IBM PC: A conversation with Dr. David Bradley
An interview with Dr. David Bradley, conducted in August/September 2021 by Henner Thomsen of Stay Forever. Stay Forever is Germany’s leading retro gaming podcast.
Dr. David Bradley, born in 1949, is one of the very first engineers who worked on the IBM Model 5150 – better known as the IBM Personal Computer, released in 1981. He joined the company in 1975 and became a part of the IBM System/23 DataMaster development team in 1978. From there he was recruited for the PC team in September 1980, where he created the machine’s BIOS code. He moved on to further engineering and management roles in the development of the PC XT as well as the PS/2 and PowerPC-based computer series.
Henner Thomsen, Stay Forever (SF): Dr. Dave, I’m really grateful that you agreed to answer a few of my questions, which are about all aspects of the PC’s development – and all the anecdotes you might want to share with us.
Dr. David Bradley (Dr. Dave): Before I start, 2 important statements:
1) I was NOT an executive or manager or strategist or evangelist for the IBM Personal Computer. I was a mid-high level engineer who was good at writing device driver code. Most of what I know about the history of the IBM PC before September 1, 1980 is hearsay and „campfire“ stories. I wasn’t there for many of the major decisions. So take anything I say with a grain of salt.
2) The answer to virtually any technical question about the IBM PC can be answered with „DataMaster“ (DM). We began working on S/23 DataMaster in early 1978. It used an Intel 8085. We wrote the BASIC interpreter and operating system (OS) for it. It was announced in July 1981, 2 weeks before the IBM PC. All of the planners, marketers, engineers, programmers and manufacturers who founded the PC development came from DataMaster, with the imperative to develop and announce a product within a year.
SF: Well, of course I was aware of the DataMaster, but always saw it as a close sibling to the 5120 and its predecessors, even though the 5120 used the PALM instead of the 8085. But now I realize the DataMaster design – despite its vastly different appearance – is actually closer to the 5150, or at least paved the way for it.
Dr. Dave: The DataMaster was not a younger sibling to the 5120, even though it may seem that way. When the DM development had been delayed beyond the plan, Bill Sydnes took the DataMaster mechanicals and the 5100 electricals and put them together in about 6 months to bring out the 5120. DM is more of a „mother“ to the 5120, with the 5100 being the „father“. Or something like that.
SF: Interesting. I had no idea about that 5120’s origin story.
Dr. Dave: There are several IBM PC stories about products that came about because of schedule delays or extra inventory. The XT 286 was a one off, based on the XT, when we had lots of 6MHz 286s in inventory. Quick redesign of the motherboard (take a little AT, mix with a little XT) and voilà, a new product.
SF: I heard that Atari offered IBM to design a personal computer for them back in 1979, and there are reports of an actual „IBM Atari 800 PC“ prototype being reviewed. There’s even a picture floating around, with a case created by Tom Hardy, the IBM PC’s designer. Is there any truth to that?
Dr. Dave: See my #1 above. Such an activity may have happened at the corporate level, but I have no knowledge of it.
An important thing to know is that there were several personal computer activities taking place in IBM in the late 70s. The DataMaster was one of them, which led to a commercial product and the foundation for the PC. But IBM was not monolithically ruled from above – lots of system groups were looking at doing something, and it’s quite possible that one of them talked to Atari. Could have even been a corporate staff group. But that had no effect on our work in Boca.
As an addendum to the many PC activities underway, the story is often told of Gary Kildall missing the important meeting with IBM on getting CP/M for the PC. While that part of the story is true, I have heard from people who were on the software side, that Jack Sams was the fourth IBM group to visit CP/M, all of whom showed up with non-disclosure agreements (NDAs) and were looking to learn a lot and offer little in return. I think that he figured Jack was just the most recent in a parade of IBMers.
SF: Both Bill Gates and Paul Allen claimed to have convinced IBM to make the PC a 16 bit machine – according to them, IBM had originally planned an 8 bit computer. Is that correct?
Dr. Dave: See my #1 above. [But by August 1980] it was clear the CPU was the 8088. There had been an IBM task force meeting in August 1980 to define the PC and determine if it was feasible. Lew Eggebrecht was the hardware guy there and did the design you can see in the charts. I believe that Lew picked the 8088 because in DataMaster (I told you) we had difficult workarounds for the 64KB address space that Lew was very familiar with. The 8086/8088 had a 20-bit address and solved the problem.
The 8086/8088 was the next evolution, but was more than a year old, so it wasn’t „bleeding edge“ technology. I wish I could say something about the relative pricing between an 8088 and an 8085 (and associated chip sets), but I don’t have any. I’m assuming it wasn’t a big difference. The 8088 was a reasonable choice.
SF: The prototype name, „Acorn“ … I don’t know if you were involved in naming the machine, but somebody from the PC team must have known about that British computer company also called Acorn, which had been in existence since 1978?
Dr. Dave: Code names appeared randomly, using techniques I had no view of. The PC was „Chess“ for a long time, and changed to „Acorn“ later on (or perhaps the other way around, I really can’t remember). I do have a T-shirt that says „I survived Acorn“ that a technician had made shortly after announcement, so I’m thinking „Acorn“ was the later name.
But code names were internal only, since the project needed to have a name so we could talk about it. There was no knowledge of the final „real“ name for most of us until the announcement.
Code names were frequently different for different suppliers and partners. While we might refer to the product as „Acorn“ internally, we would use another, unrelated name, like „Salmon“, to refer to the product when in discussions with Computerland, or „Panda“ when talking to Digital Research. There was no permanence or meaning to any of the code names („Salmon“ and „Panda“ were just totally made up by me just now for illustrative purposes. I don’t remember, if I ever knew, what other code names were used for the PC.)
SF: According to some sources, your „Dirty Dozen“ team of 12 engineers was sent to a rented building without the IBM company logo, outside the IBM campus, because
A) the project was so top secret, or
B) nobody wanted to have anything to do with it,
depending on who tells the story. Could you elaborate on that?
Dr. Dave: We never called ourselves the „Dirty Dozen“. It first appeared, IIRC, in an article in Think Magazine (an inhouse, but distributed externally to customers, etc.) article on the first anniversary of the IBM PC. Also 12 ENGINEERS met during September 1980 but quickly grew to about 50 engineers by the end of the year. And that doesn’t count the programmers (mostly working with Microsoft (MS) and others), planners, marketers, manufacturing and other groups that started in September, and grew to about 400 by midsummer 1981.
I was one of those 12 engineers working on the PC in September 1980 (I was the 12th). There are about 60 people who claim to be part of the original 12.
About A): Yes, it was very secret. IBM was always secretive about their products under development. The PC may have been more secret than most, maybe because it was a brand-new market for IBM. The DataMaster was also a secret development. If I remember correctly, there was a sign at the front of our workplace that said, „IBM Building 203“. It wasn’t hidden. It had workers from another company in one section – a bank’s IT department. We never saw them since that section of the building was entirely walled off from ours.
203 was also the very first building that IBM built when they moved to Boca Raton in 1969 (or had built and leased). It had always been an IBM building, but a leased one. And, since the original engineers came from the DataMaster development, we began working in Building 203, which was where we were working on DM. I didn’t even change offices. I did go into a different lab for PC work. Lew had gotten an Intel Microprocessor Development System (MDS) with a disk (!), rather than diskettes, and it lived in the PC lab. We had used Intel MDS for the DM work (I told you) and just continued on.
The other PC groups (software, planning, …) worked in other buildings. I interacted with the software people a lot and they were in a small building near the main site (on a golf course!). I think they were there because there was no more room on the main site. About April 1981, all of the PC development group, except for manufacturing, moved into a brand-new, leased, building about 3 miles from the main site. Here it was clearly a case of needed a new place to bring everybody together because there wasn’t room on the main site.
To summarize A): We were in a leased building because (1) DataMaster (I told you) and (2) no room anywhere else. We weren’t hiding, but we didn’t talk about what we were working on either.
About B): I don’t recall anyone not wanting to be there for the PC work. Some people may have passed on it, but I didn’t hear about them. We were excited. Mel Hallerman (known by some as the Chief Programmer for the IBM PC) heard about the PC development work while working on the Series/1 PL/1 compiler. He just left and started working for Sandy Meade, software manager. Paperwork to follow. (But Mel was a true believer. He got the „IBMPC“ vanity license plate for his car after the announcement.)
SF: Any anecdotes from this time that you want to share?
A favorite is that the IBM PC had to conform to FCC Class B protocols, intended for home products. The electromagnetic emissions had to be much lower than the Class A required of business machines. Since we didn’t have a Faraday cage room available for electromagnetic compatibility (EMC) testing, we used an approved alternative – an open field removed from any sources of electronic contamination. Since the PC was developed in Boca Raton, Florida, if you head for the nearest open field meeting requirements, you’re on the outskirts of the Florida Everglades. In those days everyone at IBM dressed like banker or funeral director – white shirt, tie, suit coat and dress pants. The guys who went out and did the EMC testing dressed in T-shirts and shorts, and came back at the end of the day sunburnt and sweaty. And since they were working on the edge of the mosquito laden swamp, debugging had a very different meaning for them.
SF: Besides the Apple II’s modularity and the ROM-based BASIC from several machines of that era, is there anything else you took inspiration from (or were asked to duplicate – or to avoid)?
Dr. Dave: See my #2 above – most of the inspiration was from the DataMaster. Perhaps the mechanical design – separate keyboard, printer, system and display – was inspired by personal computers of the time, both Apple and Radio Shack were like that, as well as most S100 systems.
The PC development group (I’m not sure who) acquired about 10 Apple II. I played around with one for a few days. It was nice but didn’t inspire me to do anything in particular. Also, there was never any guidance to avoid things. For me, I was guided by an appendix of the boilerplate Microsoft contract with OEMs. They required a certain amount of device drivers to be delivered with the system – keyboard, printer, display, diskette, … which we put in the BIOS ROM. My goal was to work to their required interface. However, the boilerplate was 8080 based so there were discussions with them on moving to the 8086 software interrupts, how to pass parameters and such. As we progressed there were additions and extensions made, most notably character operations while in graphics mode which I patented and made lots of money for IBM during the clone era. And IBM shared some of that money with me, so I was happy.
SF: Was the BIOS/OS structure inspired by Gary Kildall’s CP/M with its multi-layered architecture (BIOS/BDOS/CCP)?
Dr. Dave: Easy answer, see #2 above. No, it all came from DataMaster and Microsoft’s requirements. DM had a multi-layered architecture (and wasn’t influenced by CP/M) and MS required an OEM supplied layer of device control. From all of this I surmise that a multi-layer architecture is the natural way of assembling a large system like this.
SF: Did anyone consider a GUI and mouse for the PC, since the Xerox Alto had already proven their potential in 1972?
Dr. Dave: No. Reasons, all in hindsight:
1) [We had to] develop and announce in one year,
2) use commercial off-the-shelf (COTS) operating system (eventually MS), none of which offered GUI,
3) design a competitive personal computer, none of which used a mouse in 1980.
Closest thing in IBM at the time was David Morrill’s GLASS being done at Watson Research, and it wasn’t ready for prime time until 1985 (and maybe not even then).
SF: The CGA color palettes … I guess you get a lot of questions about the interesting choice of cyan and magenta, but well, there’s not that much you can do with only four colors if two of them are black and white (and other palettes were available of course). But couldn’t IBM have offered a better color video adapter? Of course cost reduction was one of the main goals, but in 1981 you could buy an Atari 400 with GTIA chip and a 256 color palette for about the same price as the CGA card (albeit with lower resolutions).
Dr. Dave: I’m going with one year development as the excuse. No custom chips. And 16KB on the CGA meant that 320×200 was limited to 2b/pixel. Somebody, not me, picked the actual colors, and I think that Bill Gates claimed to be in on the choice. Could have been Andy Saenz, who did the CGA design.
VGA was the first real attempt to do something bigger and better. But it took time to get the custom chip.
And I do have to point out the superb characters on the Monochrome Display Adapter. No colors, but great for business. And it was International BUSINESS Machines.
SF: What computer did you use personally?
Dr. Dave: The system that I purchased – at least the electronics – was a Digital Group. Z80 based. I put it together during Super Bowl XII (Jan 1978 for non-Americans). I used an old TV set as the monitor, built a keyboard from a kit from Radio Shack, and played around with it for a year or so. However, my day job (in Feb 1978 on) became DataMaster, which was 8085 system design and software. After doing that for 10 hours a day, doing the same thing with my personal computer didn’t seem so interesting. This is a picture of my Digital Group (covers on, unfortunately) with my Radio Shack keyboard in front.
Which leads to another anecdote:
I built the keyboard from a Radio Shack kit sometime in 1978. It worked well, but it produced only capital letters. Since I didn’t want my Tweets to sound like I was shouting, I wanted to be able to do lowercase chars also. So I examined the schematics and determined that if I cut a few wires, added another transistor-transistor logic (TTL) component (the one in the upper left oriented differently than all the others) and wired in a few gates I could also get both upper and lower case characters. That’s the value of 8 years of an electrical engineering education!
SF: Some sources claim that Intel’s 8088 was selected because the 8086 was „too fast“, and that a faster microcomputer would hurt sales of more expensive machines. Is that true?
Dr. Dave: No. 8088 was chosen for the 8-bit bus, which was slower than the 8086, but led to a cheaper entry system price. The 8086 would have required memory x16 rather than x8, so minimum entry point would have been 32KB. Also would have required 6 rather than 5 8KB ROM modules on system board (would have saved me time and effort in getting BIOS to fit in 8K). We wanted to have a low advertisable entry price (if I remember correctly, $1565 with 16K memory, you supply monitor and tape cassette).
Compared to DM (see #2 above) the PC would be significantly faster, primarily due to 1MB addressing rather than kluged 64KB extension, and higher clock speed. Also it would be competitive with personal computers offered at the time – Apple II, TRS-80 etc. We never benchmarked against any IBM products (I guess S/34, S/36 would be most vulnerable, but they offered much more in commercial software and stuff for a medium sized business). In fact, what little benchmarking we did was done by me, and I did a poor job of it. Of course, I spent about 3 days on it, so effort equaled results.
SF: Did you expect DOS-based clone machines to be available so soon – or weren’t you expecting them at all, since your BIOS should act as a „copy protection“ mechanism?
Dr. Dave: Poor planning on our part. Based on the S100 experience we should have expected clones very quickly. But that would have required us to assume that the PC would be a world changing product and we were just hoping for a successful one. So we (or at least me) weren’t expecting them at all.
The BIOS as a protection mechanism was an after the fact method. And easy to work around, as Compaq and Phoenix and others quickly showed. So we could get the low hanging fruit and force the cloners to do a little bit of work.
Patents were a MUCH stronger protection mechanism against cloners. There were about 11 patents that protected much of the IBM PC (including my 2 mentioned above) and required licensing from IBM. The cloners were the driving force for IBM changing their patent policy in the mid-80s. Prior to that IBM licensed patents pretty cheaply, primarily working with other companies and cross licensing their patents. Since cloners seldom had any intellectual property (IP) we needed to charge them more, and IBM started doing that about the same time the PS/2 and Microchannel (MC) appeared – leading to claims about IBM forcing MC cloners to pay big bucks. Nobody ever paid more for an MC machine rather than an ISA machine, but we didn’t do a good job of saying that. (See #1 above. I was not party to corporate IBM decisions on patent licensing. These are MY conclusions based on what I saw.)
Finally, in hindsight, I’m not sure what we could have done differently, we used Intel and MS as and others as COTS suppliers to get it done in a year. We published the technical reference manual (TRM) with all the „secrets“ inside since we didn’t have time for a reasoned architecture manual, and we wanted to invite the rest of the industry to join us with both hardware and software supporting the PC. (And the TRM probably saved only 2 months or so effort for someone cloning a PC.)
SF: So you’re saying IBM sold licenses for some of its patents to clone manufacturers – even before ISA or even PS2/MCA were a thing? I didn’t know that, doesn’t this mean that IBM allowed the clones to exist in the first place? IBM couldn’t be forced to license patents to competing companies, if I’m not mistaken, so why did they do it?
Dr. Dave: In the 1980s IBM licensed its patents to just about everybody (at least in the US). I’m not a lawyer (I guess this becomes #3 in my disclaimer list), but I did spend a lot of time with them during the Clone Wars. So here’s my explanation to answer your question.
IBM was much bigger than the „PC Company“. In fact, for the first couple of years we were just a pimple on the revenue sheet. IBM’s big business was S/360 then S/370 and so on, the big mainframes. There had been a decade long antitrust suit against IBM’s practices in the plug-compatible manufacturer (PCM) market. I think that one of the ways out of that suit was patent licensing. So the IBM company licensed its patents, and this meant the PC Corporation would license its patents.
The change in the mid-late 80s was to charge more for patent licenses, driven largely by the fact that many PC licensees were out and out clone makers and did NO development work themselves. In the mainframe PCM market, the PCMs were real developers. By increasing the license fees IBM forced PC cloners to spend development dollars to either a) make license fees make up for the DE they didn’t spend, or b) to do some real development, get some patents, and cross license with IBM as we did with bigger companies such as Compaq. Also, by increasing the fees IBM developed another revenue stream, almost pure profit (except for salaries to those pesky lawyers). In the 90s and beyond, it was substantial.
The PC Corporation also „suffered“ because of IBM as a whole. In hindsight it would have made sense for IBM to license the BIOS and PC architecture to other companies so as to maintain a presence along with Intel and Microsoft. But S/360 had drawn the line at licensing their basic OS/360, so no licensing the BIOS. I don’t think it would have significantly affected the final result of the PC Corporation, but it might have made a difference.
SF: The story of the clones is told very differently depending on who is telling it. Some say IBM actually encouraged the clone makers, so the PC platform might spread and ultimately defeat Apple. Others say IBM fought the clone makers with all they had, but just couldn’t stop them.
So I would like to clarify, was IBM under any legal obligation to license its PC patents because of the antitrust suit?
Dr. Dave: I think that we naively thought we could encourage people to develop products to use with the PC, hardware and software. I personally never thought of clones, although I should have – it happened with S/100 after all. I thought we’d defeat Apple with a superior product, faster, cheaper, more flexibility, etc. Or at least hoped …
When the clones arrived (slowly at first – the first I recall was a TI product, probably already in development when the PC came out, that was similar but not compatible) I was thinking, „Imitation is the sincerest form of flattery.“ When Columbia and Eagle and others first appeared, we were appalled because it was a wire for wire, line for line, copy. Where’s the engineering in that? (Clearly, I was naïve.) So I think we fought them because they were using our work product, our IP, to compete against us. We had to force them to actually do work to bring out a product.
I’m not a lawyer, although I worked with many. I don’t know the provisions of the antitrust settlement. I do know that in the early ’80s it was IBM’s practice to license patents fairly freely, not demanding much money for them. Things that IBM REALLY cared about were protected by trade secret (the IBM SELECTRIC ball was that way). The need to level the development playing field drove the changes to IBM’s licensing practice in the later ’80s, I think (see #1, way, way above).
SF: Either way, when companies like Compaq or Columbia asked for licenses, IBM already knew what they were building, right?
Dr. Dave: I can’t recall (but then again, I didn’t work in the IP shop) anybody asking for a patent license BEFORE they had a product. We knew what they were doing. The biggest difference was whether they had their own IP or not. Compaq had a (small) stack of patents, some of which we might want to use, so we compared it to our stack and negotiated a payment for the cross license. When a Columbia showed up with nothing, we said this is what it’ll cost you, with the amount pretty much aimed at forcing them to spend DE $ for licensing at nearly the same rate as if they had designed the machine without reference to ours (patents would still apply, no matter how you developed). But that’s my assumption from the years in the clone wars. I was never in the room when the strategic decisions like this were made for the IBM Corporation.
Although not really part of this question, here’s a chart that summarizes the PC Corporation difficulty with being part of IBM. This is done in the tradition of a McKinsey consultant 4 quadrant chart that breaks up the world into nice regions. I did this originally when working on the PC Corporation staff in late ’80s or so. You could add more detail and nuance, but you get the drift. PC Corporation and IBM had two very different views of the world.
SF: You keep referencing the „PC Corporation“ as if it was its own entity – are you talking about the Entry Systems Division (which was created in 1983 if I’m not mistaken) or a similar IBM branch/subsidiary? Or was it more of a nick name for the PC development team in Boca?
Dr. Dave: For me „PC Corp.“ is shorthand for whatever group was doing the PC. If I remember correctly, it started in Entry Level Small Systems, then became Entry System Division and then sometime became a business unit known as the PC Company. I wasn’t making a business statement with the name – just a way of referring to the people making the PCs.
SF: Why did IBM decide not to offer a hard disk drive (HDD) controller? An HDD seems like an obvious choice for an office machine – unlike the included cassette tape controller. I know PC-DOS 1.0 didn’t support hard drives but I guess this could have been implemented.
Dr. Dave: Time and availability. One year to develop. COTS stuff. Hard disks may have been available, but they weren’t plentiful and easy to pick up. The XT really focused on an HDD, and it took a year to get a controller and drive that we could put the IBM name on. Remember that IBM still maintained high standards of quality for their products, even PCs. Remember parity?
SF: Since IBM offered a gameport card, you obviously expected the machine to be used for gaming. This seems like the most non-IBM thing to do. :) Did you ever consider or discuss the 5150’s gaming capabilities during development? Did you play any games on it, and how did management think about selling „toys“?
Dr. Dave: Not really focused on gaming during development. I did „beta“ testing on Adventure – which means that I played the game on the PC and told the product manager about it. I was an early user of Decathlon for the PC – I helped the developer, Tim Smith, with some hardware issues, since the game relied on pounding the keyboard pretty actively. I was the only one in the engineering group who could pole vault – not well, but at all. Snipes became a game played by many – particularly since it was character based and would work on the monochrome display.
During the downtime between PC development and the beginning of XT, and PCjr and AT, I wrote a „tank commander“ game that was multiplayer. It was character based, you drove around with an overhead view of the field and could shoot at your opponent. It took two PCs that were connected, if I remember correctly, through the parallel ports (might have been serial ports) to exchange position and field of play information. It worked. Not really well.
And soon it was time to move on to the XT development where I was a first-time manager and a group of people writing XT BIOS and POST working for me. And some architecture work – how do we expand the PC? How does the HDD fit into the BIOS? The rest of IBM showed up, so how do we add capability for disadvantaged people – perhaps can type with only one finger – or a tongue stick? Did lots of that stuff. But no more game playing.
Side note, Mark Dean (well-known member of the PC AT development team) and his officemate, a technician named Ted, were relatively big-time game players. Missile Command type of things, Snipes, Frogger, things like that in ’83 and ’84. (Not to suggest that was a big-time suck for them, but when they had downtime, they rolled out the games.)
But gaming was never a big deal for the PC developers. Getting the thing to work was.
SF: Sir Clive Sinclair (Sinclair Research) once said, „On the outside the IBM PC may look elegant but inside it is board after board after board of chips. […] They have a board about the size of a large coffee-table book with God knows how many chips on it, it must be a hundred, just to do color. We do it on one chip.“
Comparing the 5150 to the tiny ZX Spectrum isn’t really fair, but what do you think – were the PC’s PCBs inefficiently designed? Could it have been simpler and cheaper with more development time?
Dr. Dave: Steve Jobs said pretty much the same thing comparing Macintosh to PC (I think that’s when he did it – certainly the Apple II looked like a PC inside), calling it „jellybean TTL“. Here again – one year development cycle. The large-scale integration (LSI) chips came from Intel, Motorola, NEC, and others. We just glued it together with TTL. The IBM PC didn’t really get into the custom chip business until VGA times.
As for cheaper, hard to tell. Given the stone axe level of tools available to us back then, it would have taken a long time to actually get to a working chip.
SF: What was your main goal while designing the XT? Were there any things you regretted regarding the 5150’s design and tried to improve upon with later machines?
Dr. Dave: Big 3 for the XT: More slots, more memory, and HDD. Users were unhappy with 5 slots when you used up 1 for display, 1 for diskette, and (maybe) 1 for printer/serial. 64KB wasn’t enough memory, that’s another slot. Our PC sales team brought customers into the lab to show them what we were working on and to help them plan their future purchases. I was frequently invited to these sessions (the sales guys always invited me right at lunch time, so it was a free lunch), and I would ask the customers about their requirements, including the number of slots. The right number of slots was „1 empty slot“. If all the slots were full, they were uncomfortable.
As for things we regret on the PC, when a journalist asks (ahem) I always say, „We should have made the interrupts active low rather than edge triggered high“. This usually confuses them, and we shift to a different topic. (I think they want me to say something about clones.)
The truth is we couldn’t change much of anything. COMPATIBILITY!!! When we tried the XT on current programs, we found some that failed because they assumed that BIOS routines were located AT SPECIFIC ADDRESSES rather than using the software interrupt vectors that pointed at them. Timing loops couldn’t run faster/slower (this obviously failed when we got to the AT). Bus and slots had to remain electrically identical. We even had complaints when we closed up a 0.5″ by 1″ hole in the back panel (I don’t recall why it was there initially) because a vendor used that hole to run out the connector to his light pen (which you could do with the video cards). Most of the XT effort was gently expanding the capabilities without upsetting anything.
A little anecdote: After we discovered the address requirement for the XT BIOS, we shuffled the stuff around, so the entry points remained the same – in some cases just a jump instruction to where the routine now lived. Wayne Freeman (who wrote the power-on self-test (POST) on the original PC) commented that jump instruction „Unnatural act for compatibility“ which of course was published in the XT TRM. There was some fallout from that.
Bigger steps away from compatibility occurred with the PCjr and AT, but all were remarkably identical to software. Those are the systems where we „improved“ on the PC. PCjr made it cheaper, AT made it faster.
SF: About the famous IBM songbook … I found the 1937 edition online, it’s fascinating and surreal. Was the songbook still being updated and used in the eighties, did you ever sing in a meeting, and if so, what was your favorite song? I would fully expect the entire original PC team to get their own songs!
Dr. Dave: I was aware of the IBM songbook, particularly „Ever Onward IBM“, but never sang songs or did anything with it. It was primarily a sales thingie, and the IBM 100% Club (salesmen that made their quota) would meet annually to pump themselves up and sing the songs (fitted in around the drinking and …).
SF: Do you still own an original IBM 5150?
Dr. Dave: I have an original PC sitting here in my office. I carry the keyboard with me when I present, normally to college students, the history of the PC. I want them to feel just how heavy it is – as well as the clicky keys. I generally hand it to the front row and say „pass it around“. It’s fun to watch a new person accept it and mouth „Oh my god“ when they realize how heavy it is.
I don’t think my PC will run. I don’t have a monitor for it. I don’t think I have any 5.25″ diskettes anymore. I also have a separate system board so I can point to „my“ part of the PC, the BIOS chip.
SF: What aspect of the 5150 were you especially proud of?
Dr. Dave: That it became such a useful product to so many people. I can claim that the PC enabled the Internet because without all of those end user terminals the internet wouldn’t have taken off. The PC moved computing to people in their homes, offices, dorm rooms (and South Sea Islands, according to one of the Charlie Chaplin commercials). And we invited the industry to join with us and write programs and build adapter cards and attachments and other things … and they did. On the first anniversary of the PC, I went to a trade show in Atlantic City, New Jersey (one week before the Miss America pageant, and some of them were walking around, always guarded by several people). A large ballroom was full of people hawking products to work with the PC – programs, network cards, improved video cards, DOS extensions, games. David Bunnell published PC Magazine which would become a twice monthly publication about the size of the Manhattan phone book (there’s a reference that will be lost on the Millennials).
Working on the PC was a great opportunity. It was a great product to work on because
1) It was a brand-new design – it didn’t have to be compatible with anything; even though it did take a lot from DataMaster, that was choice and not requirement.
2) It was a product you could show to your family – „That’s what I did!“
3) It was a great success.
Something that we did must have been really OK. That’s what I’m proud of.
SF: Well, this was amazing. Thank you so much for your time and insights, Dr. Dave!