more responsive, but at a cost. Lacking a traffic cop and nanny, every app runs in “superuser mode”, entrusted with access to everything inside the machine including the code and memory of other apps. Your app has to tread lightly; you have to make sure it doesn’t wander outside the lines and into another app’s territory, you have to play nice and pass the CPU to the next app when your turn is over…and you have to trust that other apps are similarly careful and polite.
It didn’t always work well. Old timers recall, none too fondly, the dreaded icons announcing a crash:
This gave us a clear goal: Write an OS kernel and slip it between the Mac’s Motorola 68K processor and the current Mac OS.
Simple, no? No.
We didn’t have time to write such a complex piece of code, so we looked at companies such as Hunter & Ready who would license a kernel. Problem almost solved.
But then we realized that the task of delicately lifting the existing Mac software base, disconnecting and reconnecting blood vessels and nerves — that, too, would take time that was beyond our budget; we were concerned that we’d lose the patient. (Lest one think that we were too timid or clumsy, Jobs’ team of master computer scientists — Avie Tevanian, Bertrand Serlet, Scott Forstall, et al. — were faced with the same dilemma when Apple bought NeXT in 1997. It took them a good four years to create the Unix/NeXTSTEP-based Mac OS X.)
As we got busy with the short and medium term fixes for the Mac product line, Apple engineer Sam Holland came up with another, longer term idea: Let’s develop a quad processor of our own for future Apple personal computers, one that would outperform evolutionary Intel and Motorola chips and give the Mac a (preordained) place at the industry’s top
To simulate the microprocessor, codenamed Aquarius, we bought a Cray supercomputer and used AT&T Microelectronics as our development partner. This wasn’t the AT&T that we now love to loathe as a wireless carrier, it was the technology giant whose Bell research labs gave us Nobel laureates such as Arno Penzias and Steve Chu, and a long list of inventions including the transistor, cellular telephony, the C programming language, the Unix operating system, the original non-blocking telephone switch and many more.
Although the quad processor development work didn’t produce direct results, the Aquarius project stands as an example of Apple’s abiding desire to control the future of its hardware, a yearning that would again manifest itself, successfully this time, when Jobs bought Palo Alto Semiconductor to develop the Ax series of microprocessors that power iPhones and iPads, microprocessors now widely considered the industry’s best in their category. Also noteworthy: The AT&T Microelectronics relationship led to the Newton project, and then on to another company’s hardware… but that’s a later chapter.
The fixes we started working on did get the Mac out of the ditch. In future installments, we’ll see how well, or poorly, I did with the cultural and political components of my new assignment.
JLG@mondaynote.com
