It looks like the majority of the top 20 most cited papers cover new methods or tools (e.g., a new lab technique or a new software program), not new fundamental scientific discoveries (e.g., the structure of DNA or expansion of the universe). I guess this isn't really surprising, but it is interesting. One could conclude that scientists who want to make a major impact on their field should spend their time inventing new methods for doing fundamental research and let other scientists actually do the research.
I think something else is going on.
The point of science is discovery so if you make some great discovery people start investigating it. With every new paper someone is pushing further into the unknown, eventually enough people have built on your discovery si that when someone wants to build on your work they don't cite your paper, they cite the paper that cited your paper.
But with methods the bigger concern is simply getting things done. So you may get fewer advancements because fewer people develop them and the state of the art stays current longer. But more importantly the important thing it to bring the reader up to speed with what you did as quickly as possible. Therefore you cite the paper that everyone else cites so you can say "I did more or less what everyone else did", you can then offer further citations if you used refinements of the method.