So, if you are publishing research articles that use computer programs, if you want to claim that you are engaging in science, the programs are in your possession and you will not release them then I would not regard you as a scientist; I would also regard any papers based on the software as null and void.
Wednesday, February 10, 2010
Sounds right to me ...
... If you're going to do good science, release the computer code too. (Hat tip, Dave Lull.)
Subscribe to:
Post Comments (Atom)
As long as the math is transparent there is no reason to divulge the proprietary computer model coding and undermine their value to the developer — the coding is just a shorthand for a complex set of data-based equations that could be done by hand (with some enormous effort; but, you know, tough luck on that front). Does it have to be correct? Yes, that's what peer review is all about. Should he then give it away? Not sure about that.
ReplyDeleteUnless you would want to extend this to all fields and essentially remove copyright coverage from scientific research (which is an idea I hear floated around from time to time), this seems like a double standard based in the politicization of the issue at hand. Pharma and medical companies don't release their proprietary findings either.
Dear Sir,
ReplyDeleteI was going to post something similar to the previous commenter. What is copyrightable about a computer program is the code. If the program is freeware or some shared distributed ware, then distribution of your variant of the code should be something you would do naturally. But it the code is proprietary--expensive to buy, build, or develop--then publication of the code poses certain problems.
Not all code falls under the latter case, but it is perhaps too simple a thing to say the code should be published. What should be published and clearly indicated is the route of analysis, the equations and formulae used, how these are interrelated (where does the output of one formula enter another formula or round of analysis. I think it is in these parameters that scientists are too often coy and unwilling to share the information that is vitally necessary to assess the validity of the analysis.
But it is a difficult and touchy issue.
shalom,
Steven
Well, I think we can agree that, as Steven says, it is "a difficult and touchy issue." The Hadlet CRU email titled README HARRY does indicate that a lot of influential "science" seems to have been based on a very dubious bit of programming. If you are going to tell us "the science is settled" based on stuff like that you have a problem. BTW, Daniel, when you say that Pharma and medical companies don't release their proprietary finds are we talking about just their programming or more? Their results do undergo a good deal of review, presuming that the FDA is doing its job.
ReplyDeleteThe results of clinical trials and review do go through the FDA; the proprietary aspect of drugs, being the process by which the compound is produced and refined, stays close to the chest of those companies. The FDA is essentially the peer-reviewing organ that makes sure something is proven safe enough for public sale and consumption: other fields have similar processes.
ReplyDeleteI'm not sure what this article is really asserting — for instance, if a program hits an "invalid" string, by poor design or incorrect input, then the program should just stop running. It isn't like a human, who makes a calculating mistake. Maybe the models are poorly constructed or the degree of variance too broad; it isn't "not science" for these problems necessarily, it just becomes the parameter of certainty for the results output.
It is difficult and to the public it sounds — well, people generally don't want to hear "degrees of certainty" because they're used to hearing "Laws". Science is a process of refining the certainty of a model's results tho, and good science is about the process, not the results.
Well, Daniel, I pretty much agree, and that appears to be what Harry, whoever he is, was saying in the Hadley CRU email: This program is crap.
ReplyDeleteDear Frank,
ReplyDeleteI cannot disagree with the analysis that says that settled science is based on dubious analysis, and for that matter, dubious premises. As I have often asserted, given that we are still coming off of some of the coldest times in recent history (the year 1666 and surrounding is said to have marked a little ice age), it is no great wonder that we are gradually warming. Given also that there are vast swathes of Earth History during which there were no polar ice caps, I'd say that if we were warming up we were entering a different stage of "normal" for the planet. So, the whole thing has always struck me as a little odd. Additionally, as I think I've asserted, it makes what ought to be a moral and ethical issue into a pragmatic one (admittedly, this is sometimes the only thing stubborn humans will respond to). But the question is should we continue unnecessarily to pollute our environment if means exist to reduce the impact and make things better for future generations. There is no amount of climate research, good or bad, that will give the answer to that question--which while simply stated is infinitely nuanced.
So, we are in accord--the settled science is based on dubious analysis. The computer code on which it was based has been found faulty by some (and of course, their own analysis of what is happening should also be examined to see whether or not they hold water) but the question of wholesale issuance of proprietary and copyrighted material is a somewhat different issue and one that could, understandably, take in many viewpoints. For the most part, I think that if you want to be taken seriously you should detail your method of analysis and if there is nothing to prevent you from doing so should present the actual code for analysis. However, I wouldn't hold to this as an invariable principle simply because there are a great many things that may influence it.
But we concur on the central point--the science is poorly wrought, poorly conceived, in my opinion, not science at all--but just so much agenda.
Thanks for providing this interesting forum for discussion of what may be one of the more important issues in scientific ethics of our day.
shalom,
Steven
Chris Wiggins also thinks code should be shared:
ReplyDeletehttp://www.law.yale.edu/images/ISP/Chris_Wiggins.pdf