Why “open source” is not mere marketese

Every once in a while I hear it alleged that “open source” is just a marketing device for a practice that would be just as well off without it. This is seriously wrong, but it’s a confusion I’m partly to blame for because I have emphasized the marketing utility of the term in the past.

Yes, the term “open source” is partly a marketing device. I proposed the adoption of the term in 1998 because I felt we needed a term for the practice of the hacker community less ideologically loaded than “free software” in order to sell the idea to the mainstream. It was a pretty successful marketing device; we passed out of early-adopter stage around 2003 five years later, which compares well with 20 years of failure by the FSF to make a similar breakthrough in the public perception of the “free software” brand.

However, “open source” was never just a marketing label. Marketing labels usually only have feel-good emotive content. On rare occasions they may weakly imply a descriptive theory of the merits of the product; consider “workstation”, for example. What marketing labels never contain is a generative theory of how to improve the product — because their purpose is to make you believe it’s already as good as it can get and you should buy, buy, buy, now, now, now.

But the term “open source” does entail an entire generative theory of how to improve the process. If you are calling what you are doing “open source”, you know (for example) that your process will be improved by changes which (a) shorten the feedback loop between software changes and public testing of those changes, (b) reduce the difficulty of making casual contributions, and (c) increase the size and variety of your developer group.

In this respect, “open source” escapes being mere marketese and is like a normal engineering term of art for a methodology; compare, for example, “agile programming” which is also both a marketing term and a methodological one associated with similar process ideas. (The similarity is probably not accidental; several of the founders of the agile movement cheerfully admit to having been influenced by my work.) This is one reason I have always insisted it should not be capitalized: “Open Source” looks like puffery from someone’s ad campaign, but “open source” does not.

As always when the history of this term comes up, I like to give credit where credit is due. I did not invent it. That honor goes to my long-time friend Christine Peterson, futurist and nanotechnology advocate. What I did was (a) attach her term to the generative theory I had developed in 1996-1997 (in The Cathedral and the Bazaar), and (b) successfully persuade most of the rest of the community to use it.

As usual in open source, the most important skill involved was at being egoless enough to recognize someone else’s good idea. Both I and the larger community managed to do that with respect to Chris’s “open source” label, and all have benefitted greatly as a result.

29 comments

  1. I tend to capitalize “open source” in a way somewhat akin to use of big-L or small-l in “libertarian”: it’s open source if it’s generally inkeeping with the philosophy. It’s Open Source if it uses an OSI-approved license.

  2. Would you say “open source” is a term that applies only to the developer’s perspective? That is, should computer users who don’t program care at all about “open source”?

  3. >Would you say “open source” is a term that applies only to the developer’s perspective? That is, should computer users who don’t program care at all about “open source”?

    Sure. Increased reliability, open formats, and the death of vendor lock-in matter to end-users too. Our job is to help them understand this.

  4. I have deleted some contentious and off-topic comments. Arguing that some of the consumer-visible virtues of open source can be achieved without open source is not relevant to the question of whether open source is mere marketese or not; regardless, it still describes a methodology with predicted consequences and a generative theory attached to it.

    Some of open-source’s virtues can be achieved in closed source, but require unusual conditions and/or heroic effort. The most difficult one is probably reduction of defect rates; real-world data such as the Fuzz tests and Coverity scanning show that open source has a substantial advantage there. One one project I’m personally involved in, Coverity scanning turned up just two errors in 22KLOC, a defect rate most managers at closed-source shops would consider an impossible dream. Yes, NASA gets lower error rates on embedded software for spacecraft, but only by throwing an amount of money at QA and review that no vendor of closed-source merchant software can afford while remaining price-competitive.

    And with that observation I am ending the discussion of whether open-source benefits can be replicated in closed source, to be picked up in some future post where it is on topic. Further off-topic comments in this thread will also be deleted unless I judge them to be (a) factual rather than contentious and (b) of exceptional merit.

  5. Eric, I think its pretty punk of you to delete the comment stream, rather than objecting to it, but its your blog, and you get to run it as you see fit.

    Eric says: You, Jim Thompson, are the single most severe cause of thread pollution on my blog. If you reform your behavior, the likelihood that I will have to winnow threads will drop dramatically.

    In that spirit, I offer this: http://mjg59.livejournal.com/100221.html

    in which Matthew Garrett explains how Google has failed to co-operate with the linux kernel community to the detriment of both. Google has seemingly ignored the power-management infrastructure in the linux kernel (and the hooks to same from userland) and setup their own, parallel but separate infrastructure, located entirely in-kernel.

    This seems to fly directly opposite your assertion that the process of open source generates better code quality (and destroys vendor lock-in, something very much present in the G1 phone.)

    This example should be of some interest, given that you own a G1 phone.

    Apple’s behavior with “WebKit” .vs KHTML is similar.

  6. Open source may not be marketese, but some people seem to tout it as a panacea, the way object-oriented programming were in the 1980s. The thing they all had in common was that all software problems would magically disappear if developers just adopted some methodology. The short release cycles and peer-reviewed code do allow for better software, but some projects fall by the wayside because of human factors, not technical problems. I suppose one must approach claims of the “one best way” with a healthy skepticism.

  7. >This seems to fly directly opposite your assertion that the process of open source generates better code quality (and destroys vendor lock-in, something very much present in the G1 phone.

    Google fails to follow good open source practice, unhappiness results, and you somehow expect me to take this as a refutation of my assertions?

    It is to laugh.

  8. “Increased reliability, open formats, and the death of vendor lock-in matter to end-users too.”

    Would you say that the freedom to copy and share code with other non-programmers is a virtue of open source as well? Or is this where the emphasis on the philosophical differences between open source and free software are most apparent?

    That is, from a developer’s perspective, open source is a superior model for software authorship while from the user’s perspective, free software and open source software advantages might be best emphasized in tandem?

  9. > single most severe cause of thread pollution on my blog
    it might be helpful to point out (even privately) which parts upset you enough that you engaged in censorship.

    > It is to laugh.

    and now we’re right back to what I’d pointed out before: “it takes sound design of the overall system at so many higher levels that open source is really only a minor part of the toolkit.”

    You deleted these words from this thread recently, but of course, these are *your* words.

    Perhaps the open source process (method?) helps with design, as a sort-of “massive peer review”, and the ‘ship it when its ready’ mentality helps keep this kind of “hack it in and fix it later” type of thing from happening, but, as you said, getting the design (architecture, whatever) right is of massive import compared to the actual “open source” process. It could be said, (perhaps) that the innovation happens at the design layer.

    You might blog about this. It could be an interesting distinction.

  10. >Would you say that the freedom to copy and share {code|programs} with other non-programmers is a virtue of open source as well?

    The freedom to copy and share code (which I take to mean source code) is, for non-programmers, one of mainly theoretical importance. The ability of non-programmers to share programs (which I take to mean deliverable and working forms of code) is of great practical importance. I, personally, am not very attached to the latter, but open source as a method is not supposed to be about my ideological positions. It’s suppose to be a relatively value-neutral set of claims about what makes better software and maximizes efficiency in the production of software.

    That being said, if you want to argue for open source using value-laden “ought to” propositions, I probably won’t try to stop you. Just don’t expect your value-laden propositions to necessarily be the same as mine or any other advocate’s.

    And that is “probably”, though my tolerance is large. If you try to equate open source with or justify it in terms of an ideology I consider fundamentally evil, I will have words to say about that.

  11. >The freedom to copy and share code (which I take to mean source code) is, for non-programmers, one of mainly theoretical importance. The ability of non-programmers to share programs (which I take to mean deliverable and working forms of code) is of great practical importance.

    At least nontechnical users can share programs without being subpoenaed. Remember how popular “shareware” was in the first half of the ’90s, even though the code was still proprietary?

    (All software is open source if you try hard enough. ;-))

  12. If I pirate a piece of software, if I copy MS Office for example, Is that “open source?”

    By the way, has anyone here tried Google Chrome? I’ve been playing around with it a little, you can scroll up and down a page a bit faster that Firefox enables you to do. It seems pretty slick.

  13. Darren, with the exception of things that require security and scriptblocking, Chrome has become my default browser.

    It’s faster and more responsive than FireFox, it doesn’t eat memory resources as atrociously, and it’s got several UI tricks I appreciate.

    I still use IE for Windows Update stuff, but that’s about it.

    Phil, I enjoyed the links to the Linux Hater blog.

    FYI – the Ubuntu installation that I did two months ago for the friend whose computer I wiped clean and reinstalled on…has appreciably been used twice in two months before being shut down and re-started in Windows. Fairly bright computer user. Not a programmer, no desire to be one. Just wants to read emails, listen to music, watch videos on the web, do some spreadsheets and word processing.

    I made sure that OpenOffice saves to .doc and .xls formats rather than .odf as its defaults, because she needs to be able to exchange files.

    Her answer – “I already know how to use Windows, I don’t see any reason to learn how to use Linux when it doesn’t do anything Windows doesn’t that I care about. and most of the things I do care about, take more time and are harder to understand.”

    Many of the things in the Linux Hater blog are spot on with my experience with supporting her in this.

  14. JimThompson: anybody who manages to complain about censorship in the same venue in which he claims to be censored is obviously not *actually* being censored. And anyway, how was Eric able to stop you from publishing those same things on your own blog (a requirement for actual censorship to have taken place)?

    BTW, Eric, a trademark pretty much HAS to be capitalized, otherwise it’s not a proper name, is just a word, and not deserving of trademark protection. I always capitalize Open Source, and I’ve never seen you object to it. Since it’s Too Descriptive, perhaps I should Stop Capitalizing it. Okay, in deference to your wisdom on this subject, I’ll stop capitalizing open source.

  15. >If I pirate a piece of software, if I copy MS Office for example, Is that “open source?”

    Of course not. In what way was this not obvious?

  16. > Google Chrome

    Is very slick, indeed. (And serves as another interesting counter-point. Why didn’t “the community” develop something Chrome-like? (Chrome has a very “unixy” architecture, despite being written for Windows. It would have seemed an entirely natural development of the open source process.

    @ russ nelson: two quick counter-examples, “macy*s” and “adidas”. Both trademarked in lower case. (and the lower-case is matter for the mark.)

    I’d explain ‘venue’ to you, but it would be a waste of time.

  17. Chrome is a really nice browser. I love how each tab has its own process. I’m just waiting for it to come out for Mac and Linux. Plus, I’m too addicted to StumbleUpon to switch from Firefox.

  18. > It is to laugh

    But of course, Google managed to get you to buy your first new phone in 8 years largely on the hype (that would be marketing) of it being open source. Yes, you can write programs for it, in a forked version of Java, and yes, you can get to a # prompt, for another couple weeks, until the OTA software update arrives against your will and closes that hole.

    Surely the point is not lost on you that despite all your protests about open source being more than mere marketese, you’re stuck running their less-than-optimal kernel.

    And you’re laughing at me?

  19. This seems to fly directly opposite your assertion that the process of open source generates better code quality (and destroys vendor lock-in, something very much present in the G1 phone.)

    Jim, you seem to be saying that Google’s new phone is not “open” as in “open source.” I’d agree that Google having used some open source software does not necessarily make their phone “open.”

    If Google’s phone locks any user into anything, then it is not open. That would be Google’s fault. Regardless, that does not reduce the quality of open source code/projects.

    But of course, Google managed to get you to buy your first new phone in 8 years largely on the hype (that would be marketing) of it being open source.

    According to Eric’s blog entry above, “Yes, the term “open source” is partly a marketing device.” Partly.

    Google is an advertising company. They are marketing. Google’s marketing does not mean open source is only marketing–even when Google uses the phrase open source in advertisement(s).

  20. I agree “open source” matters to the end users. Only problem is: they don’t understand it and don’t see the necessity to mistrust monopolys. That’s where marketing comes into play.

  21. >I agree “open source” matters to the end users. Only problem is: they don’t understand it and don’t see the necessity to mistrust >monopol{y,ie}s. That’s where marketing comes into play.

    How does marketing help people arrive at that conclusion, i.e. in an honest manner? Since marketing stands upon creating desire in heterogeneous groups of people to exchange one thing for another, I cannot see how marketing is the solution that will give people the knowledge they need in order to achieve an understanding why free software solutions are good for them or why their proprietary opposites are bad. The most intellectually offensive things I see on a daily basis are adverts from various markets trying to create the desire in me to trade money for something by attempting to convince me I need something that I actually don’t. I arrive at the inevitable conclusion that while it might be possible to use marketing to create enough desire in some people to switch to using free software under the banner of open source development, it isn’t possible to do so and simultaneously convey an honest reason why because honesty and marketing are natural enemies.

  22. “Open source” is really a controversial concept. It changed the whole software industry a lot during the last years.
    >As usual in open source, the most important skill involved was at being egoless enough to recognize someone else’s good idea.
    That’s really hard. I sometimes notice my simple consumer attitude to the open source projects. I wish I were more appreciating.
    I wish I always saw someone’s small or great achievements, appreciate them and enjoy with gratitude, rather than rehash all the way I need, as it happens very often.

Leave a Reply to esr Cancel reply

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