Guillaume Dumas asked a few questions to Marius Buliga, a researcher who started doing “open science” even before the term was actually coined.
1. Hi Marius, can you present yourself and what you are studying?
Hi, I’m a wandering geometer by heart, so I move from a problem to another driven by the feeling that there is a hidden geometric bit which needs to be revealed. My phd is on the Mumford-Shah functional which can be used to find edges in images and also to model the appearance and propagations of fractures. Nonlinear elasticity (think about big deformations, like in rubber) is about minimizing energy functionals over infinite dimensional groups. Convex analysis and hamiltonian mechanics can be mixed into the study of non-smooth, dissipative systems. Sub-riemannian geometry is actually only a model for a non-commutative differential calculus, not enough studied yet (despite fundamental contributions from great mathematicians like Gromov, Mostow, Margulis), where the space, even at a local level, even if trivial from a topological point of view, has a fractal behavior which is smooth in an unusual sense.
Now I’m interested in distributed, decentralized computing, biology and artificial life. It’s related because if you try to put back geometry into computation (and not trying to eliminate it with one dimensional thinking) then you’ll get models which resemble a lot both molecular computing and decentralized networks. The artificial chemistry chemlambda is my latest toy.
2. What is open science for you?
I got hooked by the web in 1994, at Ecole Polytechnique, Paris. The Mosaic browser, the possibilities… At that point I understood that we pass through a revolution like the one started by the invention of the press, but much, much bigger and faster. Everybody knows this, or it should. (Who doesn’t? Well, all these people from academia who take disastrous decisions for their students, are hand in hand with legacy publishers and eventually will go into a yummy retirement, leaving the academia in a far worse state than decades ago.)
It is natural then to try to do open science. For me this means that I put almost everything I do online, with the hope that if there is something valuable for anybody else, then this is the best strategy.
Science does not happen in isolation, but it is (it was) hard to pass the barriers of time and space. Books are wonderful means to discuss with people who lived long ago or who are not even born. But this is a slow channel. Moreover, it became very noisy, due to legacy publishers and their minions from the academia. The geographic neighborhood of the researcher was enlarged by travels and correspondence by letters, in a small, not very well connected world.
With the means of the web, communication of science exploded. Open science is a matter of evolution of ideas. I can’t stop to think about it as a mating strategy. What is best to propagate one’s genetic traits? An open behavior or an ornate and arthritic mating ritual, performed in secrecy inside a very small community? There is no question which is better.
3. You have put all your research online. Can you tell us more about the open notebooks?
I have most of my articles on arXiv.org. This is the NASA of open science, old technology but still better than anything else, for the moment. I used figshare.com. Of course, there is my homepage, which exists since 1995. Then I have an open notebook at chorasimilarity.wordpress.com. There are more than 500 posts there. From these, about 300 are mostly research notes, some of them transformed into articles, but there is a lot there. Many other posts are about OA.
More important, I have a recent GitHub account. That’s the best variant, because it allows to play with validation as an alternative to peer-review.
4. How did you decide to do put all your research online, even before publication?
I believed in the web from the first moment, as I wrote before. Besides this, it was a gradual process. I use(d) to write articles which mix several fields and I was told, repeatedly, that such articles are difficult to publish. I have articles which took years to get published (my record is 17 years) Practically, I got again and again into the same problem: by the time the article is accepted for publication I have a secret backlog of many better results and moreover I am already bored by the subject and looking for a new one. The arXiv.org variant solved only partially this problem. I still have many (physical) notebooks which have not turned into arXiv articles because I can’t stand to explain once more why and how and what about various subjects.
But for some time the arXiv was the solution for me to avoid the years lost into waiting for the rare (I wonder why so rare) reviewer who is willing to spend some time with my article. I just put them in the arXiv and moved along.
Then, in 2010 I had very strong feelings that I should go much more open, regardless the outcome. I was visiting Rio de Janeiro, a great city, and perhaps the amazing exuberance of life there made me start to try to move towards computation and biology, from a mathematical point of view. So, when back in Romania I opened the chorasimilarity.wordpress.com and entered into disputes about OA.
More recently, by a slow process I understood that programs are better (more rigorous) than proofs, another idea that everybody heard about but it is hard to internalize. So now I use GitHub as a validation means for my research and I write a lot about it on Google+.
5. Did your openness have any consequences on your career (good or bad)?
There have been good and bad consequences. Good, because all the people I met, scientifically, are among those contacted via the web. Bad, because a lot of my work does not appear on the bureaucratic radar.
6. When we first interact, it was about peer review. You mentioned an alternative or complementary concept: validation. Can you explain what is it and why it could be better than traditional peer review?
Yes, validation is one of those ideas which floats in the air.
People from biology, medicine, neuroscience, etc, remarked that despite the avalanche of peer-reviewed research, a significant proportion cannot be validated by other researchers, independently.
Validation, as I see it, means that the researcher has to provide to the reader as much as possible so that the reader can validate (by reproduction, or by reasoning, or by any other means the reader has) the research.
Peer-review is a social validation, from the point of view of the reader it looks like this: hm, I’m reading an article in the journal. That means that the editor interacted with some reviewers who told the editor that the article is good for publication. I don’t know why the reviewer thought so, nor how the reviewer arrived to this conclusion. A good guess is that the reviewer arrived to this conclusion by reading the words and nothing more deep than this. The reviewer had the same means as the reader of the article to arrive to a conclusion, with the implication that the reader has access to the article with the prestige attached by the appearance in the journal, only because another reader recommended the article for publication. Otherwise said, if the prestige stamp of the journal is ignored, the reader has exactly the same means to make sense of the article as the reviewer had. The role of peer review is to suspend the disbelief of the reader because there has been another reader before.
This is a very weird practice, right? We can enhance it by making open, perpetual, pre- and post- publication peer-reviews. This way, if it works, we might see the reviews and get input from other people impressions about the article.
But this is not solving the validation side. What are the means that anybody, first reader (the reviewer) or second reader, has to validate it? What if the reader could replay the research process, instead of just reading some words which describe it?
From here the idea of an article which runs in the browser. Suppose that you could play with the research, as you read the article. Suppose you have access to as much as possible data about this research. Then, from the point of view of the reader, this would be great. It does not exclude peer-review, it provides instead a more rigorous fundament for the opinions expressed through peer review.
This is a new form of research communication, a sort of super-article. Easy to say, but not at all clear what this means. There are proposals, one which I like is the PeerJ-paper now, but there are others as well. Open notebooks, why not? My version is simply a github.io collection of pages, with all means to run the programs instead of reading the proofs.
7. To finish, what would be the most important advice(s) you will give to young researcher and students interested in becoming scientists?
Think about the future. If you want to do research then do it.
Nobody dreamt, as a child, to spend life writing N articles/year in journals with impact greater than p.
Really, this is reserved to people who use research as a means towards something else. If you want to do research then do it, inside or outside academia. You are important. You have a dream, do it.
Thanks a lot Marius!
More information about Marius work:
- Marius’ homepage: http://imar.ro/~mbuliga/
- his arXiv.org articles: http://arxiv.org/a/buliga_m_1.
- the chemlambda project https://github.com/chorasimila
- an example of self-executing article which runs in the browser: http://chorasimilarity.github.