mastodon 4.7.3Researcher at CNRS. Computational science, in particular computational biophysics. Metascience, in particular the evolution of science in the digital era.
Favorite hash tags:
#DoughnutAcademia
#Emacs
#GlamorousToolkit
#Guix
#Pharo
#CommonLisp
#ReproducibleResearch
Lettre ouverte : Arrêtons l’adoption naïve des technologies de l’IA dans le milieu universitaire
https://aeporreca.org/blog/lettre-ouverte-adoption-ia/
Traduction française d'une lettre ouverte internationale (par @olivia@scholar.social et al.) que vous pouvez signer dans l'original en anglais ici;
All I knew about Zig before reading this post is that it's a programming language in the C/C++/Rust subspace. I have no opinion at all about the language. But this post almost makes me want to use it, just because of its credible long-time plan to become and remain community-managed technology.
Interesting remark by @k9ox@mastodon.radio in the FedWiki Matrix room: "I don't mind wiki being unreliable but I want to know why it isn't."
This looks like a good framing for reproducibility issues in science as well. If something is not reproducible but it is generally understood why, that's OK - not nice, perhaps, but it's possible to work around it. But if you can't reproduce a result that everybody believes to be reproducible, that's a problem.
You can now sign up for the next session of the MOOC
"Reproducible Research II: Practices and tools for managing computations and data"
which opens on May 5th.
https://www.fun-mooc.fr/en/courses/reproducible-research-ii-practices-and-tools-for-managing-comput/
"Hallucinated citations are polluting the scientific literature. What can be done?"
http://doi.org/10.1038/d41586-026-00969-z
After lots of technicalities:
"In some cases, hallucinated citations might serve as a sign that the entire paper is fabricated, whether by an individual or a paper mill, a business that sells authorship slots. But the extent to which fabricated papers are contributing to the fake-citation problem is unclear."
"Even more troublingly, other researchers say, the fake papers were then cited in peer-reviewed literature. Osmanovic Thunström says this suggests that some researchers are relying on AI-generated references without reading the underlying papers."
Does this still surprise anyone?
An interesting view on software use in science. What this means is
- gain in efficiency: less software to maintain, less friction in collaborations, etc.
- fewer but more critical points of failure - potentially offset by improvements in funding and governance, but I am probably dreaming
- reduced epistemic diversity: the chosen few software packages determine which scientific methods become consensual
@smallcircles I have made various attempts in the past to explain to FLOSS communities that they need to have at least their users implied in their governance, and ideally even more distant stakeholders. With little success.
The typical reaction is one of
Users get our work for free, they have no right to complain.
Users can join our community, e.g. by contributing to documentation, and then they have a say as well.
Formation "Construction et animation d'une communauté autour du développement d'un logiciel libre"
30 juin - 2 juillet à Autrans
La formation est ouverte et gratuite à tous les membres des laboratoires de recherche académique en France.
"Reliable research in the social and behavioural and sciences"
A collection of four (paywalled) articles on reproducibility and replicability in the April 2 issue of Nature: