Macworld • January 1988


Tales of the MultiFinder

by Steven Levy


Behind the new system software is a controversy that cuts to Apple's core


MultiFinder made its debut at last summer’s Macworld Expo, an event that will ultimately be remembered as the high-water mark of the Macintosh. The crowds were reminiscent of the 1982 Comdex, when people suddenly realized that microcomputers were not only real, but growing explosively, and those who got in then would reap big bucks. In 1987 the Bayside Convention Center was teeming with people in business suits who smelled money in the Mac.


The most talked-about product was HyperCard, which I’ll discuss in another column. The most important product was MultiFinder. This program, a sort of super-Finder, provides the illusion (and, in many cases, the advantages) of running more than one program at a time; and it embodies the past, present, and future of the Macintosh. It has enabled Apple to smugly proclaim superiority over the IBM world, which can only promise a similar product sometime in the future. By that time, Apple executives hinted, an enhanced version of MultiFinder would preserve Macintosh’s technological head start.


But the MultiFinder story, like the Apple Computer story itself, is bittersweet. Its resonances, its strengths and shortcomings, are, almost as if genetically determined, inextricably linked to the controversial Apple character. One could easily detect this heavy baggage when the product was introduced by Apple’s technology czar, Jean Louis Gassée. The French-born executive felt compelled to insist that none of the programming code used in the MultiFinder was held in common with code written by a former Apple engineer for a similar program that had somehow gained notoriety. It was an odd remark, a seemingly gratuitous insult. But as we shall see, it cut to the truth of the matter.




Split Apple

That unnamed programmer was, of course, Andy Hertzfeld, a name familiar to most Macworld readers as one of the prime software wizards of the original Macintosh. Hertzfeld, a resolutely upbeat fellow now in his early thirties, sees his years working on the Macintosh as the major event in his life thus far, and he takes inordinate pleasure in seeing people make the most of the computer he helped to create. Andy Hertzfeld doesn’t work at Apple anymore, but his Macintosh love affair has lingered; in the past few years he has devoted his talents to writing products that will enhance the Mac and help it in the marketplace.


While his goals would seem to match Apple Computer’s, the company’s relations with Hertzfeld have been ambivalent.


Some personal clashes might have contributed to this, but the heart of the matter is that Andy represents the Old Apple of the hackers—flaky, scruffy, passionate, and irreverent. The New Apple sees itself as energetically committed, but also professional, reliable, well-dressed, and in tune with the needs of conglomerates and robber barons. The New Apple likes the luster of its rambunctious youth, but wants the world to know that it has grown up.


The other problem with Andy is that he is associated with the reign of Steve Jobs. The former chairman and cofounder is still Apple’s unwelcome Ghost In The Machine, and no amount of success seems to be able to assuage the corporate Oedipal trauma.




A Switch in Time


What happened with MultiFinder illustrates the paradox. The program was really born from the limitations of the Finder, the file-handling program in the form of a desktop that was released with the original Mac. On early Macintoshes, closing down one program and opening another was a process that one did not undertake without setting aside some spare time. Competing computers did not have this problem, and it was something that, in certain eyes, damned the Macintosh to the status of a toy or curiosity. But no one really addressed that problem until Andy Hertzfeld, on a suggestion from writer John Markoff, hacked up a program called Switcher.


Switcher was originally a two-week project, done in time for the first Hacker’s Conference in November 1984. Apple immediately saw the program’s value, and Steve Jobs himself bought the program from Andy for $100,000 only minutes after viewing it. For Apple it was money well spent, as Switcher, in its various versions, became a terrifically useful, if not particularly robust, program.


By the next summer the Mac market was floundering. As Andy Hertzfeld roamed the sparsely attended aisles of the August 1985 Macworld Expo, he wondered what else he could do to help the Mac. He decided to rewrite the Finder, incorporating the features of Switcher, so the computer would be even easier to use and could more easily run several programs at once. So he set about writing a program to be called Servant. By the New Year he had it running. Indeed, it had multiple windows: you could open a program and then, by downscaling a window; see the desktop. From there, you could open another program, and another. Andy began to see how you could even do some multitasking—actually using one program (say a word processor) while a second program was performing a task (such as a communications program downloading a file).


Hertzfeld took Servant to Apple and showed it to the systems software group. There were gasps and buggy eyes—the previous day the group had set as its top 1986 priority to do a similar program. In no time Andy was negotiating to sell his work to Apple. The price this time was to be around $200,000. But Andy Hertzfeld was troubled. He had no confidence that the new Apple would preserve all the neat tricks and user-pleasing ideas of Servant in its own finished product. The strength of Servant was in its more powerful interface—new ways to show icons, move them around, and arrange them on screen—but Apple wanted to emphasize multitasking in its eventual product. Hertzfeld couldn’t sleep at night, torn between preserving the artistic integrity of the product and doing “what would be best for Mac.”


He thought he could do a better job of writing the program himself, insulated from the Apple bureaucracy. But he knew that no matter how good his program was, it didn’t have a chance at effective multitasking unless Apple itself marketed the program as basic system software. Only then, with the mother company supporting the product, would there be incentive for those who write applications software to conform to the inevitable restrictions of this new Finder so that the applications would “behave” together.


Finally Hertzfeld arrived at a solution. He approached Apple and offered them the program for half the $200,000—if they would take nonexclusive rights. That way, he would work on his product, and they would work on theirs. Apple accepted, and MultiFinder was underway.


Andy, meanwhile, continued to work on Servant, which he intended to distribute free to anyone who cared to have it.


The dream would be preserved.




Every Twitch Way


That’s Andy’s half of the story. Now for Apple’s.


By 1986 it was clear to everyone at Apple that the Mac needed a new Finder that could handle more than one program at once. That capability was expected on the new Mac II, and it would be a boon to Apple if the program would work on older Macs as well. Why Apple was not hard at work on this before 1986 is a mystery, and maybe a scandal, but in any case it became a priority. The person charged with creating this software was Erich Ringewald, then 26, an Apple engineer since March 1985. It was Ringewald who had been assigned responsibility for Switcher support in the latter versions of that program, though his relationship with Andy Hertzfeld had never been warm. Ringewald respected Hertzfeld’s talents, but thought he could do the new Finder, code-named Twitcher (clever, huh?), very well, thank you, without Andy’s help.


According to Ringewald, Apple’s purchase of Servant was irrelevant to those working on Twitcher. “We had the same goal, but Andy was beginning from the old Finder and interested in user extensions to it,” he says. “I was interested in a stable mechanism [to run more than one program at once].” A big difference between the two programs is how they handle windows when two or more programs run at once. Just how they differ is too technical get into here, and I suspect it’s one of those religious battles that programmers have.


As far as Ringewald was concerned, Apple’s purchase was unnecessary, and he insists that he was never even curious to see Servant.


Ringewald was later joined by a new Apple programmer, Phil Goldman, and by the end of 1986 Apple was ready to send a prerelease Twitcher to selected developers, under the name of Juggler. The biggest problem, as expected, was accommodating applications that “misbehaved” in certain ways: some programs try to hog the whole machine. Ringewald thinks he did a pretty good job in dealing with those malefactors, or at least setting up rules for their rehabilitation in future versions. By the time of the Expo, when the program was announced, MultiFinder (alas, the nifty Juggler name was already taken) was able to run an impressive demo without crashing. Judging from the educated oohs and ahhs from that demo, Apple did the right thing with MultiFinder. By concentrating on the juggling power rather than Servant-like user-interface features, it gave its customers what they needed immediately.


How soon they would have done this without Andy Hertzfeld’s original contribution is a matter of opinion. My opinion: not soon enough.



Memory Lane

So where are we now? Well, MultiFinder is a big step for Apple, which sees it mainly as that—a step. Charles Oppenheimer, Apple’s mouthpiece for MultiFinder, explains that system software is “not a point in time, it’s a continuum.” At this point in the continuum, we have a Finder heir that can kind of do multitasking. That near-hit does not satisfy everybody. In these grown-up days of Macintosh, there is no shortage of techno-weenies and MIS guys who huff and puff and say, “Can this do true multitasking?” And to those, Charlie Oppenheimer says, “Not exactly.” But he says MultiFinder will increase the utility of the Macintosh because it is now absurdly easy to have several applications open at the same time. This is especially helpful for those running a program that works best if it is “on” all the time. And what multitasking MultiFinder does do—background printing for laser printers and communications while other programs run—is the kind of thing most people will use. Best of all, MultiFinder does this on Mac Pluses and SEs as well as the Mac II, something we veterans will appreciate.


Later on will come a version in which programs not only run at the same time, but also run with each other. How much later on is anyone’s guess.


It’s a brave new world, but there is an entry fee. MultiFinder takes up a lot of memory on its own—so much that with “only” a megabyte of memory, the hopper fills up after only one decent-size application, negating the purpose of the program. And HyperCard, presumably the program that will take most advantage of MultiFinder, won’t run at all with MultiFinder in a 1MB machine. Since at this writing no Apple computer comes standard with more than 1MB of memory, we are now in a situation analogous to the days of the 128K Mac, the last time Apple shipped a computer with insufficient power to run its standard software. Now 2MB is the obvious minimum, probably 3MB if you want HyperCard on all the time.


People will eventually get the extra memory, and the New Apple will find MultiFinder a valuable weapon in its battle for real-world acceptance for the Macintosh. The present MultiFinder is a big win, and its future looks even more promising.


For Andy Hertzfeld, though, the experience has been mixed. He’s spent months working on his version of the Finder, Servant, perhaps imprudently. “I couldn’t help it,” he says, “I wasn’t working on it, it was working on me.” Servant is now available on CompuServe and other bulletin boards, and from user groups. You should check it out, see the way it handles icons, and marvel at its file-copying tricks. Have fun with it. Andy hopes that its innovations will find their way into Apple’s system software. “In a way, I’m forcing Apple to improve things. If I come up with a better way of doing something and people see it, Apple will copy it.” Or so he hopes.


Meanwhile, MultiFinder is here, relegating Servant to the status of a curiosity. MultiFinder is not so much fun as it is productive. Just like the New Apple. The Macintosh, as evidenced by the gold-diggers at the Expo, has finally made it. Nodding toward the masses of well-heeled professionals making the rounds of over 200 exhibitors at the show, Andy Hertzfeld admitted that it was the end of an era. “The Mac roller coaster has come to a halt,” he said, indicating that the days of ups and downs have been replaced by stable acceptance. “It’s time for me to get off and find a new one.”


For Apple it’s time to consolidate and improve, and to make further inroads into the serious world of computing. The MultiFinder and its descendants will do that. But I hope the company won’t forget the early creativity that gave Mac its technological head start-nor the people who, with the foolishness of lovers, gave years of their lives, as well as their hearts, to a computer and a dream. Apple will survive without Andy Hertzfeld and indeed without Steve Jobs. But there is no need for Apple to purge its zealous past in order to preserve its productive future.