Android Begins
“Ha, is this really their official response?” Bill Gates stared at the document in disbelief, even adjusting his glasses for a better look.
“Your eyes aren’t playing tricks on you,” Steve confirmed, having already reviewed the papers.
Only then did the incredulity on Gates’s face give way to rising anger. The document the two Microsoft executives were examining was the formal notice Remington had prepared based on Yoo Jae-won’s reply. Of course, Remington had not sent the words verbatim. After consulting with the legal expert Ellen, he had polished the language to suit the dignity of ID Technology’s owner and CEO.
“No investment required, and they’ll solve DOS’s memory problems for a single dollar?”
It wasn’t a distortion of Yoo Jae-won’s original statement—only the tone had been made more courteous. The core claim, that they would fix the memory issue Microsoft had never solved for one dollar instead of four, had been preserved as faithfully as possible.
“Hah. The puppy doesn’t even know what it’s barking at,” Gates muttered, taking the one-dollar offer not as a rejection but as outright provocation. Both he and Steve were convinced of Microsoft’s absolute superiority. If even they couldn’t solve DOS’s conventional memory limitations, there was no way a tiny company with barely a dozen employees could manage it.
To be fair, Gates and Steve did know how the problem could be solved: by adopting Unix-style memory management or the protected mode Windows 2.0 had experimented with. But doing so would break compatibility with the millions of existing applications—an unacceptable price.
“They probably aren’t planning to preserve compatibility either.”
Gates assumed ID Technology’s one-dollar challenge would likewise sacrifice compatibility in favor of stronger memory management.
“Of course. That’s exactly how ID Office handles memory.”
Their conclusion was based on ID Office itself. Reverse engineering had revealed that the program contained its own memory-management utility. Much of its superior performance over MS Office came from that utility. Running ID Office required first launching non-standard memory managers like QEMM 386, which caused conflicts with other software. In other words, applying ID Office’s memory methods to DOS would mean abandoning compatibility with existing programs.
“It’s been a long time since we were turned down so cleanly.”
“Haha, they’re a startup. They probably don’t know any better and just lashed out in the heat of the moment.”
“Likely. In any case, we need to push DOS 4.0 forward.”
“With improved memory stability?”
“Naturally. The late-July launch will proceed as scheduled. After that, watching how they respond should be entertaining.”
Neither man explicitly said they would ensure ID Office’s memory utility lost stability in version 4.0. Both were masters at excluding competitors without crossing legal lines, their experience with countless lawsuits having taught them exactly how to phrase things.
“By the way, did we make any recruitment offers to their developers?”
Gates tossed the document aside and changed the subject.
“Yes. Headhunters have been in contact with Mario Rogers, their lead developer, and a few other programmers. I wouldn’t get my hopes up, though.”
“Why not? Is our name not enough? Or is the compensation lacking?”
“Apparently the incentives tied to ID Office are substantial. There’s a launch-day bonus, plus running guarantees based on sales volume.”
Gates shook his head, unable to understand why anyone would hesitate to join Microsoft for such a promise. If Microsoft decided to target a rival, ID Office’s future would be a thorny path. Sales would suffer, and the promised bonuses would likely never materialize. Waiting for them would be like a frog slowly boiling in water.
“Running guarantees? Their entire company—executives and employees alike—seems incapable of facing reality.”
If the opponent was that foolish, Gates saw no reason to worry. Steve nodded in agreement. The two moved on to the next agenda item.
“Windows 3.0 will launch as planned, correct?”
“We have no choice. It’s too late to change now. Besides, they were quite thorough. They’ve filed global patents on that ribbon interface. Even if we ignore the patents and copy the concept, we’d still need to rewrite a significant portion of the code. Meeting the schedule would be difficult.”
The conversation turned to the next-generation graphical operating system. What concerned them most was the interface. Both Gates and Steve had an eye for quality; the ribbon interface used in ID Office was far cleaner and more intuitive than the Program Manager Windows had adopted. Their discussion now centered on how to capture as much of its form as possible without infringing the patents.
“We’ll have to live with it. At least we can apply the design to Windows NT.”
“Of course.”
Windows NT was being designed as a 32-bit enterprise operating system capable of competing with Unix. Currently only a conceptual project worked on by a handful of developers, it was scheduled for full-scale development with a massive team once Windows 3.0 was complete.
The conversation ended there. Even after acknowledging ID Technology’s existence, Microsoft’s response remained largely inertial, its sheer size dictating its pace.
The intercom rang.
“Ahem. Melinda? What is it?”
Gates answered with his usual gentle tone. Though known for his sharp tongue, he had been unfailingly kind to Melinda, his recently appointed secretary.
“What? What do you mean?”
The news Melinda delivered was anything but routine. After a few short exchanges, Gates slammed the receiver down so hard the sound echoed through the office. Steve, who had been watching, was startled by the violence of the reaction. Gates might have a rough personality, but he was not usually destructive. For him to lash out like this, something catastrophic must have occurred.
Gates took a deep breath.
“The computers in Alpha Lab have started throwing unexplained errors simultaneously. Hard drives are failing. They’re attempting recovery now.”
Steve’s worst fears were confirmed.
“How many machines?”
“They’re still counting, but it’s already over ten.”
It was the worst possible report. Alpha Lab was Microsoft’s elite programming team, assembled from its most capable developers. They handled the core code for Windows and DOS, as well as the reverse engineering of ID Office—tasks Microsoft considered most critical.
“We can’t stay here. We need to go see for ourselves.”
Unable to remain seated, both men leapt up and rushed toward Alpha Lab.
Early July, the sunlight growing harsher by the day.
“They really picked the wrong person to mess with.”
Back home after school, Yoo Jae-won ground his teeth as his fingers flew across the keyboard. Anger had sharpened his already formidable concentration. Normally he typed at a brisk 400–500 characters per minute; now, fueled by fury, he added another 100–200. On the screen, complex C source code appeared at more than twice the speed of a 2,400-bps modem.
What he was writing was a program designed to neutralize Microsoft’s plan to use DOS 4.0 against ID Office at the source. He was not, however, attempting to create an even more innovative memory manager or simply adopting DOS’s existing approach for safety. Instead, he was building something that would eliminate the root danger—DOS itself.
“I didn’t expect I’d have to write a bootloader this soon.”
A bootloader—also known as a multi-boot program—was a near-magical utility that allowed multiple operating systems to run on a single machine. Its principle was simple: it claimed the boot sector before DOS could, presenting the user with a choice at startup. This made it possible to install and run two fundamentally different operating systems on the same computer, selecting between them each time the machine was powered on.
In Yoo Jae-won’s original roadmap, the bootloader had been scheduled for a later stage, intended for the Android operating system that would inherit Unix architecture. Although Android already contained numerous safeguards for DOS compatibility, some programs would inevitably fail to run properly. The plan had been to include a bootloader so users could simply choose original DOS at startup if needed.
“If you think ID Office is just an office suite, you’re making a serious mistake.”
Yoo Jae-won intended to embed this bootloader directly into ID Office. Upon reflection, there was no need to wait for Android. ID Office’s desktop and taskbar already functioned like an operating system in their own right. Icons on the desktop could launch applications, the Start menu panel could register and run programs, and the file manager could handle directories and files. It could execute not only ID Technology’s own software but nearly any program written for the DOS environment. When running ID Office or Keyboard Warrior, everything stayed within the ribbon interface; when launching third-party applications, the ribbon would temporarily deactivate and hand full-screen control to the program. Upon exit, the screen would flicker briefly before returning to ID Office’s desktop.
If he added the ability to play wave and MIDI files and made the whole thing capable of booting independently, ID Office would become both an operating system and an office suite.
“Creating a DOS is hardly difficult.”
DOS was an extremely primitive operating system, composed of just three files: IO.SYS for hardware input/output, COMMAND.COM for the user interface, and MSDOS.SYS for file access and memory management. Its simplicity had spawned countless variants—PC-DOS, FreeDOS, and DR-DOS among them. Even in Korea, the Ministry of Science and Technology had once collaborated with universities to develop K-DOS in an attempt to escape Microsoft’s influence.
Building a custom DOS that remained compatible with existing software while running ID Office was no great challenge for Yoo Jae-won. The resulting system would not be perfectly compatible with MS-DOS, which was why most variant DOSes had never achieved meaningful success. Any programs that encountered errors could simply be launched through the bootloader.
“This should make the transition to Android far smoother as well.”
Even if he built it like an ordinary DOS program, it would run. But applications written with the ribbon interface and Glide X library would perform far better. For games that demanded maximum hardware performance, using Glide X would be both easier to develop and deliver superior results compared with rewriting engines from scratch.
Glide X was already gaining traction among game developers. Hardware manufacturers were not far behind. Cirrus Logic, ATI, and S3 were advertising new products slated for August that would carry the “Glide X Full Support” mark. Advanced features such as hardware scrolling, anisotropic filtering, and direct memory access had been released for free; only a fool would refuse to use them. Sound Blaster, the de facto standard for sound cards, had also adopted Glide X’s audio library. Joystick and joypad makers welcomed it with open arms as well. Smaller manufacturers often struggled because DOS drivers were unreliable, causing malfunctions in certain games. Building a universal driver was difficult for small companies, but simply conforming to Glide X’s controller library guaranteed compatibility across titles.
And if even that failed, the bootloader could always launch MS-DOS.
“Wait. In that case, should I give this a name too—call it Android?”
The DOS he was creating in response to Microsoft’s compatibility threat was being born almost by accident. For now its purpose was to run ID Office and, as a bonus, DOS-compatible programs. Yet it was undeniably a standalone, bootable operating system. Still, it felt insufficient to bear the name Android.