From short scenarios to extensive modeling of the Earth’s climate, protein structures or even space, it is difficult to imagine scientific research without software. Scientists use the program code in many ways – plan experiments; record, organize, analyze, visualize And archive data; control scientific tools and much more.
Why it is worth making computational methods easy to use
But the software is developing. Most open source programs, used in studies, are specified both it is well and collectively, and does not have a published “recording version”. Updates can be aimed at various versions and issues, which means that each aspect of software – the project as a whole, a specific version or one file – may require another way to turn to it. This creates confusion.
And therefore, the software is supplied with double connection: as data, it supports the results of the study and must be preserved and published. However, he must also remain affordable and supported, and possibly be improved, After a whileThe field scientists, librarians, research institutions and financial agencies are fighting how to coordinate these two requirements.
Recent efforts about this1 focused on the adaptation of a set of principles originally developed2 do Detected, affordable, accessible, compatible and reusable (fair) reusable (fair)The field but this approach depends on the tracking of the data, their archiving and the provision of metadata. For software, this will create a large administrative load that should have been supported for decades. This would be disproportionately laborious for those who support and improve software, and whose efforts are already underestimated. Imagine that, for example, maintaining a software package in which there are dozens of participants (which is not rare). Each release and version require a new load to the archive, with updates for metadata, the list of authors, dependencies (any other software necessary for launching programs), compatibility (with which other programs can work) and much more. Some programs have a weekly or even daily release cycle, which makes a fair approach impractical.
Researchers should be able to publish a piece of software without the need for long -term bureaucratic procedures to identify the owners of rights, choose an open license and protect intellectual property.
AI 'open source is not really open – this is how researchers can restore the term
As researchers and engineers with experience in the development of software in various scientific fields – from computer science to neurobiology, physics and chemistry – we recently proposed an approach called “Code outside the fair“This describes how software can be better processed, exchanged and supported. Here we describe the recommendations for two groups: scientists who develop software (see additional information (SI), table S1) and research institutes, sponsors, libraries and publishers who use it (see SI, table S2). Our recommendations are based on our joint experience in developing open source software, but also rely on the work of free communities and open source software (FOSS). They have many years of experience in managing projects, financing, recognition of individual deposits and training of future participants.
Teach scientists to share code
The exchange of the code that was used to achieve the conclusions of the article is important for Study of integrity and reproductionBut practice varies greatly among research communities. Permits that allow others to use and change software with a small number of restrictions are increasingly found, Especially in computer science, mathematics and physicsHowever, most of the software has not yet been published at all.
There are platforms for code exchange, such as GitHub or Gitlab, and for its archiving, for example, in Zenodo repositories or Software heritagewhich can capture the entire history of the project3Field
All researchers should know how to share and deposit. This should not mean that all scientists should spend time and resources, aware of this rapidly developing field. The search for the correct level of experience in order to make sure that researchers know how to document, share and archive code in their field is crucial.
Why NASA and federal agencies announce this year of open science
One way to improve questions is to teach all graduate students from all scientific disciplines in the basics of software development during the first year or so in their graduate school. Institutions must introduce this into all scientific curricula. Several universities, including Stanford and Harvard in the United States, Oxford and Cambridge in the United Kingdom, are already offering (and in some cases are required), at least one introductory program or a course for calculations, even for a degree that are not scientific or technical.
There are international educational organizations that teach data and computational skills for a wide audience with basic knowledge or without knowledge of software development. For example, the neuromatcha Academy for global neurobiological education-founder of computing neurobiologists Dan Gudman and Konrad Kerding during pandemia Covid-194 – It was reported that more than 2000 students from more than 100 countries through online learning in 2024. And carpenters, founded in 1998 by Greg Wilson to improve computing skills of researchers5Almost 4800 seminars in more than 70 countries facilitated or organized. These courses vary from basic computing skills (for example, in the shell of programming languages, R and Python and versions control) to expanded concepts (statistics and machine learning), as well as expanded skills specific for discipline.
Increase archiving processes
To increase the introduction of good practices, publishers must exchange code exchange and archives during publication. It is as simple as clicking the button on the programs of the software heritage or GitHub. GitHub, currently owned by Microsoft, has become a facto international center for codes.
Institutions must maintain efforts to connect portals to ensure an adequate cross link between various projects and versions, such as in European open scientific cloud Research support platform.
