Most top graduate programs require at least 3 recommendation letters. Do students who apply to such programs (and have a reasonable chance to get in) typically have such extensive research experience that they know three professors who can write in detail about their research? Or is it more common for such students to have some recommendation letters from faculty who can confirm that the applicant is competent (for example because they did very well in their class, won an award, ...) but with whom they have not worked much personally?
I understand it would be ideal to get all letters from faculty with whom you have worked closely on a research problem. But I wonder how commonly this actually happens, especially considering the fact that the final undergraduate year typically cannot be taken into account for applications in fall.
If it depends much on the discipline, I'm in physics.