Stack Exchange is a pretty popular source of papers I think because we make so much of our data available for free. (Anecdotally, I met with a group from a local university who outright told me that's why they used the network for research.) When people contact us to let us know they are doing research, we've been happy to help. But often people don't contact us and papers get written that don't quite grasp our system. In fairness, it's vastly more complicated than it appears at first glance. In the past I've tried to duplicate results, but often it's not worth the effort. In many cases, the paper isn't intended so much to understand the network as it is to further the authors' academic goals.
This particular case is complicated by the fact that we aren't reading the paper itself, but a summary written by a journalist. The two questions listed were not in the paper (because they weren't asked until 2 days ago), but just examples the journalist picked to illustrate their point. (Their point seems to be that questions on the network are trivial, I think?) Some journalists do a good job writing popular treatments, but others just don't. It's pretty unlikely that a journalist will understand Stack Exchange better than a researcher, however. There aren't nearly enough papers about the network for journalists to specialize on them. Correcting their misunderstandings is unlikely to help them write better articles about the network in the future if they never write about it again.
Ultimately, we can't make misinformation go away. But we can highlight good papers, produce our own research and encourage people familiar with the sites to use our data. There's also a role for our marketing team to promote a more accurate way for people to look at our products as we experiment with Stack Exchange advertising. Still the sites are complex enough you can pretty much count on mistakes and misinformation getting around. You just gotta pick your battles, you know?
I haven't read the paper, but I did find the primary author's dissertation online: "Essays on the Interaction between Users and
Information Systems" by Shun-Yang Lee 2016 - The University of Texas at Austin. The author clearly does understand how Stack Exchange sites work and has some interesting conclusions. For instance, here is an excerpt from "Chapter 4—Is Best Answer Really The Best Answer?"
This implies that the design of answer display rules in current CQA platforms, if relying too much on the question asker’s own answer assessment, might be inadvertently sacrificing quality for politeness, which would negatively impact the community’s collective knowledge building process. As a consequence, even though question askers seem to be the most qualified individuals to determine whether the answers have successfully addressed their questions, the CQA platform should reconsider to what extent question askers’ choices of best answers should be used in determining the quality level of answers so as to improve the effectiveness of the platform as a whole.
To put it in practical terms, the research suggests that answers that use certain linguistic features are more likely to be accepted than answers that don't use those features. Examples of these features, according the the dissertation:
- pronoun usage,
- percentage of articles contained in the text,
- length of the text, and
- number of words with more than six letters.
According to Politeness theory, these linguistic features threaten or save "face", which is the asker's (in this case) public self-image. An asker might accept an answer not because it did the best job of solving their problem, but because the answer was not an affront to their feelings of self-worth.
Or to sum it all up: maybe we shouldn't pin accepted answers.