In addition to open access

Overlay journals¶

One issue for young tenure-track researchers is the necessity to publish in high-impact-factor journals. Most of them were started in pre-Internet times, and are still owned by rent-making publishers. To avoid this, I believe we should move our quality works to overlay journals, also known as "reviewing entities" or "Peer Community in", where peer reviewing happens on pre-print repositories like arXiv or HAL. We can still put out pre-prints first, as first-mover advantage is a thing, and then iterate on them in the open, taking into account feedback from colleagues and reviewers. The record of these iterations can also become valuable information for newcomers who face knowledge gaps.

Post-prints¶

Since the beginning of my graduate studies I have fortunately been able to submit pre-prints of all my works, focussing the reviewing process on improvements by decoupling it from the ability to share my works. The habit even evolved into post-prints, where I keep updating manuscripts after their publication, because I noticed a second wave of meaningful feedback starts flowing around 2–3 years after publication. Here is a quick model:

• The first wave is peer review, it filters mistakes or issues that appear in reading. It is mandatory and requires a significant investment over a short time span.
• The second wave comes from peers trying to reproduce the work. It can question design choices and spin exciting questions. You need to maintain an active online presence to benefit from it.

I have been very happy with the interactions that came from maintaining both post-prints and an active online presence. Ideas take time to spread: putting out a manuscript or a proper code distribution are only the first steps of a meaningful journey.

Code distribution¶

As academics, we want our works to be reproduced not only by experts, who are already advanced in their various paths of knowledge, but also by newcomers eager to climb up their own paths. For them, published papers ripe with field idiosyncrasies (which come naturally from compression to a fixed number of pages) are not an efficient tool. Another channel we can cultivate to fill this gap is the source code we produce. I believe that, luckily, interests in doing so are aligned with career incentives, as quality source code is correlated to peers citing a work. (It has been clearly the case in my experience.) Here are some levels of increasing source-code value:

• All dependencies are open source software.
• The source code is packaged, for instance to conda-forge.
• A one-line procedure allows to run code related to the paper.

By distributing usable code, we help ensure knowledge gaps can be crossed by newcomers who take the time to work them out. Papers cannot be fully detailed on every point, but compilable source code has to.