Say yes to open-access + peer-review

The only model for peer-review which makes sense: the “No free view? No review!” signature campaign and why the existing publishing system needs an overhaul

There are fundamentally several problems with the current publishing model of scientific articles:

  • Authors toil and work polishing nearly every word of the article
  • Reviewers volunteer out of goodwill or to keep in touch with research
  • Editors coordinate the whole effort
  • Libraries or businesses pay to get access to and distribute the end result

And no-one above in this process gets paid. High charges do not make sense in the electronic age, when fewer articles get actually printed into hard copy. Even today with this model, as an author, you will have to pay extra to put a colour figure which would not be legible in grey scale.

Here is a documentary describing why this system is ridiculous at best and predatory at worst.

A paradigm shift

Plan-S and Bibsam consortium made a huge dent in this process. This movement gave libraries and universities the ability and courage to collectively bargain and walk out of the current subscription model with exorbitant prices. This was a top-down change and resulted in new deals being made to:

  • benefit authors to make new articles open-access without the authors paying a hefty Application Processing Charge (APC), but rather the university as a whole
  • make certain journals open-access or hybrid

What we need now is a grassroots movement, a bottoms-up approach. As authors, we should be willing to share our research at different stages to preprint servers (for example, arXiv) and similar outlets (Zenodo, Figshare etc.). The Sherpa-Romeo portal is my go-to tool in finding out what kind of article can be archived for public-consumption. Also, check with your library to find out if they could cover the APC charges.

And as potential reviewers, this campaign is relevant:

No free view? No review!

I think it is our turn as researchers and academics to say no to review articles which would end up behind a paywall. It is simply not ethical in today’s world.

Lightweight classes for scientific computing using "cached_property"

In the quest for optimization and fast computation, while using object-oriented programming, it is a typical technique to initialize the linear algebra array operators as class attributes. Especially in fluidfft we encounter pseudospectral operator classes that look like:

import numpy as np
          from fluiddyn.util import mpi

          class Operator:
              def __init__