Shared publicly  - 
 
I'll be sitting on a panel on IP where I want to push the idea that open source and academia can go well together. I'm looking for new and interesting ideas from the world of IP and OSS that can have an impact for academics. What would you like to hear about?

CC-ing my academic or OSS friends (not exhaustive, sorry if I missed you)...
/cc +Daniel Lemire +Michiel van de Panne +Robert Bridson +Kimberly Voll +James O'Brien +Vincent Levesque +Mark Crowley +Andrej Karpathy +Nando de Freitas +Michael Nielsen +Terence Tao +Audrey Watters +Seb Paquet +Michael Welsman-Dinelle +Jocelyn Houle +Nicolas Chapados +Jean-Lou Dupont +Chris Messina +Harald Pehl +Yannis Gonianakis +Thomas Broyer +David Chandler +Mohamed Mansour +Joel Webber +Brett Cannon
2
2
Philippe Beaudoin's profile photoMichael Welsman-Dinelle's profile photoMathieu Daigle's profile photoKimberly Voll's profile photo
17 comments
 
Thanks for reminding me. Exactly the kind of thing I'm trying to collect. :)
 
Let me splurt out various ideas that pass through my mind.

First of all, talking about IP and patent no longer seem to convince anyone anymore (IMO). I feel almost everyone has taken a stance, and few people will argue that the current system serves its original purpose well. Twitter's recent stance sounds to me like "we don't want to participate in the problem", yet doesn't offer a solution, and simply adds to the current inertia (albeit, less than most). So my advice would be to not make patents your main topic, else I fear nothing new will come out of the panel.

On the other hand, you talk about tying OSS with academia (quite relevant to the GRAND organization, I believe), and I think there is some interesting potential...

Academia already benefits immensely from OSS (GCC, UNIX, LaTeX, to name but a few). Also, OSS has in its DNA the very essence of what research is about: sharing! Now, I wouldn't go as far as to tell people of the conference that they've lost their way on that front (I would probably think it somewhat, but not say it!). But I think there is an opportunity to try and show that sharing more research source code is A GOOD THING, and that the benefits of having more people do it is a net gain for knowledge in general, and research and academia in particular. Sharing is the essence of academia...

That said, a good exercise would be to highlight clear examples of how OSS helped research (and vice versa), and while I believe such examples do exist, I have to admit that I do not see any of them in the last 10 years, or so... Actually, the more I think about it, the more depressed I feel. Well, maybe it means you'll need to have a proof by contradiction, i.e. that preventing sharing, through patenting, actually typically stymies good ideas.

In the end, IP is a means to benefit individuals at the expense of the community. And our primitives selves will favor self survival at the expense of others (most of the times). A sad fact, but a fact nonetheless. But I digress (I think).

A while ago, I stumbled upon this video, which might be relevant in this discussion:
La propriété intellectuelle, par Albert Jacquard

It's starts with Stallman (you were talking about OSS), and continues with Albert Jacquart on intellectual property... A good listen (if you can understand French). Jacquart reminds us that an idea is actually inspired and provoked by outside factors...
 
Amazing, Jocelyn! More than I hoped for by sharing this here, thanks a lot for this.

I think everybody will agree that improving and disseminating knowledge is the goal of academia. Here are a couple of points I suspect might come up:
- Sharing too early/openly can mean I will be scooped by a competing researcher.
- Sharing makes it harder to secure a research contract with the industry. (If it's free why pay for it?)
- Sharing can expose your early mistakes and make you look bad.
- My code is disgusting, no way I'll put it up on the net. Keeping it clean is an unneeded cost in academic research.
 
I'm a little confused by your question... it seems like OSS & academia go hand in hand. And there are loads of examples from the maker community and other innovative business models cropping up built on the foundation of open source hardware and software like Kickstarter or Khan Academy.

I guess if you're talking about new/social media, the point (IMO) is to be less about holding on to something tightly, but figuring out how to propagate the kernel of an idea through the network. For example: had I patented the hashtag, do you think it would have caught on? What about BarCamp? Of course the means should fit the desired outcome and blockbusters that are heavily controlled/protected still emerge, but if you're on a budget or looking to influence the world, packaging up ideas for distribution is a saner strategy than trying to fight the Internet. Because you'll lose.

/cc +Noah Diffenbaugh
 
Thanks +Chris Messina. I personally agree that OSS & academia go hand in hand, yet few academics open source their code or publicly share details about their ongoing research. I'm not entirely sure why, so I don't know how to approach the problem, hence my question.

Your answers are very good, and my personal experience echoes them (I would never have had a job at Google if I had not open sourced GWT-Platform). What I'm trying to do is convince academics that this reasoning holds for their situation too.
 
+Philippe Beaudoin Take into account that most Computer Science professors are not programmers. If you think about it for a moment, this explains why they don't see OSS as a natural outcome. I find that it helps to be a programmer who has given OSS a serious run to understand when OSS makes sense.
 
Non-programmers who's primary piece of validation is code (at least in graphics). But I'll grant you they might just not be thinking about OSS. How do I convince them it's worth thinking about?
 
In fact, I wish I could talk to their students. A phd candidate is a programmer and should be pushing his advisor to let him oss his code.
 
I think if you cite Dribbble.com as an emerging "open source-y design community", you might turn some heads.

That and all the design related work showing up on Github (like the three20 framework and others).

Designers are learning to code, and learning to code through open source.
 
Thanks +Chris Messina for cc'ing me into this discussion. I can't speak for other communities, but there is a long history of open source code in the climate modeling community, with many research models freely available on the internet, including the most heavily (taxpayer) supported American models (CESM for global climate; WRF for weather forecasting). This open source culture has created huge scientific benefits, with many more brains contributing to model development, evaluation, and analysis. And there has also been a long history of sharing climate data, which has yielded similar benefits.

+Philippe Beaudoin I would be interested to better understand your observation that most academics don't share details of their ongoing research. Indeed that is the essence of scientific publication, but even prior to publication we in the climate research community also share our results and methods publicly in conferences and workshops (the most significant of which are as open to the public as to the members of the scientific societies). And many of us also spend considerable time engaging with the public about our results and the process of getting there. So I guess I see the whole scientific enterprise - at least in my scientific community - as open source.
 
Hi +Noah Diffenbaugh and thanks for this very valuable input. I'll look into what goes on in climate modeling and will make sure to refer to it in the panel. My own experience, in graphics, is that very little open source code gets released (with a couple of notable exceptions). Yet this is a very code-centric community: almost every paper is accompanied by a software-generated video. The community accepts this video as the proof that the algorithm works, but to me it's obvious that progress would be much faster if the entire source code was released.

Not releasing source has a lot of very visible drawbacks:
- it makes the results harder to replicate for comparison purpose ;
- it makes it hard to really understand the details and limitations of an approach ;
- it fuels cynicism in the community as people question the validity or limitations of a paper ;
- it makes learning harder as students have to rely on opaque papers rather than experimenting on source code ;
- it favors the larger labs that have an extensive code base ;
- it makes it hard to design shared libraries of models or test sets since everybody use in-house formats.
That's just off the top of my head... What I'd like to find are counter-arguments. Why not go open source?

Regarding my observation about "ongoing research not being shared", it was meant as a response to +Chris Messina who proposed finding a way to propagate the kernel of an idea on a social network. This echoes my personal opinion that the entire research process should go on in the open instead of simply publishing a finalized paper. However, I think it's seen as a slightly heretic idea, so I'll keep the lid on it for the panel. :)
 
Partly I wonder how much of this is related to the age of good open source collaboration tools. I like GitHub, for example, the distributed git model, having a "fork" button, etc. but that's all only about 4 years old. In the past sharing code was more painful and it takes time for people to change their habits.
Add a comment...