Shared publicly  - 
4
5
stavros vagionitis's profile photoDavid Mertz's profile photoTina Caton's profile photoMary Paniscus's profile photo
11 comments
 
+Lisa I. Smith Absolutely. That's one reason why I'm so skeptical of some of these climate scientists. They even hobble FOIA requests, much less make their computer code and datasets availalble.
 
I suspect that the reason those scientists were unhappy at the FOI requests was because they felt that everything would be taken, changed, misrepresented to 'prove' the opposite of the facts.

I believe that the headline is right. Without OSS, nothing is secure or certain.
 
I do not disagree with the linked editorial. After all, I am moderately well-known as a FLOSS proponent, in several roles. However, I feel like there is less at stake -- at least in many cases -- than the editorial suggests. Quite a lot of science is done with very specialized equipment, or in specialized conditions. There is only one LHC, and only a small number of large telescopes, for example. If LHC releases all the code the makes the machine run (and all the code used in analyzing the data) that still doesn't let anyone without access to the machine replicate the results, at least not directly.

I happen to work at a company called D.E.Shaw Research that has made a specialized supercomputer for performing molecular dynamics, called Anton. It's cool and amazing work, and the code that runs the machine is strictly in-house, even though results based on simulations on the machine are published in the open literature. And frankly, I see nothing wrong with this. No one without this very special hardware could even run the software, so there is nothing to "replicate" at that level. On the other hand, the company also publishes as "open source" (maybe not an OSI compatible license, since it's for non-commercial use only; commercial licenses that include source can be obtained for money) the code for a software-only MD system called Desmond (that runs on commodity machines). We also do a good job of publishing things like trajectory sets of simulations that appear in the literature, so that users of other MD systems can run the same systems, or scientist can analyze the trajectory data using other tools (most of which are open source tools, to my knowledge). We also release various ancillary tools and libraries under liberal terms that might be useful for the MD community (or for general computing, in some cases).

In general, when science relies on specialized machines/detectors, I'm not persuaded that the code that runs the detectors themselves really needs to be OSS. Software for analysis of data, certainly should be, but "embedded code" for devices feels like a different category, but certainly makes up a lot of the code used in science. Even for commodity instruments, I think embedded code is rarely released, nor needs to be, since in those cases the proper validation comes in having other instruments (from other makers) produce compatible results when measuring the same aspect of reality.
 
+David Mertz What you describe seems like a legitimate edge case. I'm sure there are others. But where "science" is responsible for the expenditure of $Billions by governments etc. acting on good faith that the body of evidence presented to them is accurate and believable...
 
+Peter Bromberg Actually, I'd certainly support a greater source code disclosure standard for government funded research, as a general matter of public access and public funding. The principle of open access seems obviously correct, and the threats against it from folks like the so-called "Partnership for Research Integrity in Science" and Elsevier (in the form of the "Research Works Act" recently, but as a general matter) are perversions of science and of funding integrity. And also, ideally, an open source requirement for code produced under government funding of scientific research (even the code that goes into specialized devices if they are government developed/funded).

However, given your earlier comment in this thread that seems to have some climate-change denialist conspiracy nonsense in it, I have the feeling that you have some whole other agenda that is unrelated to the specifics of publishing the source code related to scientific research. The somewhat crazy accusations made by the denialists have never had anything to do with source code, at least not that I've ever seen... nor even with data sets, which indeed seem to me to fall under Open Access principles. The conspiracy stuff mostly seems to be about the status of personal email and the like, and there the issues seem to have little to do with either FLOSS or Open Access. I do think "sunshine laws" (including the FOIA) should apply somewhat to those sorts of things, but certain limitations seem reasonable also, depending on the nature of communications; in any case this is a distantly related question to the one in the linked article.
 
+David Mertz My first comment was in response to +Peter Bromberg. Why a scientist might "hobble" FOI requests. It is not because they have anything against the openness of information. It was a very case specific objection. IN the long term, they had no alternative to make information available whether they liked the requesters or not. That is a good thing.

Anything someone wants to claim as a fact needs proof. The article mentioned the supposedly FTL neutrinos but, there will be much more prosaic things such as MPs expenses that need proof.

There are many things that cannot be proved to people. The public relies on others to verify things. The readership of publications like Nature is very small but their influence is large. If things are not proved to the satisfaction of scientists and engineers, then the alternatives of big money, politics and hearsay take over again.
 
+Scott Watson I am sure that is the basis of much of the resistance to FOIA requests. However, if a scientist has a testable theory and good data, they should be a lot less inclined to resist allowing others to verify their assumptions, data, and results.
 
+Scott Watson I've read all that stuff. Personally, I think much of it is baloney. The number of accurate reports about fudged data, hockey stick charts and the like is simply too overwhelming. If somebody makes a mistake with your data, you can explain to them what they did wrong - that's part of the scientific process. The IPCC's computer models can't even reproduce historical data, much less make accurate forecasts.
 
If I have a choice between most scientists and corporately funded 'studies' I will trend to believe the scientists.
Add a comment...