Follow-up Comment #5, bug #68708 (group groff): Aha! I have it! M. Kirk McKusick is, if not quite exonerated, relieved of central responsibility.
Mike Karels is the culprit; perhaps Charles Hannum learned at his feet. > Markup languages were beginning to appear and be debated at that time and > apple had a hypertext attempt. The markup stuff looked to be a disaster, > highly dependent on each person who input the markup. This led me to develop > an abbreviated type of markup by determining the context rather than the > formatting. CSRG tolerated my stuff banging on their systems. I wanted to > fully implement everything in groff as it would have been easier and faster > than troff, but Mike Karels did not want that. He wanted it ditroff > compatible. --[https://manpages.bsd.lv/history/livingston_20_03_2026.txt Cynthia Livingston, 2026] So, yes, a BSDI principal--a founder, even--was responsible for this decision, as I suspected. (Not like it was a hard guess given the overlap between CSRG bigwigs and BSDI.) With that confirmation, I have a new hypothesis. First let's establish some facts. Clark's release of _groff_ did BSD a **huge** favor because BSD was an incomplete system without it.[1] AT&T source code, even for old (1979) Seventh Edition Unix and its ancient _troff_ ("Ossanna"), which could format man pages to the terminal but not drive any sort of device other than the obsolete-for-a-decade C/A/T typesetter, was not redistributable by the CSRG. The only alternative _troff_ was AT&T's Documenter's Workbench product, "DWB"...or one of its licensees, like those from Elan or SoftQuad, all of which were proprietary and expensive. With _groff_ BSD (and therefore BSDI) could and did ship a freely licensed _troff_ formatter that was far more capable than Ossanna troff, a drop-in replacement for the DWB 2.0 troff-based troff on SunOS 4 (a hugely successful Unix) and on par with SoftQuad troff. Bam. Several problems solved. Dust off one's hands, proclaim _groff_ the new standard in _troff_ technology, and move on, right? Wrong, according to Mike Karels, it seems. Karels's insistence on crippling Cynthia Livingston's work to maintain compatibility with a more limited software platform that was wholly owned not only by a competitor but by one that [https://en.wikipedia.org/wiki/UNIX_System_Laboratories,_Inc._v._Berkeley_Software_Design,_Inc. by 1992 was attempting to sue one's own firm into oblivion] makes **zero** business sense on top of no free-software advocacy sense. As we'll see, that lawsuit was in no sense a surprise.[1] There is something we're not being told. The darkest possibility is that Karels was secretly promised a windfall by USL if he made meaningful decisions to weaken BSDI. I have **no** evidence of this--I venture it simply because it would be an explanation _consistent_ with an engineering decision that was thoroughly unjustifiable on technical grounds, strategic grounds, and one's own putative immediate pecuniary interest. Another possibility is more consistent with the available facts and public history. While on its face it is "philosophical", I think that posture masks speculation, in the formal economic sense. Let's next review what were, at the time, current events. [https://www.latimes.com/archives/la-xpm-1988-01-07-fi-33970-story.html In 1988, AT&T had purchased a 20% stake in Sun Microsystems, in exchange for which the BSD-ish SunOS product would be murdered]. AT&T/USL had penetrated standards committees like X/Open. For example, by 1994 it had managed to make the wholly proprietary SVRr4 _curses_ library the basis for X/Open Curses Issue 4. Moreover, AT&T explicitly positioned System V Release 4 (announced that year but not available until 1989) as a monopoly play, "unifying" the BSD and older System V camps. James Clark began development of _groff_ in 1989 with the sponsorship of the Free Software Foundation. That same year, the Berkeley CSRG put out Networking Release 1 (Net/1). On 6 January 1991, Clark released _groff_ 1.00. In May of that year, [https://cgit.git.savannah.gnu.org/cgit/groff.git/tree/ChangeLog.old?h=1.24.1#n3095 _groff_ explicitly added support for _mdoc_], which [https://cgit.git.savannah.gnu.org/cgit/groff.git/tree/ChangeLog.old?h=1.24.1#n3095 shipped in _groff_ 1.02]. Also in June, the CSRG put out Net/2. "Within six months", BSDI released its operating system, adding "six files" to Net/2.[1] As an aspiring computer industry monopolist, AT&T plodded in clown shoes with two left feet. Their 3B20 model was notorious as the only computer in the world that restarted with a pull cord like a 2-stroke lawnmower engine. (This was known slightly less colloquially as the "fuse gooser".) In 1991 they executed a hostile takeover on NCR in an attempt to improve their market position, which appears not to have produced a positive result in any financial quarter. AT&T therefore made itself conspicuous by pursuing an evil goal (monopolization) with demonstrable incompetence. It doesn't require great insight to perceive that one might pursue an evil goal in a more skillful manner. (For example, D. C. lawyers are no match for the intellect of a Berkeley grad student in operating systems.[1]) A market monopoly is invariably characterized, by the Board of Directors of any firm that enjoys one, as "natural" and "benevolent". Further, in entrepreneurial culture generally and especially in Silicon Valley, one doesn't set modest fiscal goals. Why shoot for a personal fortune of a mere 10 million or 100 million dollars when you can go for a billion? Today, in 2026, we're much smarter. Just a billion? What, do you have low T or something? A trillion dollars is what it takes to play at _this_ table, homes. I think Karels understood that BSDI was poised to carry off a massive coup in capturing a huge proportion of the Unix market from AT&T. GNU _troff_'s syntax extensions stood poised to serve as a superior basis for Livingston's _mdoc_ macro package and consequently to support further development of BSD man pages. I suspect that Karels's stance against _groff_ was due to its copyleft license. I surmise that Karels perceived that copyleft licensing imposed a ceiling on the economic rents that can be extracted from software products. That ceiling makes the cost of a copy of software more closely resemble a "commodity" in economic terms, and no entrepreneur wants to market a commodity until _after_ they've made a billion dollars and cashed out their position (if they ever do). Livingston said using _groff_ to implement _mdoc_ would be "easier and faster". Why, then, reject it for that purpose? Livingston undertakes the labor of tediously backporting _mdoc_ to DWB _troff_, and Karels and his pals stand to claim a theoretical jackpot. Usenet trolls and USENIX conference attendees, and other partisans in the Unix Wars, still raging in 1991, could be trusted to assert without evidence that _groff_ was "bigger" and more complex than DWB troff. Whoever won the Unix Wars was sure to be in the catbird seat of a penthouse apartment--or a sprawling estate in Los Altos Hills--if they had a good stake in the victor's company. Because Unix was the best OS ever and there was no alternative, its license fees would be a cow that gave green paper milk forever. But the GNU Project was salting the earth of that war, threatening to wreck unrealized fortunes, by forcing people who elected to distribute its software to pass along the freedoms they enjoyed themselves. The FSF had already squeezed an Objective-C front-end for GCC out of NeXT in 1988. And then Sun unbundled their compiler from its Unix, put it under a separate licensing scheme, and drove thousands of users into the arms of that same GCC, infested with the "viral" license that managed to humble Steve Jobs himself. Who the hell is going to get rich that way? Detestable. "Community resistance", stimulated perhaps by Karels and others, could be deployed as caltrops to impede _mdoc_'s progress until something "better" came along. 1. Some BSD hacker might write a "*truly* free" (snort) DWB troff clone, BSDI could Hoover it up, throw _groff_ out, and skip merrily on its way, not releasing source code except at its own discretion (if ever), and applying proprietary patches wherever they thought it advantageous. 2. Some BSDI employee or contractor with a jones for typography might write a wholly proprietary DWB troff clone in-house...and BSDI could skip merrily on its way, perhaps charging a mere $10,000 for a source license and $75 per copy instead of AT&T's $15,000 and $120 respectively, undercutting them and still enjoying fat rents. Either way, with founder's stock in BSDI, one comes to spend weekends out on one's yacht trading insults with Larry McVoy at his fishing pole, and crow about how one's personal ability is the exclusive explanation for one's glamorous life of personal luxury. And on weekdays, there's always the back 9 at Sharon Heights. Business leadership sure is hard work. Thank you for maintaining https://manpages.bsd.lv/history.html -- it's an invaluable resource. [1] "Our standard thing was, well, you know, 'just buy our distribution [BSD] and take whatever you need out of it'. But that meant they had to get an AT&T license. Well, by the late '80s, AT&T licenses were up to, like, a quarter million dollars. And when you're in the little commodity PC business, a quarter million dollar cost just to get the software before you do anything with it was just a non-starter for them. And so we were getting requests for just that TCP/IP code and associated utilities. Clearly, since it had all been developed at Berkeley--there was none of that in the original Version 7 Unix that we had started from--AT&T couldn't lay any claim to that And so we talked to the lawyers at Berkeley, and they're like "well, how can we make money on this?"--'cause that's what they were paid to say. And so we pointed out to them that it had been developed under government contract, and we show them the government contract saying that, you know, the stuff done under this contract has to be made available to the public at reasonable cost or whatever it said. And, so, it didn't _really_ say that, but we convinced the lawyers that it did. And so they agreed to let us pull out this chunk of Berkeley code and release under this new license that the lawyers cooked up. Uh, which is the one that we think of today as the 'BSD license'. And of course our goal was to be absolutely as free to give it away as we possibly could, so really, the only thing you had to do was give due credit to Berkeley. So this was Networking Release 1. And it was a wild success. The amazing thing was that we sold a thousand--you could buy it from Berkeley for a thousand dollars--or you could just download it off of UUNET, for free. And yet a thousand people chose to play a thousand dollars to Berkeley to get it. And the only difference was that you got a 9-track tape--which many people didn't have any way of reading--and a piece of paper signed by the Berkeley lawyers essentially saying 'yes, you can use this'. And that's what they were paying the thousand dollars for. And the lawyers of course got paid something like a hundred dollars for every one of these that they signed, because of, you know, well, 'overhead', et cetera. So they thought this was the cat's pajamas. Maybe the campus wasn't making a lot of money, but their office was doing just fine, thank you very much. So, at any rate, Net/1 was wildly successful, and so there was pressure, you know. Keith [Bostic] would come in and say, 'oh well, today somebody wants _vi_' or 'somebody wants this' or 'somebody wants that'. And, you know, really, we ought to do this. And, you know, how about figuring out other parts of the kernel? Mike and I would--Mike Karels and I--would sort of look at each other and we'd look at Keith and, you know, like 'yeah, yeah, yeah right--well, you figure out, you know, I mean, you know, it's all entangled with the C library and all this and that'. So we get to the conferences and Keith is putting up, you know, this huge list of utilities. You know, 'rewrite this utility and get your name in lights!' You know, and so, he would get this stuff sent into him, and he would almost always just sort of take it and then bash it into Keith Normal Form, which usually didn't look a lot like what he had gotten. But it started from clean, so that was, you know, was good. And, you know, it was sort of easy, you know, the first stuff--someone sends in _cat_, someone like really goes whole-hog and does **all** the options to _ls_. And this is in the days before GNU software when, you know, lots of options were in vogue. And then one day, Keith comes into report that somebody is redoing the _troff_ family. And this is when Mike and I sort of look at each other and sort of gulp. Because the next question out of Keith's mouth is, 'and how's that kernel coming, **guys**?' Well, we can't go to a USENIX conference and say, 'rewrite the kernel! get your name in lights!', you know. So we actually have to, like, go through it and figure out what's Berkeley and what's not Berkeley and segregate stuff out. Long story, but eventually we get something that we feel is clean. And so we go back to the lawyers and say, 'ah, we got this little update to the networking tape that we'd like to do. We're gonna call it Networking Release 2--Net/2'. And so, actually, we do sort of come on and explain: it is practically the entire system. Lots of stuff happens, but they finally sign off on it, and we put it out there. And it's out there for a good half a year before BSDI sort of fills in the missing bits and starts selling it. And they, you know--engineers are not really that good at marketing. And, so, in particular, they think these sort of cutesy things like having your phone number be 'ITS-UNIX' and ads that say '99% off the price that AT&T charges for source code' are cute. But it really irritates the piss out of AT&T. And so they sue us! Imagine that. Well, actually, they sue BSDI. And BSDI says, 'well, we only added these six files--you know, all the rest came from Berkeley. So we'll be happy to talk to you about any of these six files that you'd like to, but otherwise, you know, forget it.' And the judge agrees with them. 'Cause they don't have any problem without those six files. And so there's no choice but to sue the university. And that's another long story, in which I become an expert witness. Which has actually proven to be a relatively lucrative thing, because being an expert witness is something that you--it's really hard to break into that, because no one wants to hire someone that hasn't done it before. And it turns out it doesn't really matter how much you know--it's really, you know, 'do you have a Ph.D.? Because if you have a Ph.D., by definition you must know what you're talking about.' Not making this up. And then number two is 'how well do you speak in public?' You know, especially people harassing you. 'Well, I've had ten thousand students, you know. There's not a question that a graduate student hasn't asked me that I haven't had to deal with.' And a lawyer is not up to a graduate student. So I did pretty well on that score, too." -- [https://www.youtube.com/watch?v=ds77e3aO9nA Kirk McKusick, "A Narrative History of BSD", D.C. BSDCon, 2009] There was no transcript available, so I transcribed the foregoing myself. I observe that McKusick, Karels, and Bostic were all employees of CSRG and founders of BSDI, and that, by McKusick's account, BSDI either existed at the time of the Net/2 Release, or within six months of it. I also find it implausible that Karels and McKusick, as BSDI founders, had no input into its marketing decisions, as if this was delegated wholly to engineers who were their subordinates--but while that might have been skin off of AT&T's nose, it's none from mine. McKusick's historical account is consistent in every detail with my speculation on the motive for Karels's veto of Cynthia Livingston's engineering prerogative. The relevant decision makers were simultaneously employees of the Berkeley CSRG and founders of BSDI. It beggars belief to assert that restricting it to the capabilities afforded by a legacy, _proprietary_ formatter owned by a hostile competitor was exercised innocently of commercial considerations. (Such neglect betrays the Friedman Doctrine, the sole ethical precept a capitalist holds.) It is consistent with a choice by a dual-hatted CSRG principal and BSDI founder, who would soon join his fellows in successfully deploying the University of California as a deep-pocketed legal shield from a lawsuit by a party (AT&T) that BSDI provoked by going out of its way to "irritate the piss out of". Such decisions are easily intelligible from the standpoint of a person with aspirations for a massive personal windfall as a shareholder of a value-added reseller extracting economic rent from proprietary software. _______________________________________________________ Reply to this item at: <https://savannah.gnu.org/bugs/?68708> _______________________________________________ Message sent via Savannah https://savannah.gnu.org/
signature.asc
Description: PGP signature
