To the Editor — I would like to reflect on one sentence “if a lab is in competition with many other groups who have similar technical capabilities, they may not want to give up any small advantage that they have” in your editorial entitled ‘A problem shared is a problem halved’ (Nat. Phys. 15, 107; 2019).
I would refine this statement because I believe that the word “similar” is not appropriate. Here in Hungary, science has poor funding with respect to other European countries, even when compared to the country’s gross domestic product (not to mention working conditions, stability and teaching duties). Nevertheless, to publish in a high-impact journal or to get EU funding, we are supposed to compete with large, well-funded and often technology-oriented German and Swiss groups. In these large groups, it only takes a short time to pick up a good idea, put two post-docs and two students on it, and finish off by publishing a number of papers on the subject, gradually forgetting to cite the original work, and then even getting the credit for the technological development. In the case of a code, such as the density matrix renormalization group (DMRG) package developed by Örs Legeza from the Wigner Institute of the Hungarian Academy of Sciences, or the non-Abelian MPO code of my student, Miklós Werner, the development can take 3, 5 or even 10 years of dedicated work. These codes are unique, and they can solve problems that no one else can, at least for a few years.
And this is the only way we can remain in competition. Code sharing would remove the last barriers in this process.
I should add that I am not really against open-source codes. In fact, we did make some of our codes accessible to the public (see, for example, the Budapest DM-NRG code, http://www.phy.bme.hu/~dmnrg/). But, as also explained in the editorial, making a code public such that it is useful is a major effort. In fact, although we planned to initially, we never posted the beta version of this open-access DM-NRG code because we just did not have the human resources for writing the manuals, building the web page and to do all the necessary work to make a code public. And human resources is a pure question of money, according to our experience. Therefore, even on the ‘scientific market’ of open-access codes, being well funded is an extreme advantage.
All in all, I believe that an open-source code policy would work similarly to the free market at the European or US level: the overall ‘production’ may indeed increase but the resulting goods will be distributed quite unevenly, states already in a good position would mostly benefit from it while local ‘production’ in poorer states would suffer and eventually disappear.
I personally believe that code sharing should be the personal decision of those who develop the code.
About this article
Süleyman Demirel Üniversitesi Fen Bilimleri Enstitüsü Dergisi (2019)