… Not politically better, not feel good better, not any of that. Well, yes, that too. But for all areas where Open Source is developed, it it simply better technologically. Anybody telling you different is selling you a bill of goods.

I tried to say something of this sort here, on Bill Hooker’s blog post “On science and selfishness,” but his #%$@$#$ (presumably OpenSource) blog software would not accept my comment!!!! (Hey Bill, get that fixed, man, you’re ruining it for everybody!!!). So, instead of saying it there, I’ll say it here. This is my comment:

~~~

Bill:

I deeply disagree with an underlying assumption made in this blog post. That assumption is that it is NOT the case that OpenSource software and OpenAccess publication are technically, systemically, and procedurally superior to closed source equivalents. That is simply untrue.

For the most part, developed OpenSource software is technologically superior to closed source alternatives. More secure, less buggy, faster and more efficient, stable features that don’t change to suit marketing strategies, support that does not rely on Orwellian tactics and so on. In the mathematical sciences, OpenSource software is the ONLY valid way to carry out studies that you expect to be taken seriously (though most math people don’t get this, astonishingly) The OpenAccess publishing sources are simply a zillion times easier to use for the end user. How is that not better?

Otherwise I agree with everything you said.

Greg

Comments

  1. #1 dean
    March 13, 2009

    Interesting points.
    Re this: ” In the mathematical sciences, OpenSource software is the ONLY valid way to carry out studies that you expect to be taken seriously (though most math people don’t get this, astonishingly)”

    - the last part, about math people. I do realize that SAS, S-plus, maple, mathematica, etc., are widely used, but I (and many people I know) use R, sage, latex, flavors of linux, and so on. (Okay, I use them because my school is fundamentally business-progam oriented, and they wouldn’t shell out big dollars for the name programs even if I asked for them.) I realize that a good number of math/stat folks use commercial stuff, but is it really “most”: I’m not sure.

  2. #2 Greg Laden
    March 13, 2009

    Dean: I think it is most, but I’m not sure. I’ve heard that but can’t give you a citation.

    Certainly, last time I noted on this very blog that journals should not publish math results produced with software for which we cannot see the algorithms, I got hammered by a zillion math people. A sensitive lot, it seems.

  3. #3 D. C. Sessions
    March 13, 2009

    Datum [1]: $DAUGHTER is the stats geek in her cohort. She’s also a hard-core Linux user [2]. The scandalous price tags and lack of Linux versions for the fancy math packages have led her to learning Octave and R instead.

    Now there’s no money for big-ticket maths packages and guess who’s in demand?

    [1] Read, “Thinly-disguised anecdote”
    [2] Talk about having the male students at her undergraduate school worshipping at her feet. And that was before they found out that she did her own Stage One Gentoo installation.

  4. #4 DuWayne
    March 13, 2009

    The thing that I love most about using open source, isn’t even something I am capable of on my own – though I thankfully have friends who are. I like the notion that the applications that I use can be easily tweaked to work a little better for the uses I put them to.

    I am kind of torn on linux though. On the one hand I would like to be a little more independent, on the other I am ridiculously focused on other things and don’t know that I need more on my plate. I definitely want to get off the fucking microsoft OS, but I am thinking in the direction of going with an open source OS, rather than learning linux at this point.

  5. #5 bill
    March 13, 2009

    …assumption is that it is NOT the case that OpenSource software and OpenAccess publication are technically, systemically, and procedurally superior to closed source equivalents

    I don’t know that I really operate under that assumption regarding software, but I can see how it would come across as though I did. (I think I’m pretty clearly an Open Access nut.)

    If asked, I’d have said that proprietary software covered more kinds of applications but what F/OSS does, it does at least as well as proprietary equivalents.

    Thing is, I’ve used very little F/OS software, for all the same reasons I bemoan in others: too busy, don’t have time for the learning curve, need this done yesterday so just going to use the software I’ve always used… Little by little though, I’m trying to wean myself off proprietary wares.

    (And yeah, something is up with my MT install I think — comments are borked and it’s doing horrible things to my entries. I’ll poke at it but don’t expect results until the Spousal Unit tires of my cussing and takes over.)

  6. #6 D. C. Sessions
    March 13, 2009

    I am thinking in the direction of going with an open source OS, rather than learning linux at this point.

    There are others besides Linux?

    DuWayne, just go score a copy of Ubuntu and boot the silly thing. It’s a live CD, so you don’t even have to install it to take it for a (pardon me) spin. It’s just not a big deal — and there are lots of us who will be glad to point you at solutions to any problems you run into.

  7. #7 samk
    March 13, 2009

    I do enjoy F/OSS. One example of the superiority that can be found in F/OSS projects is gnomad2 which I have used with a Creative Zen MicroPhoto. It’s only an app for an mp3 player but I invite anyone to compare that application with the POS bloatware that actually ships with the device.

  8. #8 D. C. Sessions
    March 13, 2009

    There are others besides Linux?

    Yes, I’m aware of the BSDs, OpenSolaris, and Plan9 (as well as some really obscure ones.) The point being that someone who doesn’t have the time for Ubuntu is freaking nuts to go down those roads instead for their first ventures into NixVille.

  9. #9 dean
    March 13, 2009

    “Now there’s no money for big-ticket maths packages and guess who’s in demand?”

    Not to sound like I’m pushing it, but suggest SAGE to your daughter as well. It runs in command line or in a browser window/notebook style. It’s essentially a Python front-end that provides access to several programs (maxima, which , octave, (it comes with maxima, but will work with the others as well).

    It also comes with a latex class that allows sage code to be embedded in latex files and executed at compilation. very easy to use.

  10. #10 D. C. Sessions
    March 13, 2009

    Not to sound like I’m pushing it, but suggest SAGE to your daughter as well.

    Thanks for the tip!

  11. #11 P Farrell
    March 16, 2009

    I am almost convinced by your argument that science can only be done with open source software. What about hardware? I am not thinking of computers. I am thinking of measurement tools such as oscilloscopes, temperature meters, optical power meters, strain gauges,voltmeters, pH meters… you name it. Should these be open source as well, or should we trust Agilent or Anritsu or Fluke or Acme to produce a device which meets a specification and we don’t care what’s inside the box. My experience is that the further you are from being an engineer or perhaps a physicist, the less you are likely to understand about the inner workings of the magical doohickey that produces your measurements. It is not very common to be able to get a circuit diagram from a manufacturer any more. It is less common to be able to get information about what is inside specialised integrated circuits which are part of the device. Does this lack of openness cause the same sort of difficulty as proprietary software?

    Can we rely on a suite of test measurements which verify that the machine does indeed meet its specifications and not worry about the details of what’s inside? If the answer is yes what’s different about software?

  12. #12 Greg Laden
    March 16, 2009

    <>I am almost convinced by your argument that science can only be done with open source software. What about hardware? I am not thinking of computers. I am thinking of measurement tools such as oscilloscopes, temperature meters, optical power meters, strain gauges,voltmeters, pH meters… you name it. Should these be open source as well, or should we trust Agilent or Anritsu or Fluke or Acme to produce a device which meets a specification and we don’t care what’s inside the box.<>

    Good question. In the areas I’m somewhat familiar with, which would be mass spec and maganetics, allmost all of the hardware is “open source” with only a few bits and pieces probably not being quite as open. It seems to me that any technology whereby the explanation for how something works or what the outcome might be has the phrase “and then here, for this part, there’s an input and an output but if we tell you how it works we’ll then have to kill you. Or have you sign a non-disclosure agreement, at least” should not be allowed in a peer reviewed journal.

    How’s that for a radical position?

  13. #13 P Farrell
    March 17, 2009

    I like your radical position. Let me try an example that clarifies my own position. Let’s say you buy a Model 1001 Mark III Thingumybob to measure insulin levels in vivo. You know what the hardware is because you can open the box and look at it. You can xray the chips and reverse engineer them and so on. In the device there is a machine learning algorithm which has been trained on test data you do not have access to and you don’t have access to the code that translates whatever the machine measures to produce the insulin level. This machine has been in use for some time, is used by lots of medical practitioners and there is a large quantity of evidence that it produces very similar or the same answers as the test strips or whatever the state of the art is. I think I would be no less comfortable trusting my life to a well tested and engineered machine like that than I would be to some other process the inner working of which I had more complete knowledge. I think I should also trust it for a scientific publication provided that the publication includes sufficient information about the model and state of the machine (model number, firmware version, etc) so that issues which arose in the future could be sorted out.

    The interesting thing about this example is that not only is it true that the user doesn’t know in detail how the thing works, it may be that nobody knows how the thing works other than a machine learning approach produced an inaccessible algorithm.

    I should say I have ubuntu loaded on to all of the computers under my control. Some are dual boot with the unmentionable. I think the only commercial software I struggle to do without is Maple.

    I am tempted by the delicious excitement of recommending rejection of a paper because of its reliance on proprietary software or hardware. I can hear the howls of outrage already.

    There is a fledgling open source hardware philosophy which I hope is growing. Have a look at http://www.opencircuits.com/Open_source_electronics