How “open source” was coined

Yesterday was the 20th anniversary of the promulgation of the term “open source”. Three days before that, Christine Peterson published How I coined the term ‘open source’ which apparently she hd written on 2006 but been sitting on since.

This is my addition to the history; I tried to leave an earlier version as a comment on her post but it disappeared into a moderation queue and hasn’t come out.

The most important point: Chris’s report accurately matches my recollection of events and I fully endorse it. There are, however, a few points of historical interest that can be added.

First, a point of fact. Chris doesn’t remember for sure whether it was me or Todd Anderson that first brought up the need for a new umbrella term to replace “free software”.

It was me, and it was me for a reason. In those very early days I was ahead of the rest of our leadership in understanding a few critical things about our new circumstances. Others eventually did catch up, but at the time of this meeting I think I was the only person there who had fully grasped that what we needed to take our folkways mainstream we needed a full-fledged marketing and rebranding campaign.

That insight was driving my thinking, and led me directly to the conclusion that we needed a better label for the brand.

It was a little tricky for me to talk about this insight explicitly, and I mostly didn’t. Because I also knew that in the minds of the people in that room, and most other hackers, “marketing and rebranding” was a negative idea heavily tied to fakery and insincerity. I couldn’t blame them for this; I’d had to struggle with the concept myself before making some peace with it. But…I was unable to evade the historical moment, and saw what was needed.

Therefore, I had the dual challenge of trying to lead a rebranding campaign while easing hackers into comfort with the idea of rebranding. That meant not pushing too hard or too fast – choosing my interventions carefully and allowing other people to catch up with the implications in their own ways.

Chris couldn’t read my mind, so she had no way to know that I spotted “open source” as the winner we were looking for the first or second time the phrase was mentioned – well before I explicitly advocated for it myself. It seemed perfect to me – ideologically neutral, easy to parse, and with just enough connection to an already respectable term of art (that is, intelligence-community use of “open source”) to be useful.

There was also something right about the use of “open” that I couldn’t pin down exactly at the time. I knew this was an adjective with a lot of positive loading attached in the hacker culture, but I was not then clear on the exact psychology. Now I think I understand that better – it evokes the Big Five trait “openness to experience”, which we value a lot. That particular term didn’t exist yet in 1998, but the connotative web that would later give rise to it did.

In accordance with my strategy of wu wei I hung back a little and let other people there gravitate to “open source”, rather than pushing for it as hard as I could have right out of the gate. I would have fought for it over the alternatives if I’d had to, but I didn’t…we all got there and that was a far healthier outcome than if I had tried to dominate the discussion.

I’m telling this story this way because, now that Chris has admitted she was being a bit stealthy about getting the term adopted, she deserves to know that I was being a bit stealthy myself – for reasons that seemed good at the time and still do in retrospect.

The rest of the story is more public. A few weeks later what was in effect a war council of the hacker community’s chieftains, convened by Tim O’Reilly, voted to endorse the new term and in effect gave me a mandate to go out and evangelize it. Which of course I did; the rest is both metaphorically and literally history.

I think Chris is fully entitled to her happy twinge. From the perspective of twenty years later, “open source” was a smashing success, fully justifying both her and my hopes for what we could do with this rebranding.

Without Chris, I would have had to come up with something as good as “open source” myself in order to get the mission done. Maybe I would have, maybe not. I’m glad we didn’t have to roll those dice; I’m glad Chris nailed it for all of us.

Ever since I was first reminded that it was her coinage I’ve been careful to credit Chris for it. I was impressed with her for the invention, and that developed into a friendship that we both value.


  1. Christine Peterson deserves much praise for the coining. It’s one of those things where, looking back it’s obviously good even though it wasn’t beforehand. As someone who wasn’t in the movement at the time though I was certainly sympathetic with it, I found the “open source” term to be immediately understandable to everyone, whereas other terms weren’t. I don’t just mean free, other projects that I was aware of and using, such as perl, had different terms and they were all less than obvious.

    1. >There are many citations, including from 1995 and earlier, that demonstrate that the term was being used in the software industry prior to the meetings discussed.

      We (that is, the people at that meeting) discovered some of those after the fact. We absolutely didn’t know of them before; it was independent reinvention, as Chris makes clear and I confirm. Actually, if we had known of those prior uses we would have referenced them as a way to get the term accepted faster.

      The earliest uses I subsequently learned of of were in the late 1980s on USENET. But use of the phrase about actual software was rarer than you might think – most of the hits referenced in the Slashdot discussion (including for example the RISKS digest one in ’89) are actually on the intelligence sense of a non-secret source but not referring specifically to software source code.

      I’ve never bothered to argue the point before because really, how do those sporadic earlier uses matter? I’m sure it’s actually pretty common for technical terms of art to be independently reinvented a couple of times before they catch on. This is not a big deal.

      I suppose some of the Slashdot carpers are bent enough to believe that that Chris and I nefariously plotted to steal somebody else’s linguistic invention and then conceal the dirty deed, but there’s no point in getting down in the mud with that sort of paranoid.

      1. > how do those sporadic earlier uses matter?

        Since none of the people at the table knew anything about them, they don’t.

        Just like how Leif Ericsson’s voyage to America didn’t really matter because no one followed up on it like they did Columbus’.

      2. Wasn’t “open systems” a big buzzword of the 80’s?

        Furthermore, when was the term “shared source” first used for proprietary-but-publicly-available source code?

        I can imagine that if “shared source” was already a term being used, and was contrasted to “closed source” for proprietary and unavailable code, then in conjuction with “open systems” being a buzzword in recent memory, “open source” would be a fairly natural coinage for “shareder-than-shared” source.

        1. >Wasn’t “open systems” a big buzzword of the 80’s?

          It was. In fact, one of our few worries about “open source” was hat it might have been traduced in advance by marketing abuse of the adjective “open” during the workstation era. In origin “open systems” actually meant something like “you can forward port your old Unix code” but API divergences in that pre-POSIX era (especially the AT&T/BSD split) rapidly turned this into a hollow promise.

          >Furthermore, when was the term “shared source” first used for proprietary-but-publicly-available source code?

          That I do not know.

        2. I’m not a big fan of open source, not anymore, because it leaves the central issue of user freedom out of the discussion.

          But one advantage it did bring was that in the early-mid 90s, “open” meant roughly “available under reasonable and nondiscriminatory license terms”. Open source changed all that; now if you try to pass off a proprietary standard as open, you’ll get a drubbing on Slashdot, Hackernews, and elsewhere. Even Microsoft is putting on a show of playing nice with open-source (free software) stacks.


          Let that sink in.

  2. Again, I follow with interest while utterly completely disclaiming any particpation in the culture.

    How did “open source” beat out “shareware” — which seems to have been another earlier antonym to “free software / freeware” ? I’m thinking I might remember Jim Button (not his real name?) as the advocate for the try-before-you-buy; widely place copies everywhere, model of the “shareware” alternative.

    It’s sort of similar, I think , to the diagonally opposite pairs of antonyms for “wild”: Feral, Domesticated, Tame. Shareware wanted, but didn’t usually get, money, while open source wanted, and still doesn’t usually get (I think) participatory improvement. But as a marketing term and a business model Shareware seems to have become extinct.

    Or I could be utterly wrong…

    1. >But as a marketing term and a business model Shareware seems to have become extinct.

      That is so, and the reason is simple. Shareware was much more brittle than open source under pressure from platform transitions.

      Consider two identical and useful programs, one distributed with a pointer to source code ans one as a shareware blob. Which is likely to survive to propagate on newer OSes and hardware? Asked that way the question answers itself – the source code form may be less convenient for use but will be far easier to forward-port.

      1. In the 1980’s, replacing a machine usually meant changing at least OS, often also CPU architecture. So it was “normal” to have to collect a new set of shareware blobs whenever hardware turned over. But that didn’t happen very often, and it happened while everything else was changing at the same time.

        In the 1990’s, I had switched to PCs. Every minor update to DOS or Windows was breaking some important shareware blob or other. After installing the third generation of entirely new blobs on the same machine, I had reached my lifetime quota of shareware blob replacements.

        Then I switched to Linux, and replaced the shareware with open source. I’m still running software every day in 2018 that I first installed in 1993, and I barely noticed the x86 to arm and amd64 transitions.

        So…clear open source win? Not as such. I have installed a _lot_ of updates over the last 25 years, and I only wrote code for a handful of them. A shareware maintainer, suitably financed, could have done that for the smaller, single-author packages.

        Where open source really wins is in the larger packages whose maintainers required specialist skills, like the kernel or GCC. Shareware certainly wouldn’t work if there was a single point of financial weakness at the package maintainer’s tip jar. In the heyday of shareware, Microsoft was actively interfering with the flow of money to anyone with a competing OS platform for the PC market, while also selectively enforcing its licenses to behave like the largest established shareware vendor in that market (or maybe they just couldn’t keep up with the pirates, with the same net result). The only survivors were niche OS products that had non-PC revenue sources and open-source products that didn’t trade in money.

        1. >A shareware maintainer, suitably financed, could have done that for the smaller, single-author packages.

          In theory. I’m not aware of any instances of shareware surviving an actual OS transition. On the other hand, I can think of many instances that did not.

          So I think your model is incorrect.

          1. Archive managers and some exotic early multimedia tools (an MPEG player for Linux and Windows that I even paid for) are two examples I can think of where shareware jumped from DOS or Windows (or Amiga of all things) to Linux.

            It only works in cases where the shareware author happened to make the same OS jump as the users, or where the shareware was designed from the start to jump OSes. Both of those were really rare in the shareware world, but not non-existent as a class.

            Today we have Android and iOS applications with free and paid versions. That’s basically the same thing as shareware, and maybe like five of them even share a common source tree on the two flavors of phone.

          2. The thing is, a lot of shareware was dead, in terms of sales or active development, before the breaking OS transition ever happened, and for the PC platform, exactly when the breaking OS transition occurred can be hard to pin down.

    2. Shareware wasn’t really killed by open source; it was killed by adware, which (after always-on consumer Internet became common) was more effective at delivering revenue to distributed-for-free proprietary software than relying on consumers to register. The “ecological niche” where you used to find shareware on consumer personal computer systems is mostly filled by adware on consumer smartphone systems . . . often enough with pay-to-disable-ads replacing pay-to-disable-nag-screen.

      And, heck, despite Windows adding native support for zip files fifteen years ago and 7-Zip being open source, WinZip still is in business as shareware on Windows, Mac, iOS, and Android.

      1. >Shareware wasn’t really killed by open source; it was killed by adware

        I think it’s more accurate to day that the ecological niche formerly occupied by shareware collapsed under pressure from open source, then some of what shareware had formerly done (the more user-facing stuff) was revived by adware when adware became a thing. Which was well after 1998 unless I’m missing something.

        1. Most of the shareware I can remember that got killed was one-shot entertainment. Practically no one plays the original DOOM or Commander Keen anymore, because there’s stuff like Starcraft 2 or Witcher 3 or Farcry 4 or Civ 5.

          Shareware in general got killed by a lot of things. Sometimes it was open source libraries – a Python lib will do everything PKZip did. Sometimes it was adware – i.e. stuff that works out of the box but keeps throwing ads at you.

          Sometimes it’s websites. Remember: a lot of things that required an app now just require putting a URL in a browser. The browser isn’t shareware (often it’s open source!); it just delivers freeware to you. Or rather, the interface to that freeware. (And some websites are adware themselves.)

          Most users don’t seem to want to tinker with software, so open source is less important to them. They might want a little control, but only a little, and the UI handles that part adequately.

          Some users DO like a little more creative control, and I notice that in many cases, that’s handled by extensive modding or plugin interfaces. Even entertainment software survives for years on this. 1992 DOOM did, via map modders. Skyrim is a modern example.

        2. Mmm. I think you’re seeing things too much from a Unix-centric hacker perspective. Shareware may have crept on to Unix systems for a while and been driven off by open source, sure. But its native habitat was the consumer personal computer running consumer OSes. And while some individual shareware packages on those systems got obsoleted by open source (or, just as often, Microsoft and Apple adding equivalent features directly to their OSes), shareware as a model itself seemed to me (as a user of Mac and Windows in addition to Linux; I swivel-chair between keyboards for all three) to be pretty healthy until adware replaced it.

  3. > Now I think I understand that better – it evokes the Big Five trait “openness to experience”, which we value a lot.

    The relevance of this trait to hacker identity, and geek identity more generally, was relevatory for me. Intuitively I (and I think many of us) recognized that the proclivity for both intellectual and creative pursuits were essential to this identity, but it was hard to understand why that was the case when, in the broader cultural context, intellectualism and creativity have been construed as opposing forces.

    Psychology has helped us understand that they are actually both aspects of the same mode of being – a mode that prefers the complex, possible, and original to the simple, factual, and conventional.

    The question now is whether there is anything outside being high on trait openness that explains geek (or hacker) identity. I think there are a couple things:

    1) It may have at least something to do with a failure (or lack of interest) to engage in the peer socialization and dominance games that result from the industrialized, mass education employed in many western countries.

    2) It also has to do with an engagement with communities that facilitate and encourage “user generated content”.

    1. >Any plans for a 20-25 year retrospective on Cathedral and the Bazaar et al?

      Haven’t made any yet. Not super-interested – I got that mission done, I’d rather write code.

  4. I very much like the dicotomy between Open Source and Free Software. Open Source focuses on the freedom of the devloper to reuse both code and ideas in the creation of new software. Free Software adds another, orthogonal layer with users rights on top of that. I like the separation of concerns, and it might be productive to coin a new term for users rights in separation from the source code concerns.

Leave a comment

Your email address will not be published. Required fields are marked *