Every time you cite a paper w/o reading it, b/c someone else cited it, a science fairy dies. (A tweet by realscientists.)
The tweet points to the paper, Mother’s Milk, Literature Sleuths, and Science Fairies by Katie Hinde.
Katie encountered an article that offered a model that was right on point for a chapter she was writing. But rather than simply citing that article, Katie started backtracking from that article to the articles it cited. After quite a bit of due diligence, Katie discovered that the cited articles did not make the claims for which they were cited. Not no way, not no how.
Some of the comments to Katie’s post suggest that students in biological sciences should learn from her example.
I would go further than that and say that all students, biological sciences, physical sciences, computer sciences, the humanities, etc., should all learn from Katie’s example.
If you can’t or don’t verify cited work, don’t cite it. (full stop)
I haven’t kept statistics on it but it isn’t uncommon to find citations in computer science work that don’t exist, are cited incorrectly and/or don’t support the claims made for them. Most of the “don’t exist” class appear to be conference papers that weren’t accepted or were never completed. But were cited as “going to appear…”
Someday soon linking of articles will make verification of references much easier than it is today. How will your publications fare on that day?