kir0ul

SciPy 2023 is looking for reviewers

The turmoil surrounding Elon Musk’s handling of his Twitter takeover has renewed concern over the perils of a public good in private hands (Nature 613, 19–21 (2023); see also Nature 614, 602; 2023). Another form of scholarly public discourse is also owned by profit-driven entities — academic publishers. We propose an answer to both problems.

The most-discussed solution for Twitter is migration to Mastodon (see Science 378, 583–584; 2022), a social-technology platform that communicates over a distributed network of servers (‘instances’ in the ‘Fediverse’), akin to e-mail, and is immune to private takeover. Similarly federated solutions exist for journal articles (B. Brembs et al. Preprint at Zenodo https://doi.org/gn6jjc; 2021), but free social interaction is still hampered by inertia in scholarly organizations — in particular, resistance by scholarly societies that rely heavily on publication income.

There is now a golden opportunity for every scholarly society to implement a Mastodon instance for anyone interested in their field. If the academic community can create a public resource protected from private interests, it could become a model for bringing the remaining scholarly record — encompassing text, data and code — into the Fediverse.

Nature 614, 624 (2023)

doi: https://doi.org/10.1038/d41586-023-00486-3

The US government has issued a new policy that in future, all government funded research will have to be made freely accessible to the public. However, they have not specified how this will be achieved, and publishers are pushing for a model in which the current system continues unchanged except that the authors and institutions pay the publishers rather than readers. This is a form of open access, but the excessively high prices they charge mean that it would exclude many from being able to publish their work in these publishers' journals. In other words, this policy which is supposed to create equitable access would have the unintended consequence of making participation in research itself less equitable.

We are calling on the US government to make sure that their policy is implemented in a way that allows everyone to participate equally in research, not just read it. Since this is likely to shake up the old business models of publishers, we further call on the US government to support or build a publicly funded and freely available publishing infrastructure to guarantee to all the ability to participate in research, and to create the conditions for a lively and innovative ecosystem of new approaches to publishing, better adapted to the modern world.

Sign the letter now, share this article, or read on for more details and references.

Nextcloud is currently used by many public services in Europe (universities, governments, cities etc.) as an alternative to Big Tech solutions, keeping their documents and communication safe in Europe. The European Commission’s Open Source Programme Office is organising a hackathon to enhance Nextcloud with additional features so that more public administrations can include such solutions as part of their digitalisation journey.

Help us build and protect Europe’s digital sovereignty!

There is no shortage of package managers. Each tool makes its own set of tradeoffs regarding speed, ease of use, customizability, and reproducibility. Guix occupies a sweet spot, providing reproducibility by design as pioneered by Nix, package customization à la Spack from the command line, the ability to create container images without hassle, and more.

Beyond the “feature matrix” of the tools themselves, a topic that is often overlooked is packages—or rather, what’s inside of them. Chances are that a given package may be installed using any of the many tools at your disposal. But are you really getting the same thing regardless of the tool you are using? The answer is “no”, contrary to what one might think. The author realized this very acutely while fearlessly attempting to package the PyTorch machine learning framework for Guix.

This post is about the journey packaging PyTorch the Guix way, the rationale, a glimpse at what other PyTorch packages out there look like, and conclusions we can draw for high-performance computing and scientific workflows.