On April 8th of 1983, Steve Jobs came by the Lisa offices on Bandley Drive with the former governor of California, Jerry Brown, in tow. I had just arrived at the office, with a full day of work already scheduled, but had to drop everything when Steve asked me to brief Brown on the Lisa and to give him a tour of the offices.

Al Mandel and I spent an hour or so with Jerry in the Lisa "Sneak Room” (a very well designed multimedia, executive briefing facility), giving him demonstrations and encouraging him to try some stuff himself. As I recall we even took apart a Lisa to show him the hardware architecture. He was trying to feign interest but his mind was off somewhere else (as it often was during his tenure as Governor).

After the show-and-tell I then took Brown on a guided tour through the Lisa facilities and we visited engineering, documentation, marketing, etc. In the past I had taken other famous people on walkthroughs and Apple employees would usually never bat an eye. We were used to this (and after all, WE were celebrities as well). However, the reaction to Jerry Brown was unlike anything I'd ever seen. People were acting almost nervous...concerned. While Jerry was talking to an engineer, someone else took me aside, looked me straight in the eye and pointed at Jerry with a look of anguish on his face. As I had no idea what he was getting at I simply replied nonchalantly, "Yeah...that's Jerry Brown". After 10-15 minutes of this Jerry had completely lost interest. I walked him out of the offices and we bid each other goodbye. But I was still very puzzled by the reaction to him.

As soon as I went back in the Bandley Lobby I discovered the source of the bewilderment. The Apple rumor mill was always incredibly active and that morning (unbeknownst to me because I'd been locked up with the former Governor) the word on Bandley Drive had it that Apple was prepared to announce a new CEO that day. And Jerry's arrival at the Lisa offices nearly triggered one of the biggest Wall Street panics since Black Monday. Some people were convinced that Jerry was to be named the new CEO of Apple!

And, looking back, at that time in Apple's history this scenario was actually very plausible...

As it turns out the Apple rumor mill was correct. Later that day John Sculley was announced as the new CEO of Apple. But that's another story for another time.

While Bill Atkinson was developing LisaGraf, the crucial, lightning fast graphics package that was the foundation of both the Lisa and Macintosh user interface (it was renamed QuickDraw for the Mac), he also worked sporadically on a simple bitmap-based drawing program for the Lisa called SketchPad. SketchPad enabled mouse-based drawing with a selection of paintbrushes and patterns, and gave Bill a quick way to test out and show off new features or improvements as they were added to LisaGraf.

In early 1983, soon after the Lisa was announced at the 1983 annual shareholders meeting, Bill switched from working on Lisa system software to writing a killer graphics application for the Macintosh. Steve Jobs thought that he should work on a structured drawing program, something like Mark Cutter's LisaDraw, but Bill thought that structured drawing was too complex, and wanted to create something that was simple, elegant and fun to use.

He began by dusting off his old SketchPad code, and getting it running on the Mac as MacSketch. SketchPad used menus to select patterns and styles to draw with, but Bill replaced them with permanent palettes at the bottom of the screen and added another large, prominent palette on the left, containing a variety of drawing tools. More tools would be added over time, but the basic structure of MacPaint was there from its earliest stages.

The first big advance that Bill worked on was eliminating flicker. As you dragged a shape or image across the screen, it had to be erased from its old position before being drawn in the new one, which caused a distracting flicker as the video sometimes displayed interim, partially rendered states. Bill completely eliminated flicker by composing everything in an offscreen memory buffer, which was transferred to the screen in one fell swoop, so the interim states were never visible.

In fact, despite the Macintosh's limited memory, he used two offscreen buffers, each the size of the document window, with one containing the current pixels of the document, and the other containing the pixels of its previous state, before the most recent operation. This made it very easy to implement undo, by just copying the old buffer to the new. It also enabled fast drawing, because it was very easy to access the original state of the document in the second buffer as an object was being modified.

After getting basic updating working well, Bill began to add more tools to the palette. He added a rectangular selection tool, to allow the user to perform operations on a subset of the document. The selection rectangle was depicted by inverting the pixels beneath it, which worked well enough over solid areas, but was confusing and hard to see when the underlying image was complex, like a digitized picture. Bill knew he had to find a better way to do it.

He was thinking about the selection problem one evening when he went to dinner at an old beer and hamburger joint in Los Gatos, the kind of place where decades worth of initials are carved into the wooden tables. He ordered a beer, looked around the bar, and noticed a Hamm's beer sign. The beer sign featured an impressive animated waterfall, with the water seeming to flow down the waterfall into the lake. Bill figured out the animation was accomplished by a mask layer moving beneath the surface of the sign that varied which portion of the image was visible below.it.

He suddenly realized that an animated border, animated with a technique similar to the waterfall, could solve the graphical selection problem because it would always be easy to spot no matter what the background. He raced back home and implemented the animation by using an alternating sequence of patterns that produced the illusion of continuous motion, just like the beer sign. It looked great, but he didn't know how to refer to it. A couple of days later, he showed it to Rod Perkins on the Lisa applications team, who told Bill the effect resembled "marching ants," which Bill liked, so that's what he called it.

Around April 1983, Bill changed the name of the program from MacSketch to MacPaint. He began to hit his stride, and started to add new features to MacPaint on a daily basis. One of the most important was "Fat Bits", a mode which magnified a small section of the document by a factor of 8, allowing the user to easily manipulate individual pixels. It was implemented by scaling the offscreen buffer as it was transferred to the screen, so all of the other tools and effects kept working in Fat Bits mode.

Another key improvement came when Bill implemented the paint bucket tool, which allowed the user to fill an entire area of the screen with the selected pattern with a single click. It used a "seed fill" algorithm to find all the neighboring pixels of the matching color, which was difficult to implement because of our stringent memory constraints. Bill ended up implementing a few different seed fill algorithms before he settled on one that was both fast and memory efficient enough. The seed fill routine eventually migrated from MacPaint into QuickDraw as part of the 128K ROM.

Rectangular selection was useful, but often you needed to manipulate an arbitrarily shaped area of an image, so Bill came up with another selection tool he called the "lasso", because it contracted around its target. It's hard to control the mouse with pixel-level precision, so the lasso tool allowed you to roughly outline the area of interest, and it would automatically tighten, skipping white pixels to cinch tightly around the target. The lasso sort of fell out of doing the paint bucket, since the seed fill routine was exactly what was needed to cinch the lasso properly.

The lasso allowed an arbitrary area of pixels to be selected, so it required a third offscreen buffer the size of the document window, that contained a mask to indicate which pixels were part of the selection. The mask buffer allowed fast graphical operations on large, irregular areas, such as dragging an image around the screen, or making the lasso selection shimmer with the marching ants effect. The first two offscreen buffers were allocated statically, but the mask buffer needed to be deallocated when it wasn't being used, so its memory could be used for other purposes.

MacPaint was good at drawing text, allowing the user to specify characters at any position, with any font, size or style. But once the text was instantiated, it just became pixels like everything else; you couldn't go back and edit it as text. In June 1983, Bill thought he could do something about that.

Bill decided to try to turn pixels back into characters when you selected them with the text tool. He wrote a lot of elaborate code, probably as much as for any other MacPaint feature. First, he wrote assembly language routines to isolate the bounding box of each character in the selected range. Then he computed a checksum of the pixels within each bounding box, and compared them to a pre-computed table that was made for each known font, only having to perform the full, detailed comparison if the checksum matched.

Bill got his character recognition routines working well, and it seemed like magic, if you were used to the earlier MacPaint, to be able to recover and edit previously placed text. It wasn't perfect, because it would fail to recognize a character if a single dot was out of place, but it was still very useful. Everyone loved the feature, and congratulated Bill for pulling off another miracle.

I was surprised a few days later when Bill told me that he decided to remove the character recognition feature from MacPaint. He was afraid that if he left it in, people would actually use it a lot, and MacPaint would be regarded as an inadequate word processor instead of a great drawing program. It was probably the right decision, although I didn't think so at the time. I was amazed that he was able to detach himself from all the effort that he put into creating the discarded feature; I know that I probably wouldn't have been able to do the same.

MacPaint was essentially finished by October 1983, long before our other key applications. The last part of finishing MacPaint involved dealing with out of memory problems, since it was really pushing the limits of the 128K Macintosh by using the three, large offscreen buffers. At the worst case, there was only about 100 bytes free in MacPaint's heap. Most of the bugs we encountered when running MacPaint turned out to be bugs in the underlying system, which were exposed by running so close to the edge of available memory.

It's interesting to note that MacPaint was a rather small program by today's standards, but I guess that it had to be to run in the Mac's one eighth of a megabyte of memory. The finished MacPaint consisted of 5,804 lines of Pascal code, augmented by another 2,738 lines of assembly language, which compiled into less than .05 megabytes of executable code.

Landing in the Macintosh group as a bitmap graphic designer was a lucky break for me, and one interesting part of the job was designing screen fonts. It was especially enjoyable because the Macintosh was able to display proportional typefaces, leaving behind the tyranny of monospace alphabets with their narrow m's and wide i's.

The first Macintosh font was designed to be a bold system font with no jagged diagonals, and was originally called "Elefont". There were going to be lots of fonts, so we were looking for a set of attractive, related names. Andy Hertzfeld and I had met in high school in suburban Philadelphia, so we started naming the other fonts after stops on the Paoli Local commuter train: Overbrook, Merion, Ardmore, and Rosemont. (Ransom was the only one that broke that convention; it was a font of mismatched letters intended to evoke messages from kidnapers made from cut-out letters ).

One day Steve Jobs stopped by the software group, as he often did at the end of the day. He frowned as he looked at the font names on a menu. "What are those names?", he asked, and we explained about the Paoli Local.

"Well", he said, "cities are OK, but not little cities that nobody's ever heard of. They ought to be WORLD CLASS cities!"

So that is how Chicago (Elefont), New York, Geneva, London, San Francisco (Ransom), Toronto, and Venice (Bill Atkinson's script font) got their names.

We always thought of the Macintosh as a fast computer, since its 68000 microprocessor was effectively 10 times faster than an Apple II, but our Achilles heel was the floppy disk. We had limited RAM, so it was often necessary to load data from the floppy, but there we were no faster than an Apple II. Once we had real applications going, it was clear the floppy disk was going to be a significant bottleneck.

One of the things that bothered Steve Jobs the most was the time that it took to boot when the Mac was first powered on. It could take a couple of minutes, or even more, to test memory, initialize the operating system, and load the Finder. One afternoon, Steve came up with an original way to motivate us to make it faster.

Larry Kenyon was the engineer working on the disk driver and file system. Steve came into his cubicle and started to exhort him. "The Macintosh boots too slowly. You've got to make it faster!"

Larry started to explain about some of the places where he thought that he could improve things, but Steve wasn't interested. He continued, "You know, I've been thinking about it. How many people are going to be using the Macintosh? A million? No, more than that. In a few years, I bet five million people will be booting up their Macintoshes at least once a day."

"Well, let's say you can shave 10 seconds off of the boot time. Multiply that by five million users and thats 50 million seconds, every single day. Over a year, that's probably dozens of lifetimes. So if you make it boot ten seconds faster, you've saved a dozen lives. That's really worth it, don't you think?"

We were pretty motivated to make the software go as fast as we could anyway, so I'm not sure if this pitch had much effect, but we thought it was pretty humorous, and we did manage to shave more than ten seconds off the boot time over the next couple of months.

In 1980, a company called Franklin Computer produced a clone of the Apple II called the Franklin Ace, designed to run the same software. They copied almost every detail of the Apple II, including all of its ROM based software and all the documentation, and sold it at a lower price than Apple. We even found a place in the manual where they forgot to change "Apple" to "Ace". Apple was infuriated, and sued Franklin. They eventually won, and forced Franklin to withdraw the Ace from the market.

Even though Apple won the case, it was pretty scary for a while, and it wasn't clear until the end that the judge would rule in Apple's favor - Franklin argued that they had a right to copy the Apple II ROMs, since it was just a "functional mechanism" necessary for software compatibility. We anticipated that someone might try a similar trick with the Macintosh someday. If they were clever enough (which Franklin wasn't), they could disguise the code (say by systematically permuting some registers) so it wouldn't look that similar at the binary level. We thought that we better take some precautions.

Steve decided that if a company copied the Mac ROM into their computer, he would like to be able to do a demo during the trial, where he could type a few keystokes into an unmodified infringing machine, and have a large "Stolen From Apple" icon appear on its screen. The routines and data to accomplish that would have to be incorporated into our ROM in a stealthy fashion, so the cloners wouldn't know how to find or remove it.

It was tricky enough to be a fun project. Susan designed a nice "Stolen from Apple" icon, featuring prison bars. Steve Capps had recently come up with a simple scheme for compressing ROM-based icons to save space, so we compressed the icon using his technique, which not only reduced the overhead but also made it much harder to detect the icon. Finally, we wrote a tiny routine to decompress the icon, scale it up and display it on the screen. We hid it in the middle of some data tables, so it would be hard to spot when disassembling the ROM.

All you had to do to invoke it is enter the debugger and type a 6 digit hexadecimal address followed by a "G", which meant execute the routine at that address. We demoed it for Steve and he liked it. We were kind of hoping someone would copy the ROM just so we could show off our foresight.

As far as I know, no one ever did copy the ROM in a commercial project, so it wasn't really necessary, but it did create some intrigue for a while. We let it slip that there was a "stolen from Apple" icon hidden in there somewhere, partially to deter people from copying the ROM. At least one hacker became moderately obsessed with trying to find it.

Steve Jasik was the author of the MacNosy disassembler/debugger, which could be used to create pseudo-source for the ROM. He found out about the "stolen from Apple" icon pretty early on, and became determined to isolate it. He lived in Palo Alto, so I would occasionally bump into him, and he would ask me for hints or tell me his latest theory about how it was concealed, which was invariably wrong.

This went on for two or three years, before he finally cracked it: I ran into him and he had it nailed, telling me about the compressed icon and the address of the display routine. I congratulated him, but was never sure if he figured it out himself or if someone with access to the source code told him.