Advent of Computing - Episode 150 - Starting Windows Up

Episode Date: January 20, 2025

In the modern day Windows is a power house, but that wasn't always the case. In this episode we are looking at the fraught development of Windows 1.0. During development it was called vaporware, it was panned in the press, roasted at at least one trade show, and even called... "eclectic". Through it all a vision in lime green would take form.

It's been said that programming is more an art than a science. This is definitely a sentiment I can get behind, and not just because that would make me a professional artist. Programming is so complex, and computers are so flexible, that there's not always a correct solution to a problem. There can be many different solutions, each with their own drawbacks and benefits. Maybe you can solve a problem with fewer lines of code than your buddy, but their solution might be just a little bit faster.

Still, another's code could be a little slower, but much easier to read and understand. It's usually the case that none of these solutions are perfect. That in itself adds another layer for the artist to work with. Choosing just the right approach, such that it fits well with the rest of your code, is an art all to itself. How is this art form possible, you may ask? Computers on their own aren't very artistic. They are, after all, just massively complex and powerful calculators.

They're full of optimal solutions and correct choices. There's one correct way to flip a bit. There's one best numeric adding circuit. Programming is where our fleshy minds run into this rigid silicon. The programmer must warp and twist their brain to find ways to express ideas to one of these machines. The art here, I think, lies in that twist. There's another layer we can add to this, one that the solo programmer will never know. You see, there's a difference between just programming and software development. Once you start working on a large program and once you start collaborating with other programmers, that's when things get

truly complex. This is a lesson that I didn't really learn until I got into the biz. There's a whole other type of art here. There are many different ways to design a program, combine its different pieces, and even go about writing it. This is when we shift from half flesh half machine to mainly flesh. You have to plan around other people, find ways to get programmers working in sync, and even compromise in some places. And I know, the horror. In some cases, this kind of work can lead to truly high art. That's how you get a Unix or a Doom.

But just like in bad painting, poorly placed strokes of the brush can cause the whole work to crumble. Welcome back to Advent of Computing. I'm your host Sean Haas and this is episode 150, Starting Windows Up. Now, are you familiar, dear listener, with how software is written? I don't mean the nuts and bolts technical side here, I'm talking more about the day to day and the management tasks. When it comes to programming, the actual task of programming is only one part of the job. There's also the larger scale plans and corporate machinations going on, which are then transformed by layer upon layer of managers and bureaucracy into the day-by-day plans and strategies that we programmers live by.

Most of this grind is completely hidden to the outside world. That's especially true in real time. A programmer comes in on a Monday, gets assigned a new feature, starts work, and then by Friday the feature's canned. Us end users will never know about that cycle. That is, unless either something catastrophic happens or we read a biography or an interview decades later. Usually all the outside world sees is what the software house wants us to see. You see demos, press releases, ads, and then the final program

one day. For many programs, this follows a pretty dull outline. A new product is announced. Photos or videos circulate. Maybe there's a beta program, might be closed or open, and then the full program is released. The early press is a little simplified, maybe some illusions of grandeur. The beta is kind of rough, but the final release pulls it all together and adds a little bit of a polish. Sometimes, however, that pattern is broken. Something weird occurs in the office.

There's one story hiding in plain sight that I think exemplifies this. Windows 1.0. In 1983, Microsoft demonstrated their first graphical environment. It seemed nearly complete, fully usable in fact. In 1985, Windows 1 shipped and it bore only a passing resemblance to earlier demos. So what happened? In this episode, we'll be looking at the development of Windows 1.0 and trying to work out why it shifted so radically. We're well into the 1980s this time, which means we're in the realm of home computing.

Standards have already been established, the PC is ascendant, and there's inspiration to draw on. This means that as we chronicle Windows, we'll be entering a spider's web of connections and perhaps stolen intellectual property. This is, surprisingly, one of those periods that's full of folklore. And I know that may sound strange because this is a lot more recent and better documented than earlier stories we talk about on the show. But there's just some wacky and fascinating events going on. For instance, in this exact time period, due to Windows 1 itself, Bill Gates and Steve Jobs would get into a full-on screaming argument

at Apple HQ. Jobs accused Gates of stealing from Apple. Gates is said to have retorted, quote, we both had this rich neighbor named Xerox, and I broke into his house to steal the TV set and found out you had already stolen it. How's that for spicy? Now before we get into the episode, I have a couple of announcements. The first is that I'm currently running a poll for the next bonus episode on Patreon. I'm gonna leave that up for I think one more week and then close it and decide the winner. If you want your voice to be heard, then go over to the Patreon. You can get to it by going to adminofcomputing.com and clicking on support me on Patreon. Sign up for an account, donate as

little as $1 a month, you get all my bonus episodes, I think it's about half a dozen right now, and you get to vote on the current poll for the next episode. Next, I'm going to be speaking at the Intelligent Speech Conference on February 8th this year. Intelligent Speech is an event I've done before and I had a great time at it. It's a lot of fun. It brings together the best history podcasters in the biz and they all talk about some shared themed topic. This year the theme is deception. I went back and forth a little bit and I decided I'm going to be talking about viruses. I

have the presentation already. Getting pretty excited about it. That's happening on February 8th online. You can get tickets at IntelligentSpeechOnline.com and if you use my promo code AOC you get a discount. Now thirdly, February's busy for me this year, I'm going back to VCF SoCal. If you haven't done Vintage Computer Fest before, once again, it's a fabulous event. I highly recommend it. I spoke, sat on a panel, ran a panel, and just hung out at VCF SoCal last year and it was a wonderful time, so I'm going back. That's happening the weekend of February 15th this year in Orange, California. If you want to go to that, get tickets and get information about accommodations, you can go to VCFsoCal.com.

So if you're in the area or you want to travel, come on by. I'd love to meet some of you and I'm gonna be giving a talk, I think moderating a panel, and I might be doing a live interview too. That's all still a little bit up in the air but whatever the case it's gonna be a really fun weekend. All right, all of those announcements done, let's start the show. Our story starts as a surprising number of stories do with VisitCalc. Released in 1979, VisitCalc was the first computer spreadsheet application. It's also been called the first killer app. You see, VisiCalc was originally only intended for the Apple II.

That was the only computer it would run on. The program was such a big deal that it actually drove sales of that specific machine. This was a critical step in the development of the personal computer. Visicalc was one of the first truly professional programs running on a micro that was targeted at professionals, not just programmers or hobbyists. Previously, these smaller machines had been somewhat ancillary to large-scale computers. If an office had microcomputers, they were there in addition to a mini-computer or a mainframe. They sat at desks and they worked as terminals or they offered a little extra oomph on the desktop. VisiCalc could turn an Apple II into a business machine. And that, as far as an off-the-shelf

experience was pretty new. VisiCalc is a particularly powerful example because it was the first computer spreadsheet full stop. There wasn't an equivalent on larger machines. But while popular, it wasn't the only business app running on micros. There were things like text editors, wonderful tax preparation programs, small database systems and programming tools. VisiCalc, however, really pushed things. This was a field that Microsoft was watching intently. In 1979, Microsoft was in a somewhat mid-tier spot. Their main product, and their only real successful product at that point, was BASIC. Since 75, they had been selling different versions of BASIC to anyone who would pay.

Just about every home micro ran some kind of Microsoft BASIC, even if it was labeled a little bit differently. It was a reasonable business model, but it didn't have too much room for expansion. As early as 1979, Bill Gates had been trying to give Microsoft's business model an upgrade. This started by licensing Unix, which, well, it was a good idea, it didn't exactly turn out well.

Their licensed distribution of Unix, called Xenix, would end up running on quite a few platforms. That included the eventual Apple Lisa, of all things. But it never really took off despite how much Gates believed that Unix would be the future, even all the way into the 1990s. There was another program entered into their portfolio in this period. That was DOS, the Disk Operating System. We've covered the story before. The truth of the matter is actually a little hazy and the story reads almost like a comedy of errors. In 1980, IBM comes to Microsoft to buy a version of BASIC for the upcoming IBM PC. Gates gives them a counter offer, BASIC, AND a disk operating system.

It's my informed speculation that Bill Gates intended to sell them Xenix. Remember, he believed the future was Unix and he had a license in hand from Bell Labs themselves. But the common story is that Gates fibbed a little and said Microsoft could totally deliver an operating system in time for the PC's launch. IBM signed the agreement. Once again, the common story goes that Gates knew full well Microsoft had no in-house operating system, so he scrambled to find someone, anyone, who will sell him some code. Gates ends up buying this thing called QDOS from Seattle

Computer Products, which gets metamorphosed into MS-DOS, later sold by IBM as PC-DOS. And as far as my speculation goes, I've asked Bill Gates directly. At press time, he has yet to respond to my email, so Bill, if you're out there, hit me up. I got some questions for you. Anyway, by 1981, Microsoft is starting to diversify. They had a few operating systems in their portfolio, even a very successful one. And they also had a new platform, the IBM PC. As Gates explained in an interview with the Smithsonian Museum of American History, the new platform made new software possible.

The PC was more powerful than the 8-bit machines that Microsoft's code had been living on. What did Gates want to use all this new horsepower for? Why, simple, it's what anyone would want. Business software. Now I know, perhaps not the most exciting code out there, but it's the kind of stuff that makes money. What's more, the market for business software was just starting to open up. Look no further than VisiCalc itself, the killer app.

That program was making a killing. VisiCalc was so successful that Visicorp, its creators, also started branching out. They had started making other applications, porting VisiCalc to other platforms, and generally doing everything that Bill Gates wished he could do. Then we get to 1982. That year, VisiCorp announced their next big blockbuster release, VisiOn, spelled like vision but I assure you, it's supposed to be said Vizion. There's a little space. This was first unveiled at that year's Comdex trade show.

Vizion was a fully graphical operating system for the IBM PC. It had windows, it had a mouse, something like a desktop, and icons, best of all, it could run VisiCalc. It could run a whole suite of office software and it could multitask. Gates saw the demo, went back to Microsoft, and started up a new project. Microsoft would make their own Vizion. At least, that's the traditional story. This is one of the issues that I have with well-covered history. If you read anything about Windows 1, you'll hear the origin story that I just told you, that Vizion inspired Gates to start Windows. But I can't find a primary source for that tale. Which is frankly strange. It may be buried away

in an interview somewhere, or from a memo, or something like that, but I don't have the exact precise source for this story. I think it's true, that's the assumption I'm going to carry forward with. I just want to point out that the link here is a little rough. Anyway, however it was inspired, Microsoft starts pursuing a graphical something in 1982. I've seen some articles that say Microsoft had vague ideas about graphics as early as 81. Specifically, Microsoft was working on something called the Graphics Device Interface in 81, which was still in an embryonic state and not really being pursued as a massive, big-scale project.

Signs really point to 82 as the earnest start for Windows. This all goes hand-in-hand with the switch into the business world. Before Windows could walk, another program called MultiPlan had to crawl. This is where we enter into the dark depths of profitable business software. MultiPlan was Microsoft's first attempt at a spreadsheet program. It precedes the much beloved Microsoft Excel,

and it followed the VisiCalc model to a T. That is, a simple but effective tool that ran on as many platforms as possible. That last part would end up being a huge downside. MultiPlan was released in 1982, but it had been in the works for a while. This is in a period just before the PC rose to its full, terrible power.

Most users are still on 8-bit machines. So the decision was made that MultiPlan should support those smaller computers. According to Gates, quote, "'MultiPlan' targeting the 8-bit machines instead of just relying on the next generation to come, the PC generation, that was a huge error. When we talk about are we aiming too low in terms of system requirements, we often

think is this another case like multi-plan because it was a great product but it was the basic strategy that was wrong." End quote. Microsoft wasn't pushing things very hard. By targeting 8-bit systems as the lowest common denominator, their software wasn't as powerful as it could have been. That said, multiplayer gave us two important things. It started a larger family of programs called multi-tools and it set Microsoft up with a certain look and feel, let's call it. Multiplane uses a very simplistic,

very plain interface. It has a big space up top and it has a command palette at the bottom, kind of like a menu strip. At all times, there's a chunk of the screen that has a list of what you can do. Hitting tab lets you cycle through and run those commands or you can hit hotkeys to activate them. Like I said, it's really close to the menu bar in Windows. This interface, while not groundbreaking, is workable. More than that, it's actually really easy to adapt to be used by a mouse. Multiplan isn't exactly a killer app, but it does see some use.

Crucially, it caught the attention of Apple. Microsoft and Apple already had a lot of business dealings going back as far as 1977. The Apple II, after a bit of a stumble, would run a version of Microsoft Basic. Gates and Jobs were already well acquainted in that capacity. MultiPlan would serve as another avenue

to further their business relationship. This part of the story comes to us from Andy Hertzfeld, writing on folklore.org. At the time Multiplan came out, Apple was working on their own GUI project. The Apple Lisa and the Macintosh projects were both in flight and both competing with each other. This chapter of history is its own story

that we've talked about before. The long and the short of it is that the Lisa and the Mac took different approaches to the graphical PC. The Lisa would ship with its own suite of in-house office programs. The Mac would not. So with the Mac, there was a need for third-party software. Jobs would turn to Microsoft for at least some of that.

Hertzfeld explains that Microsoft actually got access to Macintoshes well before the computer was announced. This was done specifically so Microsoft could prepare some Office software ahead of the machine's release. That included a new graphical version of MultiPlan. But there is a catch from Hertzfeld. Quote, But there is a catch from Hertzfeld, quote, As a condition of getting an early start at Macintosh development, Steve made Microsoft promise not to ship any software that used a mouse until at least one year after the first shipment of the Macintosh. End quote.

The launch date of the Mac was, in reality, January 24th, 1984. Keep that in mind as we move forward. Here's the timeline we have so far. In 1982, Microsoft starts working on some graphical desktop. Sometime either late 82, early 83, Microsoft and Apple enter into an arrangement, which leads us to multi-tools. As multi-plan comes out, Microsoft is already working on the next entry in its Office series,

that being Multi-Tools Word. It's better known today as Microsoft Word. The plan for multi-tools was that all programs in the family would use a similar interface and be somewhat interoperable. Vizicorp was trying to do a very similar thing in this period. But from the outset, multi-tools Word, which I'm just gonna start calling Word, would be a new breed of program. It may have looked like the old multi-plan but the internals were a huge leap forward. From Gates, quote, Microsoft Word, although it was the second application that we came out with after multi-plan, was really a milestone

because with this one we decided we'd really do something forward, very forward-looking. We'd hired Charles Simoni from Xerox PARC. We knew that graphics interface was where it was at. We knew that laser printers were going to be very big, so we designed something whose underlying structure was ready for the graphical world. In fact, we made it so that it could show italics and bolds! It could work with the mouse." End quote. Now, I had known that both Apple and Microsoft had pillaged Xerox for talent. The Apple story is classic. The business plan for the first Mac was even printed on Xerox PARC letterhead inside Xerox PARC. That's how deep the connection went. I had assumed that Microsoft did something

similar for Windows. That would make perfect sense, right? Gates starts the Windows project, poaches some Xerox engineers, and gets a nice jump start. But no, that's not exactly the case. He brings on Xerox heads for Word, a text editor. That's just neat. But this also sets up a conflict. Word used a mouse. Microsoft had agreed to not ship anything that used a mouse until a year after the Mac launched. That agreement, turns out, was very poorly written. In May 1983, Word is announced. It's graphical, draws out everything pixel by pixel.

It has an interface almost identical to MultiPlan, and crucially, it comes with a mouse. The first Microsoft mouse, sometimes called the Green-eyed mouse because of its two green buttons, that deserves its own story. I bring it up here because it starts a chain of events. This mouse ships in support of the first version of Word. It hits press in May, probably hitting shelves very soon after. This is, I think, most likely what triggered the whole Jobs-Gates fight that I quoted in the intro. I say probably because the source for that fight, Hertzfeld, gives a confusing story. He claims that Apple first learned about Microsoft breaking the whole mouse deal in November of 83, when Windows was first announced. But that doesn't make a lot of sense. There

are articles in Infoworld in May that show big photos of the mouse announcing that Microsoft is releasing software that uses a mouse and selling a mouse. Seems kind of hard to miss. I assume Hertzfeld's just mixing up the timeline a little bit. I think it may have been that in May, Job learns about the mouse, he calls in Gates, a fight ensues, and it soon realized that Microsoft is actually in the right. It turns out that the terms of the contract used an incorrect and earlier ship date for the Macintosh, one that had been missed as the project got delayed. So as the contract stands, Microsoft's fine. They'd actually keep up their relationship with Apple.

With Word now out, what exactly did mouse support look like? Well, it was basic. I mentioned that Word looked like MultiPlan. It had the same layout, a screen for editing up top, and a smaller action palette at the bottom. In MultiPlan, you had to tab around or use combo keys. In Word, you can actually point and click on actions in the palette. It turns out that the interface was very easy to convert for a mouse.

You can also use the mouse to select text and move around the text cursor. One additional feature worth bringing up in Word is windowing, at least, kind of windowing. You could open multiple text files at once, displaying them next to each other. Those windows were tiled, meaning that you had to divide up the screen

without any overlaps. The windowing is very basic, but we're seeing a very specific approach. We're seeing some kind of shared DNA with what will show up in Windows. But for that, we need to jump forward a few months. In the fall of 83, the world meets Windows for the very first time. You may have noticed that this episode is deeply concerned with the timeline of events. I think in this case, we kind of have to be. We don't get much information from the inside this time. What we do have are demos, articles, and releases,

plus a few interviews with programmers much later on. We can use that to build up this picture of what the public was allowed to see and try and combine that to read some tea leaves and see what was going on during the development stage. While the whole mouse debacle was going on, while Word was just starting to ship, and even while the Macintosh's release date was slipping by, Microsoft was working on something

wild. Internally, this project, which would become Windows 1.0, was called Interface Manager. It was planned to run on an unmodified IBM PC using a mouse over a standard serial port. It would be capable of multitasking and windowing. It would run from a floppy disk and it would be cheap, at least relatively speaking. Let me put that into perspective. In this period, home computers were just on the cusp of becoming graphical machines. There were two ways to push them over the edge. Clever software for existing

hardware or next generation hardware to support new software. Apple would go the hardware route with the Lisa and the Macintosh. The Lisa was technically announced all the way back in 81, but real details wouldn't come out until early 83. The hardware route was expensive for a number of reasons. There was the upfront cost of the machine to consider. For the Lisa, we're talking $10,000. That is a wild amount for a computer even before you account for inflation. There was also the cost of upgrading. If you were moving from an Apple II to a Lisa, then you probably had a lot of software you were already using. To actually get up and running with your new graphical office, you'd have to purchase new software. Now, in practice, I know that wasn't

an option for the Lisa, but let's just pretend that there was third-party software, that this was possible. This would make switching to the new platform a huge task. What you get in exchange is a totally new type of machine, at least in theory. The Lisa was fully multitasking, but it was slow and unreliable. The hardware, as designed, didn't quite have the power to do everything Apple wanted it to. So those who jumped to the new platform were, by and large, disappointed. The software route is what we saw with Vizion. That interface ran on very specific PC hardware. You needed a CGA card, which was a mid-tier as far as graphics cards went back then. You also needed a hard drive, which was a big expense in the period. It was also pretty expensive software.

It cost about $700 to get Vizion and a compatible mouse. That was before you bought any of the Vizy software that was compatible with Vizion. The benefit of the software approach is that you can tap into an existing market, namely the PC market. If you already had a PC then congratulations! You could jump into the new graphical world. You could keep all your existing software, but add a cool new interface. That is, if your computer had the specific specs. For Vizion, that was a bit of a high bar.

So when Gates started telling folks he could do one better. Well, that was a big promise. Specifically, he was doing the tried and true Microsoft maneuver. Gates wanted to sell the product to OEMs, to companies that made and sold computers. An end user could still buy the software, they could still own Interface Manager, but the big bucks would come from a company like IBM or Compaq bundling Interface Manager with their next PC. Gates backed up all these claims with a demo. It showed what looked very similar to the Lisa software running on a PC. Or at least something similar. Maynes and Andrews described this demo somewhat in a Computer World article, and later in

their book Gates. The demo had windows that could overlap, it ran multiple programs simultaneously, and it used a mouse. However, it was all a bit of a trick. It became known as the Smoke and Mirrors demo because it was apparently cobbled together from whatever code was at hand. It wasn't so much a running system as a cool mockup of what could be. The Smoke and Mirrors demo first appeared in April of 83.

It would get pulled out to convince OEMs that Microsoft was working on a desktop environment. Somehow, despite its dubious state, that was enough to keep OEMs at bay. It was enough to get some buy-in on the project. But this was still all private stuff. The outside world knew Microsoft was doing something graphical, but not much more. That is, until September, when Microsoft finally showed Windows to the press. This version is sometimes called the Byte demo, because it would be most widely publicized by Byte Magazine. We'll get into that part of the story, but for now I'm just going to use that name without

context. You will forgive me. The byte demo looked a whole lot like something we've seen before. It was multi-plan. Specifically a very souped-up multi-plan. From glossy print screenshots, we can see the logical next step in the multi-tools family. You have a big chunk of screen up top for displaying graphics, and then a command menu at the bottom. You can click on that menu to run commands, all the expected stuff.

Now, of course, there are some actual differences here, at least graphically. The byte demo shows off data inside neat little windows with little bars around them, just like we saw with Word. The difference is that these windows each contain the interface of a running, independent program. It's not just a different text document. It's Microsoft Basic. It's Word. It's not just a different text document. It's Microsoft Basic. It's Word. It's a chart.

All running happily alongside each other. Another difference is how these windows are displayed. In Word, everything had to be tiled. You could have two windows, but each had to take up half the screen. Or at least you had to subdivide the screen. The Byte demo has tiling windows, but it also has overlapped windows. Your windows can be any size placed anywhere on the screen

and stacked up however you want. But it's more than just glossy photos. It's the details that made this demo so exciting. Here, I'm pulling from the Byte article that showed off this very performance. Windows was able to run DOS programs. Now, this is in a slightly stilted way. If your program only used DOS's API, then Windows could capture it. You get a nice window that contained your running application. Those were called cooperative programs. The problem is that DOS's programming interface is hot garbage.

Specifically, how it prints to the screen and gets keyboard input is annoying. It's slow. So many programmers handle that kind of stuff directly. These kinds of programs were called uncooperative. When you execute them under Windows, you'd be temporarily dropped down to DOS. Once the uncooperative program ended, you'd be brought right back to Windows. Like I said, clunky but perhaps the only real way to handle compatibility. You had full access to all of your DOS software even after you upgraded. That's a fantastic value proposition. If you can get graphical on your existing machine and keep using your existing programs,

well what's not to love? Then we get to the so-called smart programs. These are actual Windows programs, software that was aware of Windows itself. These programs can have their own windows resized and scaled. You can copy and paste data between them. They also make the menu bar context-sensitive. If you run the Windows version of MultiPlan, the menu bar changes as you navigate around. If you click on a cell of the spreadsheet, you get a different command palette. The final big promise in the Byte article is the larger Windows ecosystem.

Windows, like DOS, was planned as an open platform. You could get a development kit from Microsoft and then write your own Windows software right on your Windows PC. That's in stark contrast to Vizion and the Lisa. Vizion's development tools all ran under Unix, and PCs couldn't really run Unix at the time, so you'd have to have some other development machine, some mini-computer. For the Lisa, you had to boot into a special operating system to write graphics software. But for Windows, you just had a lot more convenient options.

Microsoft also promised wide-reaching OEM support. The Byte article actually ends with a gallery of 15 different computers all running Windows, including the Apple II. Although that was only possible with a special expansion card, but still, that's pretty cool. Now there is the question of if this was a real program or more smoke and mirrors. Ever since 83, folk have been trying to track down the software shown in the byte demo. We've gotten close. Someone actually found a floppy disk with the label Microsoft Window Manager Demonstration 09-1983. But the disk had been damaged, supposedly by a magnet.

There are rumors that similar discs are floating around in private hands to this very day, so ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo answers about the demo one day. But this was just the beginning for Windows. In November, the Fall Comdex show started. This was a huge affair. It took up a whole swath of Vegas, and it ran all the way into December. As Maynes and Andrews describe it, Gates decided to paint the town red, or I guess, maybe the weird bright green that gets associated with this version of Windows. Whatever the color, Gates would fork out a wild amount of money to advertise the new software. We're talking flyers, signs, door prizes, billboards, the works.

If you were going to Comdex, you probably knew about Windows while you were still in a taxi from the airport. With all this marketing came a new build. This is where I think we see a little bit of smoke leaking out. What Microsoft showed off at Comdex bore only a passing resemblance to the byte demo. Gone were overlapping windows in favor of

only tiles. The bottom menu was replaced with a shelf for icons. Each window now sported its own menu bar with drop-down menus. It was a pretty radical shift, to say the least. As to why this version is so different after just a month, well, we have some rough ideas to work off of. For this, I'm pulling more details from Mains and Andrew's book, Gates. In June of 83, Microsoft had poached Scott McGregor, another Xerox developer. He would become one of the many people in charge of Windows 1. Leading up to McGregor and, after, there had been a bit of a rotating cast.

All they really had at that point were demos and vague plans. Remember, this is summer of 1983. This is mere months before Windows was announced to the entire world. In that light, I think it's very likely that the byte demo was either Microsoft's patented magic act, or that development was just moving so fast so early on that the demo was obsolete before it ever hit a floppy disk. It also sounds like Windows had entered a special kind of development hell, as McGregor would put in an interview for the book Hard Drive, quote,

There was this very real conflict between Bill's desire to make Windows better and the need to have a plan to execute to actually get the product out, end quote. Gates, it seemed, wanted everything. He wanted the perfect product. Part of the drive was probably fueled by watching the Lisa and Vizion fumble their way towards market. Windows had to be perfect, but that's not how software works. You have to set concrete goals with reasonable deliverables. You have to have attainable endpoints. If you don't, then you get stuck in this flapping cycle, jumping from one shiny object to another.

That could explain the radical shift between September and November. It could have been a flap in one of these big cycles. Following Comdex, we enter into another type of cycle. The Byte article is actually published in December, AFTER the show. So uh, yeah. The article describes a demo given in September but comes out in December. How's that for yesterday's news? Windows was initially supposed to be released in 1984. That would end up getting pushed back to 85. Part of that was just scale. Once

again to quote McGregor via hard drive, quote, I don't think Bill understood the magnitude of doing a project such as Windows. All the projects Bill had ever done could be done in a week or weekend by one or two different people. That's a very different kind of project than one which takes multiple people more than a year to do." This, dear listener, is what development hell sounds like. Microsoft was pushing on the edge at what was possible for them. This was, for the company, uncharted waters. As a result, we get this two year period where Windows is vaporware. But it was a very visible vaporware. Over the period, Microsoft would continue talking to the press, showing demos, and even sending out early versions of Windows to developers and OEMs.

A lot of these builds are well documented online, but my favorite source for this that brings them all together is the Beta Wiki. In February of 84, Windows shows up in a computer magazine in Japan. In May, the first developer preview is shipped. This is followed by updates in June, July, August, and September. By 1985, Windows had entered this cycle of alphas and betas. So there was code flowing. We even have preserved copies of some of these builds,

but this wasn't publicly visible. Developers and OEMs that Microsoft trusted were receiving Windows, not the public. Even in these developer releases numbered DR1 through DR5, we can see turmoil. So check this out. This is from the release notes for DR4 from late 1984. Quote, This release of Windows represents a substantial increase in functionality over previous releases. New features include a multi-tasking scheduler, a dialog box facility including a text editor, many new features in the MS-DOS window, new GDI functions including filled polygons."

It continues. It took until August, 84 to implement multitasking. That was supposed to be a core feature of Windows. It was shown in demos all the way back to 83, but apparently that just wasn't real. It didn't actually exist until the end of 1984. That's a little mind-boggling to me. It would probably be correct to call Windows vaporware in this period.

It had yet to live up to its promises, it had yet to ship, it hadn't arrived. Perhaps it's no surprise that the public view on Windows was mixed at best. An article in PCMag from the summer of 1984 says as much. To paraphrase, they say there are three keys to the possible success of Windows. Third-party software support, OEM support, and a promised launch date in 1984. Of those three, two were shaky. The launch date kept being pushed back, so that one's out. Software support was actually looking good, thanks to the developer release program.

Third-party developers were, in theory, working towards shippable software as early as 1984. In this respect, Microsoft had Apple and Visicorp beat. That's a big up, but the OEM piece was a little strange. Microsoft had a good relationship with dozens of manufacturers, but they were missing one, IBM. This comes up constantly in period coverage. Microsoft seems to have had everyone on board except Big Blue. This is, perhaps, one of the first signs that Microsoft and IBM weren't exactly close friends. The reason IBM was lukewarm on Windows was simple.

They had their own graphical project. It was called Top View. It wasn't all that remarkable and wouldn't make a big impact on the market once it came out, but it was the horse that IBM was betting on. Big Blue had, in large part, made Microsoft's fortune. Now they would be competing. As the release date from Windows continued to slip, IBM would publicly announce its entrant into the market.

Top View was classic IBM, the whole way. It was fully multitasking, perhaps more capable than anything else running on the PC, but it used an old-school text interface. Windows were even rendered using ASCII art. It's actually really interesting to see this development in the timeline. By the time the alpha version of Windows 1 is completed, internal documents start to reference Top View. In April 1985, PCMag ran a long series on TopView, which had this juicy interview, quote, Microsoft Windows is willing and almost ready to compete with IBM's TopView, says Tandy Trower, Microsoft's Director of Retail Marketing for Systems Software Products. The article continues with a quote from Trower, quote,

What you are really seeing is Windows taking Top View and some of the features you see in the Macintosh and rolling them together so you get the best of two worlds. You don't have the limitations of the Macintosh and being single tasking because Windows is multitasking like Top View. Also, you don't sacrifice the graphics as Top View does. like TopView. Also, you don't sacrifice the graphics as TopView does. You are allowed to do a lot of graphic interactions using icons and pictorial representations like the Macintoshes." Windows had started off as an attempt to capture the power of Vision. By 85, that strategy had widened, and it was even publicly acknowledged. One internal document found with the alpha of Windows 1 literally has a header that just says Mac plus Top View plus Device Independence equals Win.

It's worth noting that the abbreviation for Windows on DOS machines was classically WIN. Microsoft wants to take the best ideas they can find, scalp them, and build them into something they can sell. This can be seen in aggressive hiring of Xerox employees, but that's just one aspect. In this period, Microsoft was writing a lot of software for the Mac. Those same programmers who were learning the Mac inside and out were routinely switching over to work on Windows. While taking in multiple points of view and inspirations can be a very powerful tool, the strategy also has its downsides. The final days of the Windows project sound like a slog. It was scheduled for release

in the summer of 85. The final date, the real one this time, would be November of 1985. In the last few months, applications had to be written for Windows. Memory usage had to be squashed down. Bugs had to be found and designs had to be finalized. It was a classic mad dash. But by November, it was over. Finalized versions of Windows shipped to OEMs, and blue boxes started hitting shelves. The released version of Windows 1 actually worked, and it worked with pretty limited resources. It used only 256 kilobytes of memory while running. Now, that may sound absolutely tiny, but consider this. An Intel 8088 processor could only address 640 kilobytes of memory. So you could load up a few programs,

but not too many. It would be easy to run out of space. That said, I think it would have been hard for Microsoft to do much better. This aspect was always going to be a balancing act. As far as graphics, you needed at least a CGA card. That meant Windows wouldn't work on the lowest of low spec PCs, which usually only had monochrome display adapters. But all things considered, that wasn't super restrictive at this point. And finally, there was no requirement for a hard drive. You could run Windows from floppy disks. That was perhaps one of the bigger selling points. Hard drives were

very expensive, so the fact that Windows didn't need one of those was a big deal. The CGA thing can be forgiven. So then how was Windows received? Well, let's call a little mixed. We should start where the entire story began, at Comdex. That fall, back in Vegas, there was a mythic event, the Windows Roast. Here's PCMag, quote, The biggest software story at Comdex had already been the talk of the industry for two years. At a first-night party, Microsoft Windows was roasted by its creators and then distributed to partygoers from a golf cart." From my digging, this seems to have been a real event. These big productions will become,

in time, another Microsoft classic. So hey, maybe this is where it all started. Now, we don't have a recording of the roast. I was trying to find a video, but I think we're SOL as the saying goes. We also don't have a wild amount of information. But we do have some period articles. According to a 2010 retrospective in PCMag, the roast included a number of editors and writers from various magazines and trade presses. They had actual comedians write the joke, so in theory it was funny. Luckily, one joke has been passed down to us, which I'd like to share here today. When they began coding the product, Steve Ballmer still had hair. Eh?

That good or what? Honestly, I think the roast was a good call. Windows had started to become something of a joke in the industry. The word vaporware was commonly printed in articles. And it was finally over. It was finally here. It had arrived. It could be passed around from the back of a golf cart while Steve Balmer chuckled in the corner at a bit of a witty dig. In the coming days and

months would come rave reviews, but it's not clear if everyone got windows. One PCMag article from late November called it a quote-unquote DOS extension, and promised it would make DOS so much easier to use. Which is true, but also concerning. Windows by this point was nearly its own operating system. It had a memory manager, a task scheduler, it had a runtime environment, and hardware abstraction layers. That is just about everything you need for your own operating system. It just happens that you loaded up DOS before running

Windows, and you could run DOS programs from inside Windows. There is one stripe of press that talks about Windows like it's just a nice little thing you put on top of DOS. That belief could make all the delays seem a little ridiculous. Why would it take Microsoft over two years? Windows is just a graphical shell on top of DOS. That should be really easy to ship, right? This made Microsoft's approach a bit of a doubled-edged sword. The fact that Windows could work so well with DOS software was a huge plus, but it made this type of misunderstanding possible.

But enough about Windows itself. Remember that this didn't happen in a vacuum. How did it compare to the competition? Well, sales-wise, mixed. It's all really mixed when it comes to Windows. It did beat Vizion, but that turns out to be a low bar. Vizion actually flopped at the market completely. Apple, however, would put Microsoft to shame. By April 1987, 500,000 copies of Windows 1 had shipped. That's about 18 months to reach the milestone. It had taken Apple 20 months to ship that many Macintoshes. But by spring of 1987, there were a million Macs in the wild.

It's also a good time to mention that the unit price for a Macintosh was much much greater than the unit price for Windows. Apple hit a kind of inertia that Windows 1.0 couldn't really keep up with. There are a few ways we can look at that. Perhaps Microsoft was too slow to come to market. Things could have been different if Windows hit shelves before the Mac. But looking at the development to market. Things could have been different if Windows hit shelves before the Mac. But looking at the development history, I think that would have been impossible. Perhaps the software route failed. One of the core ideas behind Windows was creating software for existing hardware, for an existing

user base. That in theory would be cheaper and would have a built-in consumer base. Maybe that idea just didn't work. Maybe it was better to create custom hardware, sell new machines. But on the other hand, later versions of Windows would essentially succeed using the same model. The final option is that Windows 1, despite all the work that went into it, just wasn't very good. In 1988, Stuart Alsop, one of the attendees of the Windows Roast,

published a special article in PC Letter. That article called the War Over User Interfaces, WUI, gives a rundown of the current state of GUIs on microcomputers. 1988 is about a year after Windows 2.0 was released, but Alsop's critique holds a lot of truth for Windows 1 as well. The section on Windows is called Eclecticism in UI. That, perhaps, gives you an idea of where this is going. Quote, In trying to design a user interface for IBM-style machines, Microsoft's two biggest problems were that IBM considered system software optional and that IBM managed to sell more than 5 million

computers before it even knew that there was such a thing as a user interface. Apple had a much easier time with the Macintosh because it was a brand new system." According to Allsoft, Microsoft was trying to fit a square peg into a round hole. The PC was a perfectly fine DOS machine. It could revolutionize the office, but it didn't have the power to really do a GUI. By targeting the PC as the lowest common denominator with its tiny Intel processor, Microsoft severely hamstrung Windows. It was a repeat of multi-plan.

The result is slow, sometimes unresponsive, and it couldn't really run all that many programs at once. IBM also hammered at least one nail into the coffin here. The main reason for MS-DOS's success was that IBM shipped it with PCs. You get a PC, you get docs. It became ubiquitous. Microsoft was trying to do the same thing with its cadre of OEMs. You buy a computer from Compaq, you get Windows. The IBM PC had been cloned by 85. You could buy a PC and never know that IBM existed, except that your computer was compatible with those three letters. But in this period, in

this specific year, not many people were that isolated. IBM was still a gold standard. Clones don't overtake IBM hardware until 1986 or so. Since IBM doesn't back Windows 1.0, it doesn't get the IBM bump. It doesn't hit ubiquity. Sure, if you buy an HP, it comes with Windows, but if you buy IBM, you just get DOS. And no one ever got fired for buying IBM. Back to the topic at hand, to the Windows 1 sucking theme. Capability-wise, it could be on par with the Mac, but it was clunky. It was a bit of a mishmash of features. It was eclectic, as Alsop said.

It also didn't look all that nice. One of the big complaints was tiling, so I'm gonna single that out. Very early versions of Windows 1 supported overlapping windows. It may sound like a small detail, but it's the kind of thing that's super convenient and super flashy. It's showy. detail, but it's the kind of thing that's super convenient and super flashy. It's showy. That was dropped early on in favor of tiling.

That works, but it's clear no one liked that as a solution. The programming staff even wanted to switch back to overlapping windows, but there wasn't enough time to win that fight. One theory I've heard is that Microsoft went with tiles to avoid a lawsuit with Apple, but I don't know how much I believe that. It makes more sense as a time-saving exercise to me. Once again, it's small, but it makes Windows look more primitive than the Macintosh's Finder. There's also things like Windows 1.0's wild use of colors. Seriously, you should look up a photo of Windows 1 in full color.

On a color monitor, when Windows 1 boots, you're greeted with a bright green background – yellow, blue, and red in addition to the expected blacks and whites. Sure, it's a celebration of what color graphics can do, but it's all a little loud. It makes Windows 1 look like a toy. A lot of early articles on Windows show it in monochrome, which makes it seem a lot more serious. The Premiere Edition, the last release before launch, actually defaults to black and white. You only see color as small pops in some applications. The Macintosh, famously, starts as a black and white machine.

I personally think that made it look a lot more serious. At least when you put Windows 1 and the Mac's Finder next to each other, Windows looks far less mature. Folk even picked up on this in the period. That same Alsop article sings the praises of the maturity of the Macintosh over the hodgepodge of Windows. Additionally, once Windows 2.0 rolls around,

the colors are very much toned down. A final nail to drive home is third-party software. From my digging, I think Windows only had 60 third-party titles, that is, software written explicitly for Windows 1.0 that didn't come from inside Microsoft. That listing from kernel386.com comes with an interesting note. Quote, despite the fact that some programs were already announced for Windows in 1986 or 87, by the time they actually shipped Windows 2.0 was already almost done or already available. So many of these programs ended up targeting

Windows 2 instead. Though some applications will run fine under Windows 1, I've only included those that officially supported Windows 1 in at least some capacity in the list below. Often only a short-lived initial release supported Windows 1 before the application was adapted for Windows 2. This makes many of these applications below very hard to find these days." Developers weren't quick to adopt Windows. By the time they were on board, Microsoft was already moving on. This presents this weird snowball effect. Windows didn't have much third-party software. It's that software that usually draws in users, like how VisiCalc sold the Apple II.

You don't use DOS so you can look at disks, you use DOS so you can run VisiCalc or Word. Similarly, you're not going to use Windows to look at icons, you use it for some killer app. Windows 1 didn't have a killer app. It didn't have many applications at all, so there was a lack of pull for users. With fewer users, third-party developers had less incentive to publish programs for Windows 1. The cycle continues as so. In the end, Windows 1 only really lived for two years. At the end of 87, a new version was released, and Microsoft would move on to 2.0.

Hopefully they learned a lesson or two along the way. Alright, that does it for our look at Windows 1. Slippery though it may be. The initial idea was for Microsoft to make their own Vizion, or at least to compete in the same market. In that sense, Gates had a winning idea. He was able to see a GUI, realize it was the way to go, and act on that idea. The path to that future, however, was a little difficult. For years, Windows was vaporware, but as I said, it was very visible. That's one of my enduring fascinations here. The publicly displayed images of Windows 1

shifted so much over the years. The beta wiki has a list of different versions and builds that appeared in the press. That comes out to 30 different builds before the retail launch. I suspect there were more. The fact that we have this kind of visibility into the development of Windows in real time is unique. It's a pretty interesting and wild state of affairs. We don't have full eyes on what's going on inside Microsoft, but we get these 30 glimpses of how the system evolved.

Some glimpses show a wild leap, like the shift between the byte demo and the first context build. Others show smaller steps. Along the way, Windows 1 as we know it was formed, for better or worse. Thank you for listening to Adjunctive Computing. I'll be back in two weeks' time with another piece of Computing's past. In the meantime, you can find links to my Patreon, information about the show and my sources and anything else you could want.

Contact information, though works, over at AdventOfComputing.com. Until next time, have a great rest of your day.