IRC channel logs


back to list of logs

<zimoun>I just got a paper submission rejected.
<zimoun>So a bit bitter. :-)
<zimoun>What I find interesting is that one component about replication crisis is because software.
<zimoun>Even, many «The Software Pillar of Open Science» is poping everywhere.
<zimoun>So, when you explain to biologists why software is vitally important, using their words and frames targetting a generalist publication, then you get rejected.
<zimoun>“Too specialized”, sic.
<zimoun>On one hand, biologists are saying “we have issues about software and reproducibility” and on the other hand they are also saying “we are not interested to understand why and how to fix”.
<zimoun>One of the issue never (or barely) discussed is that the publications are now niche, where the authors write to people already aware.
<zimoun> Depressing circle!
<efraim>yikes, that sounds really frustrating
<rekado_>I wonder if in the same journal there are articles casually mentioning Docker or Conda for reproducibility.
<rekado_>maybe it just takes more normalization of mentioning the software infrastructure
<zimoun>we are submitting to another “generalist” journal. Well, this process and the replies seem interesting by themselves. :-)
<rekado_>we are similarly stuck with EU funding applications
<rekado_>we have a contact for “pre-screening” and he told us outright that software reproducibility is a solved problem (“just list the version numbers, duh!”) and our application is too specialized
<civodul>hey there!
<civodul>zimoun: i sympathize with the frustration! (and with the enthusiasm for reproducibility :-))
<civodul>i feel like there's a window these days for explaining our vision
<civodul>there's a lot of hype around the empy "open science" word, but there's also honest interest in understanding failures and improving things
<zimoun>maybe software reproducibility is a solved problem (or let assume that), but the implementation is not solved. One of the key point is teaching what is the problem and how to solve it. Because journals are niche, where does this “teach” happen?
<rekado_>not there
<rekado_>conferences maybe?
<rekado_>but they are also self-selecting
<civodul>yes, or dedicated workshops like the one rekado_ & i went to in 2015 (time flies!)
<civodul>i think that's how it works in general: "fresh ideas" first mature in workshops and eventually become commonplace
<zimoun>for instance, we with MDs wrote a short stuff targetting broad audience of other MDs or impatient biologists about “what is a computational environment”. It is a pillar but never in the niche of the journal. Or it can be published in a specialized journal focusing this topic but not read by MDs or impatient biologists. Therefre, how to speak to this broad audience?
<civodul>yeah, i don't know the answer
<civodul>there's probably a fraction of this audience that's more curious and willing to listen to what you have to say
<civodul>and those in turn will spread the word among their peers
<civodul>but... these are generalities that probably don't help you much :-)
<zimoun>yeah for sure. The co-authors are these people. ;-)
<zimoun>Well, I think the process and replies are intersting by themselves. Let see the next feedback. :-)
<zimoun>bah, usually, it is easier to attract attention with stuff producing nice pictures about cool topics than plumbing details about boring topics.
<zimoun>Maybe it is a corollary of ? ;-)
<civodul>i hear that from researchers working in linear algebra
<civodul>but i have mixed feelings
<civodul>sooner or later, it had to happen
<zimoun>civodul: ahah! let mimick ’module’ cli. :-) Now, I understand your question from yesterday. ;-)
<zimoun>do you have examples about arguments by «researchers working in linear algebra» for ’module’?
<zimoun>because IIRC, it is only by habits. Typing “module load foo” instead of “guix shell foo” (or similar).
<zimoun>In 2019, to ease to teach Guix to some users here, I planned to write a transparent “wrapper“: type ‘module load foo’ but under the hood run ‘guix environment --ad-hoc foo’ but extensions was not around. And then I thought it was useless. :-)
<zimoun>civodul, well I do not know if you will read that. The ’module’ file points to a store item. Is this item protected from GC?
***ChanServ sets mode: +o civodul
***civodul changes topic to 'GNU Guix in high-performance computing | this channel, it is logged: | activity report: | events: |'
***ChanServ sets mode: -o civodul
<civodul>zimoun: re "modules", i still need to add GC root registration indeed
<zimoun>civodul: do not. User of modules should have the same experience as their usual ‘mdoules’ habits: broken. :-P