“Top journals” in my field — and I assume the social sciences in general — have unreasonable expectations that submissions make novel contributions.
One of the harms that doesn’t get enough attention: it encourages authors to inflate their contributions through omissions and misrepresentation of the literature.
We not only incentivize questionable practices, but we make it harder for readers to make sense of how out understanding of phenomena is evolving. [1/3]
Yes, I did the same thing when I was an editor. Novelty is an attractive heuristic for keeping control over the backlog.
Yes, I do think novelty is an appropriate metric for addressing the problem of finely sliced “minimal publishable units.”
No, I don’t have a better proposal. The pressure to publish in journals is too intense, and hence the submission numbers too large, to not appeal to some kind of over-and-above standard. [2/3]
@dhnexon I've become convinced that we (in sociology especially but social science more generally) need to give much more value to replication. Novelty is overrated, especially when it is a third-level interaction effect that has no possible chance of replicating.
I had a paper rejected where we used published approaches on real world data and showed that none of them worked at all.
I found that this advanced the art by showing that a decades worth of literature was not only not reproducible, but wrong.
Reviewers rejected it since the approaches did not show statistically interesting results, which was my point.
This is in computer engineering.
