In the course of doing some graduate admissions and writing many rec letters, I've been thinking about the value of undergrad research experiences. There is no question that doing one or more reasonably big science research projects can be of real benefit to undergrads in multiple ways. Most importantly, the student gets to see how real research works - it's very different from problem set exercises and canned labs where you know that there's a well-defined answer or solution. The student also gets in-depth experience in a particular subfield, so that they can get a sense of whether that's a specific area they might (or might not) like to study further. In my case, my senior thesis helped me appreciate that I didn't really want to do computational modeling exclusively. It's also good for students to see how much effort really goes into a big project, and gives them experience (ideally) in budgeting their time, planning, making presentations, structuring and writing a lengthy document, etc. We've recently had a really nice insight into some mysterious data coming from an undergrad project in my lab, and it's been very fun to go through the process, with the student, of figuring out what the heck is going on, and to have the student come by my office with the confirming data in-hand.
Sometimes undergrad research can also lead to big scientific results. Here is a paper (see press release) by Dave Hall's group
at Amherst College, where they have used ultracold atoms to create
(effective) magnetic monopoles. Note that this work was done at a
liberal arts college by undergrad researchers. Outstanding!