Every student should work a year before coming to grad school. (And then, a la Columbia, one should be limited to five years except in exceptional cases.*) Why? Because one only learns to recognize a bad fit by experience. Also because advisors have no incentive to get people out in five when they're so much more productive in the 6th year, but that's another story.
Not a one of the straight-from-college little darlings know what to look for. Even if they've worked in a research lab at an R1, it is NOT THE SAME. Of course one can still spot grossly abusive/ neglectful/ crazy labs, even as an undergrad. But the chances of observing the full range of things which will affect one's own research... are low. Unhappy labbies are unproductive labbies. Joining a lab that makes one miserable, even if it is a good lab for many other people, stretches out the PhD.
Of course, some of them are bright enough to observe and catch on. But a lot of them, especially this crop (decreasing funding means this Uni admits 'better' students, i.e. snootier ones, who tend to have little to no real work experience, and also to be extremely sheltered)- a lot of them will learn the hard way.
Alternatively, every PI could step up and do their jobs. The university could establish standards, teach mentoring, make regular checks of progress (not those lame forms we fill out now, which are read by a historian, so helpful). Advisors would see that new students are properly trained in techniques, guide their experimental designs so they learn how, provide feedback on a project's success, and so on.
But don't hold your breath.
This post brought to you by the letter D and the rotation students, who make me grind my teeth. D is for data, that's good enough for me...***
*Extreme bad luck, ill health, working on plants,** etc.
**Because gene mapping in plants typically takes years, and there's NOTHING YOU CAN DO ABOUT IT. Or, why to work with a short-generation-time organism.
***I have a hard time not empathizing with their future selves.
