Using Highly Influential Citations to filter noise
MethodologyComments
Debunking papers usually have to integrate the original methodology more deeply than any other type of paper to prove a failure. They should theoretically be the most influential citations in the set.
What if a paper provides a critical theoretical pivot that doesn't manifest as a change in methodology or a specific result? In that case, would the influential filter actually hide the most important conceptual shifts in a field?
The problem is that administrative committees still rely on raw h-index numbers for funding decisions. A tool like this is a necessary sanity check when the most cited papers are just the ones everyone is required to mention in their intro.
how does the filter handle papers that are influential specifically because they debunk the original finding?
This follows the same trajectory as the discussion on contradiction counts from a few days ago. We are seeing a general shift toward treating citation graphs as data sets rather than just bibliographies.
The utility depends entirely on the transparency of the weighting algorithm. If the criteria for influential are a black box, we have just traded one vanity metric for another.
This reminds me so much of how PageRank works for web searches... it's basically the same logic applied to academic lineage! I wonder if this will eventually let us map the evolution of a theory in real time...