Preprint linking at ICSE 2017

Off

As many of you already know, the International Conference on Software Engineering (ICSE), the flagship conference in our field, started sharing links to the preprints via its programme in 2013.

After ICSE introducing a preprint mechanism (supported by Dirk Beyer) in 2013, Arie van Deursen analysed and reported on how many researchers eventually made use of it. Below are the numbers Arie reported (a detailed discussion can be found here):

Track / Conference #Papers presented #Preprints Percentage
Research Track 85 49 57%
ICSE NIER 31 19 61%
ICSE SEIP 19 6 31%
ICSE Education 13 3 23%
ICSE Tools 16 7 43%
MSR 64 36 56%
Total 228 120 53%

As reported by Arie van Deursen.

That is, roughly 50% of all conference attendees made use of the possibility to share their preprints. As pointed out by Arie, authors might not have been too familiar with the possibilities or they may not have been aware of the benefits of sharing preprints; In fact, there might have been many reasons why nearly half of the papers presented at ICSE 2013 did not include a link to their preprint version. So, one natural question is whether the picture has changed 4 years later…

Preprint sharing at ICSE 2017 – 4 years later

At ICSE 2017, I am co-chairing together with Daniel Graziotin the social networking activities and we both wanted to further foster the idea of preprint sharing via the ICSE programme. Except for the ICSE 2016 edition, all other follow-up editions included links to the preprints on their website. So we wanted to know whether the picture has changed to 2013 and the extent to which the community makes use of the possibility to share their preprints when such mechanisms are established.

To give authors the chance of preparing and sharing their preprints with minimal effort and with time [read: to not approach them while they’re preparing their presentations in last minute], we started collecting the preprint links directly after the notification phase. To this end, we have set up an online (google) spreadsheet, and approached the authors via social media and directly by email via the chairs of the corresponding tracks. We asked them to add their link to their paper in the sheet.

The web chair then collected all the links and added them to the programme. Further, we highlighted the most important information for those maybe not familiar with the possibilities and constraints regarding preprint sharing directly in the spreadsheet.

In the online programme of ICSE 2017, we then shared, same as in 2013, all links to the authors’ preprints, looking like this (on an unrelated note: the last paper is the best one):

 

How many preprint links did we receive? Let’s have a look:

Track / Conference #Papers presented #Preprints Percentage
Research Track* 96 65 68%
ICSE NIER 14 10 71%
ICSE SEIP 32 27 84%
ICSE Education 20 16 80%
SEIS** 12 6 50%
Total 174 124 71%

*) Including invited presentations of journal first articles.
**) Not covered in ICSE 2013; MSR and ICSE Tools not covered by ICSE 2017 preprint initiative.

Overall, we have 71% of the presentations in the main conference coming with a publicly available preprint, which constitutes an increase of 17% compared to ICSE 2013. Not covered here are posters, simply because we approached the authors far too late (my fault) and barely anybody made (respectively could make) use of it. If we compare the number of shared preprints in those tracks in scope of both ICSE editions only (i.e. Research, NIER, SEIP, and Education tracks), then the total increase amounts to 20%. Interestingly, for some reason the number of shared preprint links has increased by 53% for the SEIP track and by 57% for the Education track.  Yet, among all preprints shared, only 6 are shared via arXiv. At least, only 3 preprints made it exclusively to ResearchGate.

What is the overall conclusion?

The overall impression is that there is quite an improvement and this should be reason enough to motivate the organisers of the next ICSE editions to keep sharing preprint links. Although I have to admit that I was a bit more optimistic at the beginning, I am very thankful and also impressed by how well the authors added to our initiative. Needless to say, the number of links collected do not necessarily reflect the actual number of available preprints, because some authors might not have made use of the possibility to share their preprint links with us. In any case, there still seems to be a lack of awareness among the authors regarding open access. This lack of awareness was at least reflected in a few personal conversations with authors who did not disclose their preprints and concerns the possibilities in open access as well as the actual benefits to the research communities (and consequently the authors themselves). A related study on author-based self-archiving awareness, practice, and inhibitors can be found here.

What still strengthens our confidence in a change towards more openness in our field is that we can already observe many initiatives driven by colleagues beyond the ICSE ecosystem. Yet, we still have much to do to increase the awareness in the community and to reach the overall goal of fully open and transparent science ranging, ideally, from open access to open data (and maybe even open reviews?).

I expect ICSE to continue this path in fostering preprint sharing, and I sincerely believe that we should (and can) start taking additional steps here soon – what, for instance, about sharing empirical data? If the major conferences do not set such open science standards, the smaller ones will only be able to do so in homeopathically dosed steps. PROFES’ newly introduced open science practice is such an example.

In any case, I want to conclude with a big shoutout to Sebastian Uchitel, general chair of ICSE 2017, for pushing this initiative and to Leandro Nahabedian, web co-chair, for his work on the online programme. Thanks a lot! As a final note, all links to preprints can be found here.

___

Update 21.05.2017: Added a link to the study by D. Graziotin on Green open access in computer science – an exploratory study on author-based self-archiving awareness, practice, and inhibitors, 2013.