Graduate Education
Graduate education is in need of some improvements. It takes too long to get a PhD, and there is too little mentoring and training during that time. Many students are just thrown in a lab and required to make a "significant" contribution to be able to finish. I've been meaning to write more about this, and eventually I will. But in the mean time the blogger "Adventures in Ethics and Science" has written a few great words about this topic. I put a great a quote from the article below.
Principal investigators have too much power over the fate of their graduate students.
What if departments decided that it takes a village to raise a scientist? What if no single professor held the power of scientific life or death over a graduate student? If the standard mode of graduate training involved collaboration between faculty members directing and supervising research projects, and regular interactions with other faculty members, it might be easier for students to stay on track if one of those faculty members had to leave in disgrace (or even if she left on good terms). It would also expose the students to different styles of mentoring and lab management, which might well be useful for them when they become PIs. And, it might help the department faculty to regard the graduate students as budding scientists whose strengths and weaknesses they know first hand, rather than simply as their colleague's serfs.
Another shift that might help in situations like this would be to consider that a series of small, distinct research projects could also constitute an important contribution to a scientific field. If we only count projects that take 5+ years in a lab as Ph.D.-worthy, the chances of getting back on track if a project falls apart in year 4 or 5 are not good. They may be low enough that people close to the end of their time have a significant incentive not to notice or look for flaws in their research, even obvious ones.
Post-doc committee
Good ideas Geoff. The most important part is the involvement of an advisor. Too often the advisors are not truly involved in the student's progress.
The idea of a post-doc committee is also good. I would like to see it go even one step further, make post-docs independent investigators who have an advising committee. Why does a post-doc need to be tied to one investigator's lab? If a post-doc can write a grant and get funded, then they don't need to be tied to one PI. This would dramatically change the structure of post-doctoral research, but I think it would benefit the researcher and the research.
Thesis Committee
There are some good suggestions and ideas here, but I think most senior scientists would read that and say, "We're already doing that, they're called Thesis Committees." I think the idea that the author is trying to promote though is more involvement of a committee in the training of a PhD student, rather than a meeting once a year. Expanding this idea a little further, I think the committee idea should be extended to post-docs. I think that as students we have help from a lot of sources and we do have thesis committees, whereas post-docs usually only have their advisor. This shifts the power squarely to the advisor. I think post-doc committees would help give post-docs mentorship as well as expand their network.