1/ Lit surveys and discovery exercises serve 2 purposes, not 1: providing idea fuel and containing downside of potential blindside errors
Conversation
Replying to
2/ Conflating the two purposes leads to misery and terrible work because very different subsets of discovery findings drive them.
1
1
Replying to
3/ For idea fuel, you want to throw to choose from the ends of the distribution: most-cited and least-cited.
1
3
Replying to
4/ The most cited work gives you foundations to build on. Obscure works give you novel starting points that have chance for high ROI
4
4
Replying to
... but are then followed by lots of other papers that co-cite the same literatures. 2/
1
1
Replying to
I've sometimes wondered about using this smoking gun to data mine really important papers automatically. 3/
1
1
Replying to
do it! that would be an invaluable service. You could probably sell the product to Google Scholar.
Replying to
A higher opportunity cost-to-benefit ratio than I'd go for (partially 'cos good cross-field citation data isn't so easy to get)
2

