Advent of Computing - Episode 125 - US v IBM

Episode Date: February 1, 2024

My coverage of the IBM System/360 continues! In this episode we look at US v IBM, and the fallout that surrounded the release of the System/360. By 1969 IBM already had a history of antitrust litigation. What was IBM doing to upset the Department of Justice, and how does it tie in to the larger story of clone computers?   Selected Sources:   http://www.cptech.org/at/ibm/ibm1956cd.html - 1956 Consent Decree   https://supreme.justia.com/cases/federal/us/298/131/ - 1936 Consent Decree   https://archive.org/details/foldedspindledmu00fish/page/n5/mode/2up - Folded, Spindled, and Mutilated

To start with, let me apologize for my tardiness. This episode turned into a bit of a beast, just as I had initially feared. Today, we're talking about USVIBM. It's one of the most storied trials in the annals of computer history. Now, despite its place of prominence, it's not very accessible history. Let me explain what I mean by that. accessible history. Let me explain what I mean by that. Many of the big events in computer history are very well documented, often from multiple angles. Take, for instance, the IBM PC. I kind of have IBM on the mind lately, as you may have noticed. The development of the PC is recounted

in a wild number of articles, books, and interviews. You can get everything from primary sources up to pop histories of the PC. You even get these ancillary stories about the effects of the event itself. There is an entire book about the development of the first PC clone at Compaq, written by one of the co-founders of Compaq himself. It's a story that is very much painted in color. US v IBM is different. For one, it's a court case, so the primary documents are barely in English. The interested parties are lawyers and economists, The interested parties are lawyers and economists, so the books and papers on the trial are focused on legal and economic theory. The history here isn't documented in some nice, linear story. Rather, it's been picked apart for mind-numbing details. Most of the coverage is also biased

towards IBM's point of view. That's a fact that's been making me, honestly, very frustrated and very angry throughout the development of the episode. I can't recall the last time I was so, well, so mad at a book. For two, US v. IBM is a wildly complicated story. The lawsuit spans 13 years. The lead-up is even longer, involving over a dozen other cases. Each corner has its own rabbit hole to go down. For instance, at one point, IBM tried to declassify a CIA report on computing in the USSR. This was an attempt to submit that report as evidence in one of these trials. That should give you an idea of how attractive these distractors and rabbit

holes are. Just figuring out which parts of the story matter and which are dead ends, well, that's kind of a job in itself. Allow me to present the fruits of all this labor. I think I've been able to capture what matters to us computer nerds while pointing out the intricacies of the case. But I'm no lawyer. In this field, I may in fact be the ultimate layman. So watch out as I misinterpret speed. Welcome back to Advent of Computing. I'm your host, Sean Haas, and this is episode 125, US v IBM. Today, we're continuing our System 360 series with a discussion of the fallout of IBM's biggest computer to date. Before we go any further, let me give a quick recap and explanation as to where we are in this larger trilogy. Last episode, we dove headfirst into the System 360, the first family of compatible

computers. This was IBM's attempt to break out of a cycle of incompatible machines. The result, announced in 1964, is the 360 family. This was a collection of compatible computers that could all use the same software and peripherals. 360s also had some measure of backwards compatibility with earlier IBM computers. This new family wasn't just technologically groundbreaking, it was also hugely successful. By some measures, the release of the 360 grew the entire computing market. The success of the 360 and some of IBM's policies led to a series of sticky lawsuits. This culminates in 1969 with US v. IBM, a massive antitrust suit. This is where we are today, looking at what led to these suits and IBM's reactions in this period. All of this is built

up to a very strange chapter in IBM history. Starting in the late 60s, a number of companies released clones of the System 360. Further, IBM did not eviscerate these companies. A shallow understanding of history makes it appear that IBM's practice is to crush third parties and stamp out competition. So this should be the case, especially when it comes to clone machines. One theory, and my personal guess, is that US v IBM kept these clones alive. I think that IBM was trying to get out ahead of the Department of Justice, so they just let the 360 clones run wild. Or, there might be some more complex factors in play. Really, we're doing all of this to get from our shallow understanding of IBM's relationship with clones to a more deep

and nuanced understanding. So here's the thread that we're picking at. In 1968, during a spate of private lawsuits and just prior to the federal suit, IBM unbundles hardware, software, and support services. In doing so, they, overnight, create a third-party market for software. It's been speculated, I think rightly so, that this unbundling was IBM cleaning house. They were trying to address a possible angle that could be used in antitrust trials. This leads me to believe, or at least to guess, that IBM may have ignored clones for a similar reason. It could have all been wrapped up in these antitrust cases. So, to that end, this episode will be a deep dive into the US v IBM trial.

Just to set expectations, I'm not going to give a legal blow-by-blow of events. That would, well, that would be almost impossible to do. There's just too much. Rather, I want to look at what charges IBM faced in this period, both from the DOJ and private parties. I also want to examine how those charges are connected to earlier antitrust actions lobbied against IBM. I want to see what IBM was doing that led to these allegations, and crucially, I want to examine how IBM responded. This should give us a deeper understanding of the unbundling, of why IBM would do something

that may hurt their business, and hopefully, a nice springboard into next episode. Just very quickly, before the episode proper, I do have an announcement. Just very quickly, before the episode proper, I do have an announcement. On February 17th, I'm going to be speaking at the VCF SoCal Festival. Now, the festival itself is going on February 17th and 18th this year in Southern California, in Orange specifically. If you want the details on the actual venue and precise times and schedules, go to vcfsocal.com. I myself am going to be speaking on Saturday the 17th.

I'm going to be discussing the Cryotron and some of the recent revelations that me and one of my co-conspirators have dug up. So if that sounds exciting to you and you're going to be in the area, then I highly recommend checking out VCF SoCal. I think it's going to be a really good time this year, and I'm personally really looking forward to being there. Now, with that said, let's get on with the actual show. To begin, we must first discuss antitrust law. I know, riveting stuff. In order to prepare for this episode, I've been reading a truly horrible book. It's called Folded, Spindled, and Mutilated, Economic Analysis and US v. IBM.

It's the only book that's totally devoted to the trial. It's also written by a group of third-party economists that were hired by IBM to give an impartial analysis of the situation. The book, I assure you, doesn't feel impartial. It's very pro-IBM, very technical, and very long. I honestly kind of think that the authors have a whole section explaining how they were paid by IBM, but they're impartial just to make their pro-IBM arguments seem more valid. Anyway, with those caveats, the book does a good job explaining the complications of antitrust law. The goal of antitrust laws are to prevent monopolies in the market and encourage healthy competition. Something that Folded,

Spindled, and Mutilated points out time and time again is that monopoly doesn't necessarily just mean a majority market share. Rather, a company has a monopoly when it can exercise power in the market to prevent competition. This power can be used in a few ways. You might be dealing with monopoly if one company or a group of companies is able to prevent new players from entering the market. Or one company could use their power to force competition out of the market by, say, controlling a set of patents or preventing some type of federal compliancy. Something crucial to note here is that monopoly can be formed through totally legal means.

A company can break no single law while building up a monopoly. It's that end stage, the monopolization of a market that constitutes the crime. That's what laws like the Sherman Antitrust Act prosecute. This, however, makes antitrust law very slippery. There's this aspect of intent and outcomes that are hard to gauge. Is someone filing very broad patents just because they don't really know how to use the patent system, or are they doing so to prevent competition in order to form a monopoly? Legally speaking, both of those actions may look the same, but one is a crime and one is not. But that's kind of the mechanical nuts and bolts of a monopoly.

The biggest symptom, and what consumers actually feel, is a lack of choice. When you try and buy something from a monopolized market, you either have no options or very few options. Take, for instance, the widget market. If you go shopping for a widget, then you only have one choice. It's WidgetCo. There are some small companies that try and make widgets. There are a few regional competitors, but your only choice from a well-established company is WidgetCo. That's an example of a monopoly. There may be some small company out in Kansas that's making technically better widgets, but you're never going to be able to buy those on the full open market, since WidgetCo owns all the patents. We own all the rights to distribution. Thus, the market is monopolized, and you don't have a choice. There's no competition.

For a more real-world example, well, we can recall a fun little phrase. No one ever got fired for buying IBM. Looked at in the light of a monopoly, well, maybe that catchphrase takes on a more sinister light, but that's all conjecture. We need to get into the real meat and potatoes of this case. To do that, let me take you back. Before the US v. IBM of 1969, there were a number of near misses between the DOJ and good ol' business machines. It turns out this company has a bit of a history with antitrust law. The first cases took place in 1936. This is some real classic IBM. We're talking punch card stuff. The core of the issue here came down to IBM's leasing policies. You see, IBM never sold hardware, at least not if they could help it. In general, IBM leased their machines. If you used an IBM punch card tabulator, or any punch card equipment for that matter,

you had to pay IBM on a monthly basis. You had to sign a leasing contract. By leasing equipment, IBM, in this period, is able to keep tighter controls over their customers. This is a three-fold type of control. First is financial. Since the lease is ongoing, IBM is able to extract money from their customers every single month. That gives some leverage over the arrangement. Second is relational. Since the lease is ongoing, IBM is able to keep in tighter communications with their customers. This can help with things like servicing and education, as well as the flow of information. Third is contractual. IBM and its

customers enter into an ongoing contract, which gives IBM legal controls over what their customers can do with their IBM hardware. Remember, customers don't outright own their tabulators. They were still property of IBM. That gave IBM the power to, say, terminate your lease if you violated a contract. This, dear listener, is where things get dangerous for all parties involved. Initially, IBM's contracts stipulated that customers could only use genuine IBM punch cards with leased equipment. The only supplier of those cards was, shock of shock, IBM themselves. This case appears messy, but it's actually a case of well-settled law. The crux of it all comes down to the punch card. The crux of it all comes down to the punch card.

In 1928, IBM patented the familiar 80-column punch card. At least, it's familiar to punch card enthusiasts like myself. And really, we're all punch card enthusiasts around here, aren't we? Now, this new card was a huge improvement over earlier cards. It was more standardized, and it could hold more data. You see, initially punch cards used round holes. While easy to manufacture, this topology limits the data density of the card. Circles don't tessellate. This new 80-column number, technically called the IBM card, used rectangular holes. This allowed for holes to be more tightly packed. So, technically speaking,

this was a major step forward. That's all well and good on its own. IBM has these fancy cards and fancy machines, and they've decided the customers have to use the two together. But that's actually illegal. To read directly from the Supreme Court's decision, quote, Section 3 of the Clayton Act declares it unlawful for any person engaged in commerce to lease machinery, quote, whether patented or unpatented, end quote, on the condition that the leasee shall not use supplies or other commodities of the leaser's competitor, where the effects of the condition may be to lessen competition substantially or tend to create a monopoly.

IBM could not control where customers got their punch cards, legally. The final decision is interesting because it, in effect, voids the patent on the IBM card. Now, let me explain this. At trial, IBM produced their patent on the 80-column punch card. They had been trying to use the 80-column card patent, plus a pile of other patents on punch card equipment, to stranglehold the market. But, according to the Supreme Court, the card patent just kind of sucked. It was a lame patent. The ruling states that the patent only applies to cards that are already punched, not to unpunched cards. I mean,

that makes sense. It's also just kind of funny. An unpunched card is just a slip of paper. It has specific dimensions, specific kind of weight, but that's too vague to patent. When a user bought punch cards from IBM as per their old lease agreement, they received a box of unpunched cards. A box full of cardstock cut to specific dimensions. The contents of that box are not patentable. Also, as the court points out, the validity of the patent doesn't matter. The Clayton Act does not care about silly patents.

The long and the short of this is that IBM just gets wrecked. Specifically, they get consent decreed. Now, this is a very specific type of getting wrecked. In a consent decree, there's no admission of guilt. So IBM isn't technically guilty of violating the Clayton Act, even though it's clear that IBM was in violation of the law. IBM does, however, have to follow certain guidelines moving forward. They have to stop violating the Clayton Act. They have to let

customers buy punch cards from any vendor. IBM must also help third parties make their own punch cards. The consent decree is a little vague here, but basically IBM just has to pass out the specifications for their 80-column punch cards. Thus, a third-party market for punch cards is created. IBM can still compete. IBM can still say that their cards are the best around, but they cannot force their customers to use IBM punch cards. This also has an effect on how IBM has to treat media in the future. The consent decree covers punch cards,

but the Clayton Act itself doesn't care. The letter of the law is much more general. When IBM gets into computing, they can't, say, force customers to buy IBM-branded tape reels. That would be a very blatant violation of the Clayton Antitrust Act, and IBM would get sued again and they would lose. What we're seeing here is a history between IBM and the iron law. IBM dances close to a monopoly, the feds get involved, and IBM has to back down. There isn't a specific trial that forces IBM to follow the Clayton Act in regards to tapes, but IBM chooses to follow the law later on, most likely because they know that the feds would come knocking. They know they can't get away with that specific trick. They can't break that

specific law. Before moving forward, I should probably take a quick step back. Why does it matter if IBM is courting a monopoly? Why is that bad? Well, in the strictest sense, a monopoly is bad because it prevents competition. Under a capitalist economy, something we are very much living in, competition is one of the main driving forces of progress. The theory goes that competition forces companies to innovate in order to make better products. It also drives costs. If two companies are selling the same punch cards,

then consumers, in general, will go for the cheaper option, as long as the two options are roughly the same. This drives down prices until, in theory, the quote-unquote correct price is reached. In a monopoly, there's no competition, or at least very minimal competition. There's no real control on pricing. IBM could sell their cards at whatever price they wanted. There's also no drive towards innovation, since there's no incentive to beat a competitor. There's no competitor. IBM wouldn't have to keep ahead of, say, Remington Rand or National Cash Register. Big Blue could churn out the same model of tabulator year after year

and face no real consequences. This should all be sounding kind of familiar, at least somewhat. The computer market isn't exactly a monopoly in the modern day, but it's similar. It's a monoculture. You do have a choice of manufacturers, but when you get down to it, you just don't have a choice of architecture. You get any kind of computer as long as it's a PC. We've seen, time and time again, how that monoculture has similar issues to a monopoly. Innovation is stifled unless you're innovating a new kind of PC. It's hard for new companies to enter the market unless they're creating a PC. That aside, well, aside, let's get back on track. Fast forward to 1956 and IBM's back-in-court. This is another

antitrust case, this time ending up in the Southern District of New York. Throughout the 1950s, IBM continued to dominate the back office. According to the Department of Justice, IBM had a 90% share of the tabulating market. That raw level of domination, plus IBM's old ways, led to another antitrust suit. This time, the suit was a little more broad. It focused on IBM's practice of leasing in general, plus a side of patent law. The leasing stuff here is pretty simple. IBM would only lease hardware, never sell. This gets us into some tricky territory yet again. The argument is that this lease-only stance was often used by corporations trying to form monopolies. So if IBM is following this practice, and they have a 90% market share,

well, maybe there's a correlation there. Another side to this leasing argument is that it prevented experimentation with tabulators. Now, this might sound kind of odd at first. Alesi didn't actually own their IBM tabulator. The physical machine was still owned by IBM corporate itself. That meant that, among other things, a customer couldn't repair a tabulator on their own. They also couldn't mess around with its guts. The lack of control could, in theory, stifle experimentation. It could stifle competition, or put in a more charitable way, it could limit new developments in computing. There is a bit of an anecdote about this. Way back in the 1930s, one John Atanasoff was trying to automate some complex calculations.

At one point, he took a stab at the problem using an IBM tabulator. The machine just wasn't up to the task, but it was kind of close. So he decided to crack open the machine and add in some of his own circuits. At Nassoff, he shared this story in a 1980 lecture, but apparently he ended up publishing his results mucking around with his tabulator. At the time, IBM kept quiet. But years later, documents were discovered during the course of one of IBM's many trials. There was a memo that said something to the effect of, Keep Atanasoff out of the IBM tabulator! At the time, Atanasoff wasn't a huge figure. He was a professor at the Iowa State University.

But IBM was so paranoid, or so well-informed, that they kept a record of Atanasoff's slight against them. This whole situation was a direct result of leasing hardware instead of selling it. In IBM's viewpoint, Atanasoff had no right to touch their tabulators. Does this mean that men in blue suits would descend on you if you opened up a leased tabulator? Well, not necessarily. For IBM, this was more a matter of intellectual property. IBM backed all their products with patents. That's just kind of how tech companies have always worked.

But patents can only protect so much. Specifically, a patent lists a number of claims. You have certain legal rights over those claims. But, and here's the crucial technical part, there's some wiggle room in all of that. You can't patent overly broad ideas, or rather, you can, but those patents aren't enforceable. It's also possible for some third party to come very close to violating a patent without actually stepping on IBM's legal feet. The whole punch card debacle from earlier is a perfect example. The patent was ruled to only apply to pre-punched cards. IBM still had a valid patent on the specific methods for reading and punching cards, but they didn't have patent over

the physical shape of a piece of cardstock. So third parties could legally make blank cards without violating any patents. This all gave IBM an incentive to keep the inner workings of their tabulators at least quasi-secret. In theory, a competitor could swoop in, tear apart a tabulator, and figure out how to make their own. If they were smart, they could even do so legally. This dastardly third party would just have to tread carefully over IBM's patents. The leasing system acted in this way as a safeguard against just such a disaster. If someone did break down a tabulator and they used that information to make a competing machine, well, that would violate a lease agreement. IBM would have grounds to sue. Hopefully, you're starting to get a feel for what's going on.

IBM is using legal means to exert control over the market. The means themselves are legal, but the ends are not. That's why IBM keeps getting hit with antitrust lawsuits. The 1956 case also ends in a consent decree, which comes with pretty broad sweeping effects. The first is that IBM has to sell their equipment. Gone are the days of lease-only hardware. IBM can still offer leases, but they have to give consumers the option to outright buy equipment. This is backed by a pile of legal jargon about how IBM has to be reasonable about the value and incentives of leasing versus buying hardware. From this, we get a cascade, so to speak. One part is servicing. The court ruled that IBM had to break off their service department to form a separate company.

At the same time, IBM had to publish manuals to allow third parties to service tabulators. Suddenly, people outside IBM could own IBM tabulators. That meant that IBM could no longer force its customers to use IBM technicians. This formed a new third-party market for servicing. The next part of this cascade was the used market. Once again, you could now own an IBM tabulator. That meant that, in theory, you could sell an IBM tabulator. You could resale it. In order to make resale possible, these new tabulator companies needed servicing information on IBM machines. So, once again, IBM had to provide technical information about their tabulators.

The final piece of the dissent decree, the last part of the cascade, is patent cross-licensing. last part of the cascade is patent cross-licensing. Now, this is about as complicated as it sounds. IBM was forced to license any patent to a third party. The caveat was that IBM could charge a quote-unquote reasonable licensing fee, and IBM could expect that third party to also license patents to IBM. That's the cross part of cross-licensing. I realize that still sounds a little confusing, so let me lay out an example. Remember, the whole consent decree is about opening up the market. So let's say I wanted to start a new venture at WidgetCo. I want to develop a board that plugs into an IBM tabulator and beeps when the number 13 is passed through the bus. tabulator and beeps when the number 13 is passed through the bus. I call it the beep board, and it's going to make WidgetCo the first multi-trillion dollar company. Prior to 1956, I couldn't do this.

IBM would sue me if I tried, since I'd be violating a lease and infringing on their patents. But let's say I wait until 1956. The consent decree is well in hand. That means all I have to do is write a letter to IBM, it has to be in writing as per the decree, and ask to license the patent. Big Blue would then respond with licensing terms and some low fee. I'd pay the fee and probably also buy a tabulator for testing out the whole beep board idea. To finish the deal, I'd have to state, in writing, that WidgetCo was willing to license any beep board patents back to IBM. Once the beep board was patented, and probably after it went to market,

IBM would see how valuable the idea was, and then someone over there could write me a letter and ask for a license to that patent. I would also charge them a very reasonable fee for the trouble. I hope that kind of clears things up a little bit. The whole point is, like I keep saying, to open up the market. The beep board would be an example of a plug-compatible market, a market for third-party devices that would plug into IBM hardware. This was a totally new type of market, a market for third-party devices that would plug into IBM hardware. This was a totally new type of market, one that would provide better options to consumers and spur innovation. In effect, IBM was no longer able to keep folk like at Nassau out of their

tabulators. So why have I been taking us through all these old IBM court cases? It's simply to establish IBM's history. Over the years, Big Blue had been flirting with Monopoly. They had been caught and punished a number of times. The consent decrees meant that, technically speaking, IBM had not been found guilty. But the fact of the matter is that IBM had violated the letter of the law, and they had done so in order to create a monopoly. But this was all during the punch card era. 56 is very early into the digital game. That would all change once we reach the

1960s. This would all change in the era of the 360. The world of the mainframe was very different than the world of the tabulator. Computers are leagues more complex than punch card equipment. It all hinges, at least in my mind, on the fact that computers can actually execute software. A tabulator can't run a program, it can only crunch numbers. A computer can crunch numbers and make choices. That last aspect sounds small, but it makes all the difference in the world. It means that a computer is a general purpose tool, while a tabulator simply is not. So it's no surprise that the computer market, even the early market, was much more complex than the tabulator market.

Now, there are some caveats here that I want to be clear about. There were actually a wide range of uses for tabulators and a surprising amount of variety of punch card machines back in the day. It's just that all that equipment basically did one job. Eat the card, read the card, punch the card. Computers can just do so much more. So they were put to use in many more places and in many more ways. The bottom line here is that the computer market was a lot more complicated than previous markets. Let's just look at things as they stood in 1965, the year the System 360 shipped.

On one side, you have mainframes, huge hulking machines that take up entire rooms and require dedicated infrastructure. That's the realm of companies like IBM and the Seven Dwarves, RCA, GE, Honeywell, Univac, and all the others. On the exact other end of the spectrum are tiny computers, like the LJP30, a machine the size of the desk that can plug into a wall outlet and be used by one person. Then you have everything in between, or even to the side of the spectrum. In 1959, DEC releases the PDP-1, a machine that's smaller than a mainframe, but larger than a desk. These kinds of machines end up being called minicomputers. They aren't as powerful as a mainframe, they can be operated by a single person instead of a team,

and they don't need the same type of infrastructure to run. You also have companies like Ferranti, which manufactured computers specifically for industrial control, or some of the later PDP machines, which were built for timesharing with multiple users. That's not to mention all the peripherals that fill out the market. It's almost a fractal once you get down to it. We can see a microcosm of this complexity in the System360 family itself. On release day, there were three different models of 360 computers, but to be more accurate, there were three different processing units. This is a mainframe, which means the full system is composed of a handful of distinct modules. Each module had its own power requirements, its own physical footprint, and connected up to the

processor over these big chunky cables. When you bought or leased a 360, you didn't just get the computer. Let me run down the list, because it gets pretty expansive. If you wanted to work with legacy data, you needed punch card peripherals. The most reasonable choice was probably the 2540 card reader-writer, but you can also get a standalone card reader or card writer. If you wanted to do interactive things, your 360 install would need a terminal. Either a typewriter terminal, an operator terminal, or a display terminal. Maybe a combination of all of those. Depending on your installation, you might need many of those terminals even. For long-term storage, you could get a 2311, a 2314, or 2361 hard drive. Or, if you prefer something more low-density, you could also get paper tape reader-writers.

More medium-term data required a magnetic tape drive, of which IBM offered multiple models. Everything could be buttoned up with a modem and some kind of power supply. All told, a 360 install could be composed of maybe as many as half a dozen or more cabinets. Now, I want to tread carefully and be very specific here. When you bought into a 360, you weren't just buying a computer. You were buying an entire room full of equipment. It was all interconnected, and it was all manufactured by IBM. This wasn't like the modern PC monoculture, at least not at first. In 1965, you couldn't just buy a 360 processor and then add on, say, a Memorex hard drive and a Telex drive for

tapes. Everything in your data center would be badged IBM. Your software, too, would have largely come from IBM. That includes operating systems, compilers, linkers, text editors, the works. compilers, linkers, text editors, the works. Anything you didn't write in-house would have, initially, come from IBM. This was all provided for free. If you bought or leased IBM hardware, then you would never have to pay for IBM software. The same went for a slew of other services, usually just called services. Now, here I don't mean repairs. There were laws in place about what IBM had to do when it came to repairs. Here I mean things like installation, education, and consulting. This, too, was free. An IBM customer could have their entire 360 installation put together free of charge, and even be taught how to use it.

So, let's think about a big question here. How would you define the market that IBM is operating in? What is the exact market that the System 360 is being sold to? Or, looked at in another way, or looked at in another way, what could someone buy in place of a System 360? Who is IBM's competition in that market? Think long and hard about this one, as it forms one of the core arguments in US v. IBM. Depending on how you quantify that market, IBM was either a massive monopoly or just another company. First of all, and I think at the core, we're looking at a mainframe market. IBM is specifically making big, powerful machines for big institutions and businesses. Customers are spending at least hundreds of thousands of dollars with IBM. So right off the bat, many computers and other small machines are totally off the table.

While someone like MIT or the Department of Justice might make use of smaller computers, the core of their digital work would get done on a mainframe. These are totally different types of machines. IBM is also operating as a full-service shop. When you buy IBM, you are only buying IBM. You don't need to go to another vendor for anything. There will be some exceptions, especially later into the 60s, but in general, your IBM processor connects to your IBM terminal and spins up an IBM hard drive. That means that a competitor, any company that offers a true substitute for the 360, must also be a full-service vendor. That means hardware, software, and services. Everything from CPUs to disk drives to terminals and all the code that runs on it and all the help setting that up.

I guess that kind of takes us to the beginnings of USVIBM, and this also takes us deep into the realms of confusion. Prior to the federal case, there were a number of smaller private lawsuits. These were much less broad lawsuits, which leads to a weird hiccup in the story. It's easy to say that this pile of small lawsuits tipped off the feds to IBM's tricks, which led directly to US IBM. Therefore, the larger lawsuit can be explained as an extension or combination of these earlier suits. That's true in some ways, but not in every way. The trouble truly starts with the 360. This is something I didn't clock during last episode, but the System 360's design actually opens up a chink in IBM's proprietary armor. It was a

modular computer. One terminal could work on any System 360. You could mix and match peripherals with ease. This meant it was profitable for a company to offer 360 compatible peripherals. A third party could tap into a huge market. That, however, was risky business. Let me direct your eyes to the Telex Corporation and their attempt to move into this new market. Now, there are memos and testimonies that tell the full story, but I'm actually pulling from Telex v. IBM, implications for the businessman and the computer manufacturer from the Virginia Law Review, as well as some parts of Folded, Spindled, and Mutilated. I'm just not going to go back to primary on this part because my mind's already kind of reeling

from all the legal documents. Talix was a company that was already well-established by 1965. When the 360 came out, they decided to start manufacturing and selling third-party peripherals. came out, they decided to start manufacturing and selling third-party peripherals. These third-party companies were called plug-compatible manufacturers, or PCMs. All peripherals connected to a 360's CPU using some big plug, and you just had to make something compatible with that plug, hence the name. Telex entered the PCM market in 1966 with an IBM-compatible tape drive. The Law Review article points out something interesting here. At that point, IBM had a 100% market share on the market for 360 compatible tape drives. They were the only show in town. This is a case where market share doesn't necessarily equate with

monopoly. I think anyone can see that. It would be kind of dumb to call this a monopoly with just this information. So, Telex muscles in and starts selling tape drives. This goes well enough that Telex starts producing other 360 compatible hardware. The crew over at Big Blue, well, 360-compatible hardware. The crew over at Big Blue, well, they can't stand this. So a scheme is hatched, a two-point plan, undercut Telex, and aggressively use lease agreements. The first part might sound fine. Competition is supposed to drive down price, right? So what's the problem here? Well, IBM is the problem. The company is so big that they can afford to take a hit. They're so large that their margins are much higher.

IBM has to spend less money manufacturing a tape drive than Telex does. So Big Blue can do a cool sleight of hand. They drastically cut prices on peripherals while increasing the price on CPUs. Critically here, the CPU market isn't seeing much competition, so IBM can jack that price up to increase profit margins. The peripheral market is where newbies like Telex are. Due to manufacturing costs and scales of economy, Telex just can't afford to lower their prices. But IBM can. Thus, Telex can no longer compete on price. Second is IBM's favorite trick, using a lease as a weapon.

Here, though, the weapon is turned against Telex. IBM introduces these new fixed-term leases in this period. These are a type of contract where a customer can sign on to a one- or two-year lease for IBM equipment. These customers are locked into IBM until the lease ends. That eats up the market, making it even harder for Telex to be competitive. They can't really win over customers if those customers already have IBM tape drives and they're going to have them for the next two years. During trial, documents are uncovered showing IBM's intent. There are memos and studies that lead to these price cuts and new lease contracts.

IBM is, knowingly, trying to keep PCMs out of the market. They're trying to squash Telex. They're trying to maintain PCMs out of the market. They're trying to squash Telex. They're trying to maintain a monopoly. That is illegal under the Sherman Antitrust Act, even though slashing prices and drafting new lease agreements are both totally legal. But here's the funny thing, and it's one of the patterns that I warned you about. IBM loses this case and then appeals. that I warned you about. IBM loses this case and then appeals. That appeal goes to a circuit court which rules for IBM. Telex then appeals to the Supreme Court, at which point IBM did something that I think is easy to see as dirty. IBM claimed that Telex was in violation of IBM's patents

and threatened to countersue. Of course, IBM would back off if Telex was willing to do the same. It should come as no surprise, then, that Telex chose to retract their appeal to avoid a lawsuit with IBM, or rather to avoid being sued by IBM. In this case, IBM is, according to the letter of the law, in the right. They won the appeal, but it's clear to see that they took active steps to keep a monopoly. They applied pressure, the kind of pressure only a huge corporation could apply in order to end the lawsuit. That speaks to something. There are more cases in this period, about 12 in all. Telex is the only case where IBM, for a moment, is found guilty.

All these cases hinge on market definition, but they all use a different definition. In the case of Telex, that market is specifically for plug-compatible hardware. In another case, CDC versus IBM, the market is specifically for mainframes. But be aware that these are all separate cases brought by private parties. When the federal case starts, the DOJ defines the market differently. Now, to be 100% clear, because this does get confusing, during different parts of the 13-year long case, the DOJ is looking at different parts of the market. They're looking at different markets. What I'm talking about here is just the most broad, high-level market. It's the definition that the DOJ uses to show that IBM may have a monopoly. The DOJ states that IBM is in the market

for a combination of mainframe hardware, software, and services. That includes peripherals. It's this huge one-stop market that I defined earlier, and I think that's the only reasonable way to look at this. A consumer wouldn't replace a full room of System 360 hardware with a PDP-1 and a Xerox printer. The only substitute for IBM is another megacorp that works as a one-stop computer shop. According to the Department of Justice, there are only three companies of any size that fit that market. Univac, Honeywell, and Burroughs. Doing the math, that gives IBM a 70% market share. That's enough that IBM could be found in violation of antitrust laws if they were doing something monopolistic. I've been harping on this

whole market thing because whenever you read about US v. IBM, you will undoubtedly run into papers and books written by IBM supporters. Folded, Spindled, and Mutilated is the perfect example of this genre. They argue that US v. IBM is totally unfounded, that it was a waste of taxpayer dollars, and a crime against IBM itself. One of their core pieces of evidence is that market definition. The authors of Folded claim that the DOJ just didn't know jack about computers. They claim that IBM is in a much more broad market. Folded even goes so far as to provide a table of this proposed market. It's huge. It spans everyone from Xerox to Boeing to Citicorp. This is, frankly, an argument that I think misses the point entirely.

And also, frankly, it makes me very mad. Reading all these law papers, it's clear that the DOJ doesn't understand computers. And neither do these economists and lawyers. Let's take Xerox as an example, because it's illustrative of this flaw in logic, I think. It seems cogent at first, but it contains very few brain cells. Xerox, through a company they own called SDS, technically manufactured mainframes. They built and sold mainframes in the same time period as the 360 and beyond. So they have to be in the same market as IBM, right? Well, no, they really don't. SDS mainframes were primarily designed for timesharing. Companies bought an SDS in order to run timesharing software, and specifically,

these machines were marketed towards universities and research labs. And now here's a dirty little secret. The 360 wasn't good at timesharing. There was only one model of 360 that supported virtual memory. That's a feature that you kind of need to do timesharing. Customers didn't buy 360s for timesharing. They bought them as big, lumbering, traditional mainframes. So is a timesharing machine from SDS a substitute for a dedicated mainframe room filled with a 360? No. I think it's very clearly not if you spend a second to think about it. It's not a substitute, so it's a different market. There

may be overlap if you look at very specific models of 360s vs SDSs and you tilt your head the right way, but that's a wasted argument. By playing with the market definition, books like Folded, Spindled, and Mutilated claim that IBM didn't have a majority share at all. In fact, that text calculated that IBM had around 40% market share. So they can't be a monopoly. The numbers just don't work out. IBM is an honest business that doesn't have a single anti-competitive chair on its board of directors. It's also important to be clear about what US v. IBM was actually about. The case was, specifically, that IBM used anti-competitive means to maintain a monopoly. The DOJ measured that monopoly in a very specific market, and they measured it based on the

value of installed computers. That last part is another sticking point for IBM Defenders. The market share that the DOJ calculated was based off the value of all computers actively in use. That put IBM at about 70% market share. Defenders will say that's a bogus number and that the DOJ should actually have looked at sales figures. But to be 100% clear here, IBM made money on sales and leases. Those leases were engineered to keep customers with IBM, thus making a steady source of income. The install base was the best way to capture the market. Marketing competitors, Univac, Honeywell, and Burroughs, also used this mixed lease and sales model.

I really do not comprehend how you can make a good faith argument for defining the market share by new sales alone. That can only be intended as a manipulation of the facts, or just a total lack of understanding of the computer market in this period. The whole argument also sidesteps the fact that the DOJ was, in fact, flexible in their market definition. There are parts of the suit that discuss markets larger than the four-company mainframe sector. There are parts that get as granular as looking at just plug-in compatible computers or just third-party resale and leasing companies. The point I'm getting to is, there was reason for the DOJ to file this lawsuit. There was, if not a monopoly, at least a majority market share held by IBM.

When you run into people saying that, oh well, technically, if you define the market a different way, just don't listen to them. That's not a good faith argument. That's the best way I can put that. All right, I'm going to take some deep breaths here. I hope I've established that IBM had enough of the market that, given the chance, they could commit some antitrust. So, what are the allegations? How is IBM breaking the law to maintain a monopoly? To quote from the feds themselves, quote, the 1969 case alleged that IBM illegally acquired and maintained its monopoly of general-purpose digital computers through exclusionary and predatory conduct going beyond the 1956 decree.

That's specifically from a Department of Justice memo summarizing the case. The large charge here was that IBM had broken antitrust law in order to create and maintain their monopoly, but the majority of the case and the actual charges focus on maintenance of a monopoly because proving the acquisition of monopoly is actually kind of hard. Further, those crimes were not covered by the earlier consent decree. In other words, this was new stuff. The specific period covered by the suit was 61 to 69, so the lead-up to and the heyday of the System 360. So, what do we mean by maintain a monopoly? Simply put, it was what we saw with Telex.

The charge is that IBM made certain business decisions in order to keep new companies from entering the market, or to prevent existing competitors from winning market share. The part that makes this an antitrust violation is that IBM used their huge size to make good on these anti-competitive dreams, and they had the intent to maintain that monopoly. Part of these allegations we've already discussed. IBM made strategic price cuts and changes to lease contracts, both in order to keep smaller companies from competing. This was ruled, at least at one stage, to be illegal. IBM was able to make huge price cuts because they were IBM. They had enough money, thanks to their huge market share, to push through a hit, knowing that more profits would lie down the road.

More specifically, there were allegations that IBM had wielded these price cuts like a scalpel. According to the DOJ, Big Blue had, in some cases, quote, engaged in various below-cost and discriminatory discount conduct in marketing its products to educational and scientific institutions in order to injure peripheral manufacturers In other words, IBM had moved in on specific sectors of the market and done so very aggressively. Since IBM was so big, they could afford to, say, sell hardware below cost to a few universities. to, say, sell hardware below cost to a few universities. Their huge market share made it possible for them to take a loss in one sector while still turning an overall profit. A possible competitor, like Telex, could not do that. They did not have enough capital. The next allegation I want us to consider is the use of so-called fighting machines. That's what the court document actually

calls them. I swear I'm not making up a dumb term. The idea is that IBM announced new machines without the intent to release them anytime soon. Why? Well, simply to thwart competition by making customers think IBM was about to release better hardware. One example of this, according to the DOJ, is the case of the CDC Model 6600. So, let me sketch this out as a short story. In late 1963, CDC, a relatively small mainframe company, announced the 6600. It was, for the time, the fastest computer money could buy. Watson Jr., hereditary lord of IBM, was infuriated by this announcement. How could such a small company ever deign to reach the heights of IBM? And how could IBM let this happen?

Big Blue had to retake their crown. The response from IBM was the Model 90 project, an attempt at an ultra-high performance System 360. This was to be the CDC destroyer. Over the course of development, the 90s initial designs were extended, so much so it was renamed the Model 92. so much so it was renamed the Model 92. According to author Emerson Pugh, in IBM's 360 and early 370 systems, the Model 92 was announced, or at least discussed in public, at the 1964 Fall Joint Computer Conference. But the 92 never shipped, as the design proved to be too ambitious. The design team went back to the

drawing board. They stripped out some of the fat and came back with the Model 91. This was a less powerful machine than the proposed 92, but still made to compete with CDC's beast. The Model 91 was properly announced in 1966. Manufacturing issues led to some delays, so the first 91 didn't actually ship until late 67, with the first consumer model running by 1968. That's the IBM side of the story. The 90 series saw delays and cancellations due to design and manufacturing issues. Clean, simple, and totally legal. According to the DOJ, however, IBM announced the 91 and 92 for the express purpose of hurting CDC. I hope the logic here is clear. By IBM announcing these machines, customers would, in theory, be less likely to

hop on the CDC bandwagon. Sure, CDC had a faster computer, but IBM was about to catch up. Why even consider switching to a new manufacturer? No one ever got fired for buying an IBM after all, and IBM was going to deliver a faster machine next month, or the month after, or the month after that. Once again, this is an intent crime, and we don't have very clear smoking gun evidence of the intent. Announcing products way too early isn't illegal, even if it is kind of scummy. But the DOJ and CDC argued that this was done in service of monopoly. That would make it illegal. And this brings us to the big one. Or, at least for my wild spiraling narrative, this is the big one.

The DOJ argued that IBM's practice of bundling hardware with software and services was anti-competitive. As we discussed earlier, when you bought IBM, you weren't just buying a computer. You were buying a way of life, a whole ecosystem, a whole army of dudes in little blue suits. Your hardware would arrive and be installed by IBM technicians. Your team would be trained by IBM technicians. Your software would all display an IBM trademark as it loaded. But crucially, the only line item on your bill would be for hardware. but crucially, the only line item on your bill would be for hardware. Software and services were free. This made IBM a very special kind of one-stop shop. It was yet another example of how IBM customers had this ongoing relationship with the vendor. It wasn't just a lease or a long-term

contract, it was a continuous back and forth. This ranged from tech support to help with system design to, in some cases, custom software. Everything that wasn't physical, that wasn't made up of wires and semiconductors, came at no upfront cost. This is something that IBM could do because they were big, and they could extract those lost values in other ways. As a bit of an industry ghoul myself, I can certainly see the allure of this model. I can imagine administrators and C-suite folk salivating over the possible savings. Software, usually, costs more than hardware. At least, if you sum everything up. So the prospect of free software, well, that's huge. That's one of the reasons that open source software is so big in the computer industry.

I can't think of a better way to explain it than that. It's just a wild value proposition. This means that, in effect, there was no reason to ever buy software. If you're an IBM shop, then you'd never bother with third-party software. That would just be a waste of money. Anything that IBM doesn't have, they'll probably write for you, or you can write yourself. Put another way, a software company cannot exist in this climate. There is no possible way to compete. Burton Grad, in A Personal Recollection, IBM's unbundling of software and services, points out that bundling helped small clients. It made IBM more attractive to companies that either didn't have a huge amount of money or

didn't have experience with computers. Let's say you run an office that's modernizing, but you don't really know how to do that. You might have some staffers with computer experience, but no experts. You could hire on someone to handle the digitizing effort, or you could just go to IBM. Once you're in Big Blue's embrace, all your problems are solved. They know just what you need, they train you to use it, and they even install everything. So, let me ask a question. So, let me ask a question. Why would you go with anyone but IBM? Why would you buy, say, a CDC mainframe? This leads to an interesting set of circumstances. Now, I'm speculating here, so watch out. Some other manufacturers bundled as well. Notably, companies like Honeywell were bundling hardware, software,

and services well into the 1970s. But remember, all other full-service mainframe shops were tiny compared to IBM. They didn't have the same kind of power to shape the market. So in effect, IBM was forcing other companies to bundle. In order to compete in any meaningful way, software and services had to be free. This, the DOJ argues, was another way that IBM created and maintained their monopoly. Small companies were simply priced out of the market, since they couldn't afford to bundle. Software is expensive, after all. But IBM, with its huge size and market share, could eat that upfront expense. The benefit was that, down the road, Big Blue would only grow bigger. A further complication is that IBM didn't offer a standard bundle to each customer.

Instead, the bundle could vary from customer to customer. It was a very bespoke kind of thing. And while this may look nice on the surface, it has an insidious side to it. You could say that IBM was meeting client needs, or you could say that IBM was changing their prices for each client. By changing the bundle, the actual value of a lease or sale could be tweaked until the client would sign. This is also where we reach a slippery move on IBM's part. The so-called unbundling.

As far back as 1967, IBM was looking at unbundling software. That is, they were discussing actually selling software as its own product. This wasn't a new practice. There were some companies that sold software as early as 1955, but there wasn't yet a fully differentiated software market. It had been stifled, in part, by IBM's own actions. The possibility of a new revenue stream must have been enticing. own actions. The possibility of a new revenue stream must have been enticing. Grad explained in his affirmation paper that these studies were inconclusive. IBM's top brass just wasn't super interested in unbundling. At least, not at first. It took a new development to change their minds. During 1968, IBM was well aware that the DOJ was pursuing the possibility of a suit against it,

alleging monopoly in the computer market. IBM assessed its position during the fall of 1968 and concluded that it probably could not successfully defend itself against an antitrust charge stemming from tie-in sales because of its bundling practices. Grad continues, IBM executives, including Thomas J. Watson Jr., Vincent Learson, and particularly Burke Marshall, held discussions with senior DOJ attorneys and believed that IBM could preempt a DOJ suit by announcing it would unbundle its service, then doing so promptly. IBM unbundled software and services specifically in an attempt to avoid the DOJ's wrath.

They knew it was anti-competitive. They knew this was a violation of antitrust law, so they got rid of the practice. This wasn't an attempt to get right for the right reasons, it was purely motivated by profit. As Grad further elaborates, a successful antitrust suit would be devastating. It could result in IBM being broken up into smaller corporate entities. It could cascade into dozens of costly lawsuits with competitors. It could destroy Big Blue. This is a side of IBM we don't usually see.

Not dominant, not in control, but making wide-reaching decisions to try and protect itself. This event, the unbundling, would have profound impacts on the digital world. This created the third-party software market. IBM software now cost money, so other companies could conceivably sell software. That's the first layer here, but the unbundling is like an onion. As you peel away one layer, you see yet another. Let's think all the way back to last episode and Gene Amdahl, the primary architect of the System 360. In 1970, he left the IBM fold and became a competitor. Amdahl planned to make computers that were compatible with the System 370, the successor to the 360. Now, the funny thing here is that the 370s were actually backwards compatible, but

I digress. A core part of the business plan was that Amdahl's new computer would be able to run IBM software. That would be crucial to getting off the ground. The unbundling and the swirling antitrust case had a huge impact on Amdahl's decision to leave. These circumstances actually made his new company possible. To quote from an oral history taken at the Computer History Museum in 2000, Just one year earlier, well, in July of 1969, IBM announced internally that it was putting its old software in the public domain and was going to price the new software separately, and that immediately told me that if we made a compatible machine, Once again, this is showing the secret underbelly of IBM. The Apex predator of the business world has to tread carefully under the glare of the Department

of Justice. The question is, how careful were IBM's footsteps? All right, thus we come to the end of today's story. So how does USV-IBM actually end? Well, in 1982, the case is dropped by the Department of Justice, ending a 13-year-long lawsuit. This leaves us in some weird territory, but it's not without precedent. Officially, the DOJ decided the case had no merit. It was a waste of time and treasure. It should have never sued IBM. But the fact of the matter is, IBM dodged a bullet. It's clear that IBM was engaged in at least some anti-competitive behavior.

The unbundling is proof of that behavior, and of the fact that IBM knew what they were doing was violating antitrust laws. Grad, who was at IBM for the unbundling, says as much in his recollections. For me, that's the bottom line. You can argue all you want about the economics of the matter, about how markets are defined and what actually constitutes a monopoly on paper. At the end of the day, IBM had established a pattern of surviving litigation. I think their earlier antitrust cases gave them some practice. They skirted a line that other companies just could not.

That seems like something only a monopolist would do. So, was US v. IBM unfounded? Was it without merit? No. There were legitimate concerns about IBM's affairs. They were doing some shady things, and they made adjustments to their business to try and avoid the lawsuit. That said, I think the case was mishandled. It became this huge circus. I think the DOJ could have won if they had a more focused case. But then again, it's hard to have a focused case against a titan of industry like IBM. Where does this leave us? Well, we get this time period where IBM is being cautious.

We get this new software market. We get self-imposed restrictions that maybe lead to some strange circumstances. Next episode, we're looking at the possible result of those circumstances. The System360 clone market. Thanks for listening to Adren of Computing. I'll be back in two weeks time with the conclusion of my System 360 series. In the meantime, if you like the show, there are a few ways you can support it. If you know someone else who's interested in the history of computing, then please take a minute to share the podcast with them. You can also rate and review the show

on Apple Podcasts. If you want to be a super fan, then you can support the show through Advent of Computing merch or signing up as a patron on Patreon. Patrons get early access to episodes, polls for the direction of the show, and bonus content. You can find links to everything on my website, adventofcomputing.com. If you have any comments or suggestions for a future episode, then you can shoot me a tweet at Advent of Comp on Twitter or just email me. And as always, have a great rest of your day.