Since at least the early 1990s, many universities have  recognized the value in involving undergraduates in research. At my university,  several provosts and presidents have praised my physics department, because we  have several active research programs that involve many students, despite our being  a small department. I'll admit that involvement in research can benefit  undergraduates: it make all the difference we me, as an undergraduate. Research gives undergraduates face-to-face interaction with faculty, useful for  getting letters of recommendation. It also gives them practical experience and  skills, also useful for finding jobs upon graduation. Perhaps most importantly,  it inspires students to dream, by showing them how what they're learning in  class is useful.
 
Nevertheless, as Peter Feibelman points out in "A Ph.D. Is  Not Enough," "Only some of your graduate students will really contribute to  your research. Others will break your equipment, contaminate your samples, and  install bugs in your computer programs." Undergraduates are harder to mentor in  research, since they know less. 
It's even harder at my less-selective university. Our best  graduating seniors have a tough time scoring above the 50th  percentile on the GRE physics exam, the minimum that many R1 universities  consider for admission to their graduate programs. My best student in the past  15 years scored in the 65th percentile, which got her into a good,  but not great program. Some of our students score in the single digits, but the  most recent student who did so is now happily developing automation systems for an aerospace company.
Still, mentoring many of our more typical students has  taught me a lot. It includes:
- Never once have I had a research student who pulled their own weight. Every last one of them, even the best I get, has been more work  for me to mentor them to do a research project than just doing the project  myself. It's not unusual for this ratio to exceed 5 to 1, or even 10 to 1. It's been a while since I've had a student who achieved nothing at all: maybe it indicates I've been getting better at mentoring, since instituting weekly research meetings as well as weekly journal club with all the students working for me together. Still, for cases like these, I suppose the ratio would be infinity to 1.
- Only very rare cases can program a computer, in any  language, at all. This means that only some projects are doable.  We can analyze samples of 100 or even 1,000  objects, but the latest, computer-generated samples of 60,000 are  out of the question. My only hope is that R1 researchers will see our papers,  wonder why we can't do what we really should be doing, and then do it  themselves, with their superior resources.
- I give my students lots of credit for writing things  that I mostly wrote. It's not unusual for an M.S. thesis I've mentored to be  about 80% written by me. I was more idealistic with my first M.S. student,  holding him to the standard I was held to in grad school, and the resulting  draft was embarrassing: it took him two semesters to do something a colleague  of mine at Yale figures out in 15 minutes. So, before submitting the thesis formally,I redid a lot of the writing, and  I continue to do lots of writing for my other students. I do have them write a  preliminary draft, but it's rare for any page of it not to require substantial  revisions, by me. I wouldn't be so generous with my students if I didn't have  tenure.
- Every university administrator of course wants me to  bring in as much external funding as possible. I'd admit it can be handy for  running a research program: who do you think pays for all the travel around the world, to get data, and to conferences, to show off results? Still, my conscience bothers me sometimes.  Frankly, I don't think involving students of this caliber  in research is a very effective use of taxpayers' dollars. This is why,  whenever a student asks to do a research project with me, during the first  semester I always pay them with academic credit, not from a grant. Students who  do work out and get to be paid from grants nevertheless still need to be  carefully watched. I've also had no shortage of students who are dishonest with filling  out their time sheets, by claiming they'd put in substantial hours immediately after telling me they have nothing to report.
- Projects can get dragged out for many years. It took me  eight years to get three so-so students to complete a project that should have  taken a good one, or me, three years to complete. Still, I did publish it in a refereed journal, with all the students listed as co-authors, much to the pleasure of my administration, and  all these students went on to good jobs, two in nuclear power and another for the  government.
- Never have even my best students put heart and soul into  research, the way I was when I was an undergrad. What they give me is almost  always not that much different from term papers by undergraduate non-majors in my  general-ed class: really, what they think is the least they can get by with. 
- Some of my students have subsequently gone to grad programs, usually in low-tier programs. What usually happens is that they get used for a year or two as cheap labor to run labs, and then bounced out by qualifying exams. They subsequently go on to a not-bad record of getting jobs in K-14 teaching. 
- My best students always  seem to be busy with something other than what I have for them, such as  maintaining their GPAs. When these same students take my classes, they get  `A's, but again, only by doing the absolute least they need for it. It's never anything near 100%, in the way I often impressed my elders since I was in Kindergarten. I was often asked, "How come you know all this?" I'd answer, "Reading."
- There are many projects that students can't help with at  all. Some of these are left over from my postdoc days, receding into the  ever-more distant past.  I hope I can  turn them into papers published in refereed journals during an upcoming sabbatical, but some of them are  getting old enough to worry me whether the science won't have become beside the  point.
- Half-baked ideas about what specific research to do, and  how exactly to do it, won't work with my students. A symptom of having been helicopter parented  is when students expect you to micromanage them. For such students to be  productive at all, I have little choice but to micromanage them: refusing to do so and  insisting it's "not my problem: do your own homework," the way I was told when I was a student, simply will not work. Every time I have I mentored exceeding mediocre students in research, they always get tangled in something,  requiring me to cut them loose. To be fair, students can't read your mind.  You will need to have something specific planned, since expecting students to  have ideas, use their imaginations, take the initiative, read the manual, think  creatively, etc., the way Henry Moseley did for Rutherford, the way good  scientists do when making good scientific progress, is expecting a bit much.  No, we have to work lower on Bloom's taxonomy.
After all this, you may wonder: why don't I just give up? I  get paid the same whether or not I do research, as too much tenured deadwood in  my department so amply demonstrate. 
The answer is that, at least, it keeps ME active. With my  4/4 teaching load, which whenever I request release time to make it 4/3 I feel  like Oliver Twist with his bowl out, it would be easy to allow teaching to  consume 100% of my time. 
What university professors know quickly gets out of date, if  they don't stay active in research. My undergraduate education was badly marred  by old proffies who were abusing their tenure and coasting to retirement. They hadn't  done any research in 20 years, and taught us a whole bunch of stuff that was  out of date, such as how to develop photographic plates. 
Research is especially valuable to the few of my students who do  go on to graduate programs in my field. It is also valuable to the many of my  students who don't, particularly the ones going into K-14 teaching. It does  help them to have some experience of what it is that scientists do. Also, everyone  around me says they like what I do. So, I keep at it.
- Froderick Frankenstien from Fresno