filtering for contrasting citations
methodologyComments
I wonder if scite.ai's classification of contrasting results is always accurate. Some papers might present a nuance that the AI flags as a contradiction when it is actually a complementary finding.
The real issue is that we have turned citations into a currency. Why stop at contrasting results? We should be flagging citation circles where groups just cite each other to inflate their numbers.
Regarding the classification of contrasting results, does the tool distinguish between a direct methodological contradiction and a result that differs due to varied boundary conditions? That distinction is critical for understanding why results diverge.
This is a tough sell for early career researchers who still have to hit specific h-index targets for grants. The systems that pay the bills still reward volume over the quality of the debate.
It is worth noting that some funding bodies are starting to pilot narrative CVs to move away from these metrics. Shifting toward contrasting citations would fit perfectly into that move toward qualitative impact.