Command Line Heroes - Floppies: The Disks that Changed the World

Episode Date: March 10, 2020

The floppy disk was one of the greatest breakthroughs in computing. It helped spin up the software industry with a format that endured for decades. And in some cases, it’s conserved treasures once thought to be lost forever.  Before floppy disks came along, computing was weighed down by punch cards and magnetic tapes. Steven Vaughan-Nichols describes the magnitude of the changes brought by the floppy disk. Dave Bennet explains how the need for permanent storage, which was also easily mailable, led to the first 8-inch drives. George Sollman recalls how he was tasked with creating a smaller floppy, and what unexpected sources inspired the next design. And when Sollman showed it to the Homebrew Computer Club, a couple of this season’s usual suspects asked him to see more. And the rest is history.  Or is it? Matthew G. Kirschenbaum points out that floppy disks are still in use in some unexpected places. And Jason Scott and Tony Diaz tell us how they brought some source code from the sneakernet to the cloud. If you want to read up on some of our research on floppy disks, you can check out all our bonus material over at redhat.com/commandlineheroes. You’ll find extra content for every episode. Follow along with the episode transcript.

Jordan Mechner is a pack rat. He's the creator of games like Karateka and The Prince of Persia, and he meticulously saved everything along the way. Journals, sketches, and storyboards, all of it. So it came as a bit of a shock to him when he couldn't find something he'd saved. And that something was a pretty big deal. Back in 2002, Mechner was working on Prince of Persia,

The Sands of Time. The programmers wanted to add the classic version of the game to their PlayStation 2 update as an Easter egg. So they asked him for the original source code. But when Mechner looked in his archives, he couldn't find it. He searched everywhere. The source code that he'd written on his old Apple II, the code he was positive he saved, had vanished.

Fast forward ten years. Mechner's dad is cleaning house. And buried at the back of a closet is a ratty-looking shoebox holding a bunch of dusty old three-and-a-half-inch floppy disks. One is labeled Prince of Persia Source Code, copyright 1989, and in brackets, in all caps, the word original. The long-lost code, found at last. It had sat in that box for a quarter century before being unearthed like some archaeological discovery. But this was 2012. How would he be able to get it off those old disks? And would the data still be intact? Was it, in fact, too late to save his work? Saving our work. These days, it often happens automatically, with programs regularly

pushing stuff into the cloud. We don't worry about manually saving things anymore. In fact, a whole new generation doesn't even know what that save icon represents. Side note, it is not a vending machine. But for many decades, saving, storing, and transferring our data had to be done using some physical medium. When the personal computing revolution took off, which we heard about in our last episode on the Altair 8800, there was one piece of technology that became synonymous with saving. The floppy disk. It seems so simple now, but floppies

changed the course of our history because they helped turn microcomputers into personal computers. I'm Saran Yitbarek and this is Command Line Heroes, an original podcast from Red Hat. Let's put a pin in Jordan Mechner's floppy disk discovery for a moment. Welcome back to it. First, though, I want to learn how the floppy disk was born in the first place and how it became such a crucial part of the tech world for almost 40 years. Our first stop, 1890. Before electronic computers existed,

there were electrical mechanical computing devices. And the method for getting data in and out of them was through punch cards the size of a dollar bill. When electronic computers came along in the 50s, IBM standardized those punch cards with 80 columns and 12 rows. A punched hole would form one type of character. No hole meant another. For a long while, those punch cards were the main method for data input, but handling hundreds of cards for bigger, more complex programs was hugely cumbersome. There had to be a better way to save and transfer information. Next up, paper tape, which came along in the 1950s. To hear how paper tape played a central

role in the origin of personal computing, listen to our last episode. Paper tape had the same punched-hole method of reading data as punch cards. But because it's all one tape, no one had to worry about getting cards mixed up. It could carry more data and was much faster to use. But as many computers grew in capacity, they needed more and more tape to store programs. Like punch cards, paper tape eventually met its limit. Enter magnetic tape. The key ingredient was mylar, a tough, flexible material coated with magnetic oxide to make the tape recordable.

Nine tracks could store up to 175 megabytes per tape. That was a big deal in the 50s and 60s. Magnetic tape drives of 10.5-inch reels became standard issue for businesses. But the problem with tape is that while it's great for moving large chunks of data from one place to another, it's really hard to search on them to find anything in particular. And we would install software on our mini computers and mainframes using tape, but it really wasn't that good for anything small and portable or if we wanted to do anything interactive with our data. That's Stephen von Nichols,

contributing editor at CBS Interactive. Sure, magnetic tape could store a lot more data, but it was too big and slow. It was only practical for the mainframe world, really. Again, there had to be a better way. And that better way came along in 1956, when IBM launched its very first disk drive, the IBM 350 Disk Storage Unit. It was a component of the 305 RAMIC mainframe computer, a machine that filled an entire room. Here's Dave Bennett, a former IBM disk and storage product engineer. There was storage in core memory. In fact, the disk storage

device, of which RAMAC was the first, was a storage device that permitted random access to a given record as opposed to a tape drive. Interesting thing, that disk drive almost didn't see the light of day because it threatened IBM's punch card business. But the project was eventually approved. Problem was, the drive contained disks made of solid metal. RAMIC literally weighed a ton. It had to be moved with forklifts and transported by large cargo planes.

Not the most convenient storage method, but out of that came a better solution. Well, the floppy disk was originally developed for a new need. And the reason was that there was an intermediate kind of storage. Originally, there was the computer code, and then there was the computer memory, the working memory. But with System 360, there was a new class of memory in between, which they called firmware. And in System 360, there was unique technology for the firmware in various forms. It was either a special kind of a punched card, or there was a thing

called transformer read-only storage. But the new need was the desire to go from these technologies to semiconductor technology in the days when semiconductor technology was volatile. That means that the memory in semiconductors went away when the power was removed. So there had to be a way of recharging, bringing the program back into that memory when the power was restored for loading what was called a micro program or that intermediate memory. And the need for such a device is what caused the development of the floppy disk drive. So in 1967, a small team of engineers led by David Noble started developing an inexpensive system for loading those micro programs into mainframe computers. The codename for their project was Minnow.

Noble personally went through all the things that he could think of, including various forms of punch cards, including use of tape cassettes. And I don't know what else he went through, but he hit on the idea of using an inexpensive form of disk drive based on a flexible disk and a very inexpensive read-only mechanism. The Minnow team wanted to be able to mail their micro program to various locations that needed to load it. So the product for sending that program around had to be durable enough to fly through the mail without having its data damaged. Some kind of casing.

Now what they actually had to do in order to make it mailable was they decided to put it in a plastic container that was fairly rigid and they would actually read and write the disc while it was inside of this plastic container like an envelope, a plastic envelope. And when you have a coating on a disc and a rigid head, you're going to have wear. And when you have wear, you have wear particles. And the problem they had was that as the wear particles built up, it's kind of caused an avalanche effect. The particles would act as additional abrasive. And then pretty soon, with the particles being loose in there, you'd wear the recording track out and it didn't work anymore. So a really smart guy that was on that program, a man named Herb Thompson, came up with a plan that was based on a household dusting fabric that 3M sold to housewives for dusting their furniture.

And he put a sheet of that in there between the envelope and the disc, and that material picked up the wear particles, and they embedded themselves in that fabric and prevented the avalanche effect and really saved the day. Flexible 8-inch Mylar discs, One could even call them floppy. And it was all housed in a slim but sturdy envelope with that repurposed dust-wiping mechanism. Altogether, it was easy to pick up, easy to mail,

with a storage capacity of 80 kilobytes. Each new floppy disk could hold the same amount of data as 3,000 punch cards. A huge leap forward in storage capacity. IBM released their 8-inch floppy disk in 1971. The Minnow team had done good, but they had no idea how good. Their breakthrough was about to evolve into a totally game-changing technology. The head of direct access storage products at IBM was a man named Alan Shugart. The middle project was his baby. Shugart was quite a character and entrepreneurial by nature. He wore Hawaiian shirts

as his daily uniform, never a jacket and tie. Shortly after the 8-inch floppy was released, Shugart left IBM, joined Memorex in 1969, and then founded his own company in 1973, Shugart Associates, specializing in computer peripherals. Don Massaro followed Shugart from IBM, becoming VP of Operations at Shugart Associates. And a couple of years into that gig,

Massaro received a request from a man named An Wang. He was the head of Wang Laboratories, one of the leading computer companies at the time. Wang had come up with a method

for magnetic core memory that computers would use for the next 20 years. When Wang reached out to Massaro in 1975, he gave him a challenge. George Salmon, an engineer at Sugart at the time, remembers. Dr. Wang said, I really want to do the small business system and word processor, but the disk drives are out there. Your big 8-inch ones are too big. We want a couple of small storage devices that will sit next to the CRT. And we can't have the big 8-inch ones because the magnetic fields will corrupt the image. So he thought we could take the 8-inch floppy and condense it down.

So we knew we had to come up with something that was low-cost and we could execute. We put together a plain old garden variety flip chart that described what it was. And we took it back and met with Dr. Wang, who said, I like it, but you cannot use a motor in it that uses AC current because you're going to distort the image. So he said, why don't you find out who makes the motors for the tape players in, say, the General Motors cars? So we went back, we implemented some very small motors that would fit in the mini floppy that would drive a belt that would rotate the diskette. George Salmon had to figure out the specs of this new floppy disk, how much smaller it should be and what it should hold.

What we did is we looked at all of the tape players at that time, and we figured out their average dimensions. When you get all done with it, you end up with a five to quarter inch diskette size. We knew we probably had one time at bat. I think it was when Dr. Wang pulled Don and I into his office and said, let me show you your disk drives in our new word processor. And they also want to give us an order for some significant amount, 10,000 disk drives. So it was like, wow, we're now there. Now, in historical tech lore, there are a few variations to the story of how that five and a quarter inch size came to be. One states that weighing one of the floppy disks to be larger than pocket size

to avoid unfortunate bends and breaks. The favorite folktale takes place in a bar over several drinks with Al Shugart and a eureka moment when he spotted a bar napkin that happened to be five-and-a-quarter inches square. Here's Terry Shugart, Al's daughter. He liked to drink, and what I can tell you is most of his planning in any company he ever started always did take place in a bar.

And so it's not unlikely that that might have actually been the case. But what really changed the game for Shugart and his team, and frankly, for all of computing history, came down to one key moment, when George Salmon and Don Massaro decided to demonstrate their five and a quarter floppy disk at the Homebrew Computer Club. There was a Homebrew Computer Club meeting at the Stanford Linear Accelerator meeting room, which turned out to be about a mile from where we lived. So we brought our mini floppy over and demonstrated it. And there was a guy named Steve Wozniak in the audience. And Steve wanted to talk to me on it and said, I've got to have some people pay attention to this.

After the demo, George and Don went back to their regular routine. But a few days later in the office, Don called him aside. And said, you're the marketing guy. You're supposed to keep the front lobby clean, George. And I said, yes, Don, what's your point? He said, we have a bum in the front lobby. Would you get him out of here? So I went out and shook this guy's hands. And he had the most intense eyes. And I talked to him a little bit more.

And he was very interesting to talk with. And he said, I would like to walk out in the lab with you and see what this thing is that Steve talked about. I didn't know whether he was legit, but he was one of the more interesting people I've met. And it was pretty clear he had a high IQ and a huge curiosity. And that was Steve Jobs. Steve Jobs hated the cassette drives in Apple's early computers. They were always failing and took forever to go back and forth to get information. For the Apple II, he wanted a floppy disk drive, and Steve Wozniak wanted to build it. But size was key.

The floppy had to be shrunk down to make the computer more compelling to customers. So in a two-week span over the Christmas holidays in 1977, subsisting mainly on a diet of McDonald's, Woz built the prototype. It was done in time for the New Year's Consumer Electronics Show in Vegas. Woz loved building that floppy drive, describing it as the finest piece of engineering he ever engaged in. Jobs hired Shugart to manufacture the Apple floppy. The disk 2 became a major hit for Apple, helping drive sales of the Apple 2 computer and changing the industry.

Here's Stephen Von Nichols. If you bought an Apple 2, this was an option that very quickly became wildly popular. Really, you can give Steve Jobs the credit for being the person to introduce the floppy drive to the PC world. People are running this operating system called CPM-80, and there are literally dozens of small manufacturers and some big ones like DEC making CPM80 computers. And all of these computers use the five and a quarter inch floppy drive. And then when IBM comes along in 1981 with the first IBM PC, it has become clear to everyone that this is no longer a hobbyist market. This is a real business market and that this business is going to be driven by the newfangled five and a quarter inch floppy drives. The floppy disk and the personal computer revolution were symbiotic. One fueled the

other and vice versa. They evolved together. Thanks to the floppy, companies could write programs, copy them onto disks, and sell them through the mail or in stores. It did for early PCs what app stores would do for smartphones, opening things up for developers. In fact, the floppy made the whole software industry possible. And as those software programs became larger and more complex, floppies changed again. In 1981, Sony introduced the next evolution of the floppy disk, shrunk down to three and a half inches. These smaller siblings were encased in hard plastic

with a built-in metal shutter to prevent dust from getting in. They were more sturdy and they could store more data. Funny thing, despite their hard case, they were still called floppies. And once again, it was Apple that made the three and a halfs the new standard when they released their new Macintosh computer in 1984. Well, we all know about Steve Jobs and Steve Wozniak. The creators of the floppy drives like ShoeGuard are not household names by any means, but the work they did was absolutely vital. If it wasn't for that, PCs would have been adopted much more slowly. It's impossible for me to see how the PC revolution could have happened the way it did without these early drives. It just made getting data in and out of computers an order of magnitude easier than it would have been otherwise. Because floppies were hand-sized and easily shareable,

they had a social quality, like an early form of social media. You could meet up with people and trade discs. This was before people had easy access to modems, and before the internet was even a thing. People shared floppies full of programs, the same way they shared mixtapes full of Cyndi Lauper tracks. And this way of transferring data even had a name, Sneakernet,

because you literally had to use your feet to transport digital information from one machine to another. And what people put on those shared disks could be anything. Software, sure, but so much more. Especially with all the new digital creation tools that started to spring up. Jamie Levy was one of those new digital creators back in the late 80s. She's now a professor at the University of Southern California and the author of the book, UX Strategy. I certainly thought it was amazing that you could make something that was interactive and nonlinear on a floppy disk. I was making what became a disk zine.

It was at the same time as desktop publishing. I think PageMaker was coming out. People were just getting Macintoshes and people were slowly gravitating to digital. No one had ever heard of an interactive zine before, not even Jamie. But she combined animation, art and sound in an interactive interface and made them fit on a floppy. She made a bunch of copies, packaged them up and took them to a local bookstore. In most cases, they did not have a Macintosh to even look at them, and so they didn't know what was on them. So I'd say, well, here, take 10, take one home and check it out. And if you

like it, then sell them. And they would keep them by the front counter because they didn't fit nicely on a rack. And then immediately people bought them. As soon as anyone with a Mac walked in there and saw this thing for six bucks, they were like, yes, I want that thing. The bookstores kept selling out. Jamie received letters from around the world and started getting national media attention. Soon, she made a business from selling her discs through mail order. Then her career as an interface designer took off. Her story is a testament to the floppy and the power of Sneakernet. There was a time when you could run an entire operating system off a single 160 kilobyte floppy drive. But by the mid-90s, you needed

multiple floppy disks to run anything. Files got bigger and bigger, and it was annoying to do the floppy disk shuffle in and out of your machine. The writing was on the wall when the 1998 iMac was the first personal computer to ship without a floppy drive. When modems got better and the internet was easier to get on, people moved away from floppies to newer storage technology, like the CD-ROM. And from CDs, we moved on to DVDs, SD cards, USB flash drives. Turn around and there's a whole graveyard of obsolete storage technology behind us. And now we have the cloud. Poor old floppies. They didn't stand a chance. But here's the thing. Floppies linger. They have staying power. They're still the save

icon for starters. People still have shoeboxes full of them. Floppies have, in fact, proven to be one of the most resilient pieces of technology out there. You might be surprised to learn they're still used to maintain legacy machines like the U.S. Intercontinental Ballistic Missile System. Up until recently, that system relied on 8-inch floppy drives, for real. Matthew Kirschenbaum is a professor of English and Digital Studies at the University of Maryland. So there was a story that was making the rounds about the U.S. Air Force phasing out 8-inch floppy disks from its nuclear command and control systems. And so this was accompanied by pictures of people down in the missile silos plugging floppy diskettes

into sort of old-style mainframe computers. And I think that surprised a lot of people to see that something like nuclear weaponry was still controlled by eight-inch floppy disks. The government, the military is finally phasing those systems out, and I certainly think that's a good thing. But it's worth remembering, too, that floppies tend to be pretty durable and resilient as storage media goes. They're low aerial density, meaning that they're relatively coarse or low-fi as compared to hard drives nowadays, which actually made them a little bit more resilient and reliable. And they were also, because of the very age, the very obsolescence of those computer systems, they were ironically more protected against some of the kinds of networked threats that we see

nowadays in the forms of viruses and hacking and that sort of thing. The phrase people use is the air gap. And so the floppies were not physically connected to the internet in any way. And there's a way to think of that as a pretty desirable security feature. Even though it made people a little bit queasy, it's not completely absurd that they've persisted in use for as long as they have. Sure, when you look at floppies now, they look flimsy and a bit ridiculous. But with the right equipment, disks from decades ago are still readable today. And who knows what treasures can be found when that data is retrieved.

A couple of years back, there was a press release that there had been a dozen new Andy Warhol images located. It turns out that the images in question were actually digital images that Andy Warhol had created on an Amiga computer back in the 1980s. He had gotten the Amiga and was experimenting with it and began to use it for some of his artwork, and the resulting images were stored on floppies that were at the Warhol Museum in Pittsburgh. And so a team from Carnegie Mellon University was able to rescue those files. And we now have a dozen additional pieces of artwork by Andy Warhol. Speaking of treasures, it's time to find out what happened to Jordan Mechner and his long-lost source code. To recap, Jordan lost the original source code for Prince of Persia. Years later, his dad unearthed the shoebox at the back of a closet

with a bunch of dusty three-and-a-half-inch floppies. One was labeled Source Code. But was the source code really on that disc? Would he be able to recover his Prince of Persia work? To find out, we talked to two people Jordan enlisted as his Save the Code crew. The first person Jordan contacted was Jason Scott, free-range archivist at the Internet Archive. It was going to work or it wasn't, but you always want to prepare for the worst. And there's this guy I know, and his name is Tony.

And Tony's relationship to apples is astounding. He knows them inside and out. I would call him an Apple II whisperer. Via a long-running IRC channel, Jason tracked down Tony Diaz and recruited him for the job. Jordan invited this pair of computer archaeologists to his home in the Hollywood Hills. So Tony loaded a bunch of gear into the back of his truck for the long drive from Oceanside to L.A. I took a couple of Apple IIe systems that I have sitting out, packed them up, filled about half the pickup truck bed full of various stuff. And I got disk drives that are modified for reading difficult

floppies and various tools like that. Disk drive controller that does things a little differently. And some stuff on more modern computer that helps with the floppies if I have to do it. But yep, loaded it all up and hit the road. In Jordan's living room, Tony set up an impressive array of vintage Apple II computers and 3.5-inch disk drives. Then Jordan brought out the shoebox in a plastic bag, like it was the Holy Grail. This grail had rubber bands around it to try to keep the cardboard lid from falling off. I've been at the receiving end of this a lot of times where I just crack open the old box of discs and it's all got that same kind of musty smell that,

I don't know, it's hard to describe, but for those people who have been on a Navy ship, they all have the same smell. You go to an airport, the backside of it, they all smell the same. The floppy discs have their own unique, what's the word, luster to them. Anyway.

Tony pulled out a couple of the discs to see what condition they were in. He was ready to take off the protective cover if he had to and clean them with Joy dish soap. The discs hadn't been rotated for a long time. So he put his fingers in the center, gave it a wiggle and rotated it once to check for mold. Then he and Jason split off into two teams. I was going to do flux readings, which is where you take the magnetic information, pulling every single wavering of magnetic info off of a floppy. And that produces a very large and hard to decode image of the floppy. But the idea was that if anything else went wrong, or we were stymied in any single way, that we'd be able

to use that in the future. Tony understood what the information on these disks were, and he was using utilities on a machine to read the disk as if it was back in 1990 and the data was something that made sense to him. We went after the disks that were very obviously standard copies of files, trying to make sure we weren't touching any one-of-a-kind discs until we knew what we were up against. Some of the things that were on these initial simple discs were things like Death Bounce, a game that he made that never came out, along with an Asteroids clone. And they worked, and we were able to make them play on the machine, and he was looking at games he hadn't even thought about for decades, and they were running fine. So far, so good. Now for the disc labeled Prince of Persia. Tony started analyzing it and realized it was a backup of a hard drive that had been split across five floppies. He needed to put the whole thing back

together. So to restore those images, you have to have a hard drive volume the same size as the one that you wrote it from. And, of course, I brought drives, but what are the odds we're going to have the exact same size? Because he didn't always. But that's okay. I'll make a 10-meg partition on my card here, and I will make a copy of these disks, and I will tell it that you have been backed up from a 10-meg drive. And we proceed to restore it. And after that, we've got a nice hard drive full of source code files to go through.

But the big reveal happened when they did the first catalog and looked at the files. Yep, that's it. Merlin files. They all end in.s. And there's the OBJ files for when it's compiling or when it's assembling. It's all here. And, oh my gosh, there's so many different versions of it. We have to figure these out.

But all the heads basically turned pretty much concentrated right back towards the monitor and the computer. And it was kind of silence, more like, okay, here, let's see what's on this one. How about that one? Ooh, I remember this one. Once he did that and we started scrolling through the text files that were on that disk, Jordan recognized them immediately as his original source code. So we just kept going through all of the disks, even though we knew we had struck gold, just to make sure we weren't missing some other versions.

And among the things we found were other work in progress from Prince of Persia, where he had tried different graphics and so on. Amazing! The moment the team saw the directory with recognizable source code files, they let out a collective sigh of relief. Prince of Persia had been rescued from the sands of time. But their work was not quite done. Jason uploaded the salvaged code to GitHub,

where it was instantly accessible to fans from around the world. Word of the event had leaked out, and people were logged on and waiting. Our main goal that day was to get uploaded to a repository on GitHub as soon as possible. We were connected in that same IRC chat room the whole time, and we had all kinds of people asking us questions. What is it? What are you seeing? What do you got? And we're giving live feedback as we're going. John Romero, who worked on Doom and Quake, said that he learned a few tricks

looking at the source code. And a bunch of other people, of course, were completely confused as to what they're looking at because code written in, you know, the late 1980s is a lot different than today. And the fact that he is, you know, byte by byte moving memory, trying to clear space so that a sprite looks like it's supposed to. You know, that's a whole other world. Since that eventful day back in 2012, the Prince of Persia source code has been studied, shared, commented on, and cherished. Here's Matthew Kirschenbaum with some final thoughts. This is another example of the way in which aspects of what we would think of as digital culture, sort of the artwork of today,

the creative expression of our own moment, a lot of it is locked away on these dusty, obsolescent floppies, but with a little bit of luck and engineering, there are some pretty remarkable stories that have emerged about how we've gotten things back. The save icon is as close as some of us ever get to floppies these days. But when you see it, I hope you'll remember that it's part of this amazing history we should never

take for granted. A history of sharing and rescuing the things we create. A history of saving. Command Line Heroes is an original podcast from Red Hat. Check out our floppy disk research notes at redhat.com slash commandlineheroes. By the way, if we saved this episode, which is about 51.5 megabytes, we figured it would take 36 3.5-inch floppies. I'm Sranjit Barak. Until next time, keep on coding. Hi, I'm Mike Ferris, Chief Strategy Officer and longtime Red Hatter. I love thinking about what happens next with generative AI. But here's the thing. Foundation models alone about what happens next with generative AI.

But here's the thing. Foundation models alone don't add up to an AI strategy. And why is that? Well, first, models aren't one-size-fits-all. You have to fine-tune or augment these models with your own data, and then you have to serve them for your own use case. Second, one-and-done isn't how AI works. You've got to make it easier for data scientists, app developers, and ops teams to iterate together. And third, AI workloads demand the ability to dynamically scale access to compute resources. You need a consistent platform, whether you build and serve these models on-premise

or in the cloud or at the edge. This is complex stuff, and Red Hat OpenShift AI is here to help. Head to redhat.com to see how.