Waste and abuse are so intrinsic to the American medical condition that calling out an instance so specific would be tantamount to picking out a single needle for removal from the haystack that has been dumped in your living room. But, here I go.
http://www.sas.com/ is an ostensibly reputable closed-source software for data-analysis; the http://www.r-project.org/ is a well-respected, if complicated, open-source software for data analysis.
SAS licenses are expensive. How many public dollars flow into SAS's coffers every year not only from the CDC but also from public universities and other government institutions? Tallies of government outlays for SAS and other closed-source software are not very easy to come by. What are the benefits of the expenditures?
Instead of investing public funds into the development of technologies in the public domain (R), these public institutions invest funds into private companies that hold a monopoly on their proprietary product.
http://www.cdc.gov/std/SAS/SASLicense.pdf
Can anyone help dig up some figures or provide an argument in favor of the outlay of public capital to the SAS corporation?
I argue that the scientific process of analyzing statistics is fundamentally at odds with a black-box statistical software. To be truly scientific, mustn't the software used for calculations be transparent and peer reviewed?
Instead of holding data in proprietary binaries, why not hold the data in universally-accepted, standard formats like csv? Well, that would provide an easy way for the victims to port their data to another software.
http://www.sas.com/nextsteps/index.html
The actual cost of a single SAS license is something of a secret too.
You make some very excellent points. The R project has gaind a lot of worldwide respect. Now we need all the U.S. public institutions to climb on board.
<lukeman> bsd
<jtauber> are there things it does that you can throw away because Pinax already provides them? I assume it was designed as a product rather than a library of apps
<lukeman> yep--user registration and stuff is stripped out
<lukeman> tab refactor was glorious for this--it uses the exact same style of sub nav tasks
<jezdez> jtauber: ftw!
<jtauber> i didn't think anything changed in subnav, though :-)
<lukeman> heh
<lukeman> probably not, but still... tab refactor is win
<jtauber> true
<jtauber> was glad to finally do it
<bartTC> http://github.com/pinax/code.pinaxproject.com/commit/a8152881e49c3024f7500a72f43863bdcff9236c
<bartTC> mind i18n brosner :)
<bartTC> (means less work for me :P)
<brosner> oh, i thought you had a fork...
<brosner> ;-)
<jtauber> :-)
<jezdez> heh
* ntoll (n=ntoll@88-110-185-199.dynamic.dsl.as9105.com) has joined #pinax-dev
<skyl> so does an event need an optional 'end' or is that just a can of worms?
<lukeman> GPL v3 is getting me down--looking for some type of inventory management reusable app... anyone know an alternative to http://code.google.com/p/django-inventory/
<lukeman> I don't mind to start a fresh app, but not going to reinvent the wheel if I don't have to
* rizumu has quit (Read error: 54 (Connection reset by peer))
* rizumu (n=rizumu@209-188-117-199.taosnet.com) has joined #pinax-dev
* ntoll (n=ntoll@88-110-185-199.dynamic.dsl.as9105.com) has left #pinax-dev
<skyl> so the comments app only allows a certain number of characters but does just goes to a white screen if I try to post with a lot of characters...
<skyl> I guess because there is a template there for the comment form but nothing in it?
<skyl> comment | text | not null ... in Table "public.threadedcomments_freethreadedcomment"
<bartTC> skyl: it raises a badresponse, I thought you simply add a 400.html template for it but that doesnot work
<skyl> bartTC how can I make it allow long comments?
<jtauber> i think it's a bug / bad implementation in comments
<bartTC> where is the limit, in the model or in the form?
<skyl> bartTC I don't know but it seems not to be limited in the db
<bartTC> btw; this one http://www.mahner.org/posts/django-comments-per-site-caching/ shows you the "white page" as well and nobody knows why :(
<bartTC> skyl: comment = models.TextField(_('comment'), max_length=COMMENT_MAX_LENGTH)
<bartTC> skyl: COMMENT_MAX_LENGTH = getattr(settings,'COMMENT_MAX_LENGTH',3000)
<bartTC> there it is ;)
<skyl> some small security from DoS or something?
<skyl> so I guess I can just remove that max_length and it should be fine ...
<bartTC> yes it did skyl :P
<bartTC> (I mean http://www.mahner.org/posts/django-comments-per-site-caching/#comment1185)
* rizumu has quit ()
<skyl> my comment fields don't have a max_length option defined
<skyl> I do have DEFAULT_MAX_COMMENT_LENGTH = getattr(settings, 'DEFAULT_MAX_COMMENT_LENGTH', 1000)
<bartTC> mmh? do you have a custom comment sytem?
<bartTC> ah ok, I copied that from django 1.1
<skyl> bartTC I have threadedcomments from the social project
<bartTC> err.. ok. Don't use that
<skyl> why not? :)
<skyl> it's a little late for this site ... save yourselves ... just go on without me ....
<jtauber> bartTC: yeah, why not?
<jtauber> it's Pinax's commenting system (although I'm open to change it in 0.9)
<skyl> I rather like the threadededcomments
<bartTC> jtauber: depends what you mean.. the old "good" one or the fucking new one which bases on contrib.comments
<jtauber> which one is in Pinax right now?
<bartTC> I guess the one, which is standalone
<skyl> that's one of the things I mention that are really cool OOTB with pinax .. there's friends and groupings .. and ... they all have threadedcomments!
<jtauber> bartTC: the old "good" one?
So here's my second go at a response. Hope I remember everything I wanted to say...
Really, I just wanted to point out a couple of things.
First, I'd say that SAS and SPSS and other closed-source statistical packages are pretty well peer-reviewed, in terms of use in the scientific community. The only differences between packages, that I've seen, is usability. I imagine the same holds true for R. The stat formulae are plenty transparent, as I've had to sit through a prof writing them out on a blackboard so we could ostensibly understand the math that underlies a t-test, or linear regression, or what have you.
As far as why R hasn't gained any ground in the community, I'd guess it's all due to consensus. Institutions chose to use SAS/SPSS before open-source was really around, and so set up a standard for the research community. Universities teach SAS and SPSS because their students will go on to careers with CDC or NIH and use those same apps. So it all perpetuates itself.
Which is not to defend closed-source stat software. I'd rather see an open-source app, mostly due to the licensing fees. SPSS, for a lowly grad student, runs about 50 USD. If all the calculations are the same, then I'd rather save the 50.
Now, the real question is why CDC uses Epi-Info, an epidemiological app they designed in-house, but which could really benefit from some serious attention. But at least it's free.
$50 doesn't really sound that bad. I thought that SAS was hundreds even for a student.
Well, 50 for SPSS. Not sure about SAS. To circumvent every student having to plunk down half a bill, the university bought a license and installed SPSS on a remote desktop.
