I think that the reason for this is quite simple. Poor graduate students are often so excited about going to a meeting that they get their slides ready before the meeting and actually rehearse, often compulsively, before their peers, their mentors, and before their long suffering moms and siblings. They cut and prune their text to make sure that the talk ends within the allotted 12 minutes, and that there is time for questions. If you add to this the fact that they are often nervous and talk faster than they have done in the conference rooms of their home departments, they always end well before the bell goes off.
Most P.I.s seem to harbor the unjustified conviction that they can give any talk, anywhere extemporaneously. They always misplace a critical slide, and tell you that they left it in the airport in Zurich (just so you know that they are busy, world travelers, and you are lucky you caught them here), and try to tell you what there was in the slide that is missing. This of course takes 12.5 minutes more than it would have had they not misplaced the slide. Add to this the fact that they put half the slides upside down and try to read what is on it by standing on their heads on the stage, and the others simply don't belong to the talk at all (I once sat through a talk by a famous scientist who presented details about T cell receptor rearrangement at a B cell workshop), they happily go on well after the timid little bell goes off--which leads me to the next verity about meetings:
Being the chair in a symposium or a professional meeting is, of course, a highly prestigious thing, something that can be added to your CV and works wonders when you have to ask your department chair for more space. Of course, no one seems to know exactly what it is a session chair is supposed to do, and there are no good role models. Most chairs don't bother to familiarize themselves with the names of the authors before the sessions, and blissfully butcher the names so badly that the authors sometimes don't realize that it is their turn to speak. And this is the only rule in all meetings: chairs never enforce the rules. So, when the above-mentioned P.I.s go on and on, the chairs (there are often two of these hapless folk at each session) look at each other sheepishly and wonder what the hell they supposed to do.
In the old days, when I used to watch John McEnroe play tennis, I would wonder why tournament organizers did not hire a few 8-foot-tall, 600-pound football linebackers as lines people. And then, when John began to rant and rave when the calls went against him, I would like to have seen one of these gorillas stand up, and pick John up, shake him up and down and ask, 'Do you still think I made a wrong call, John?'
You can always tell when a famous speaker is giving the talk at a symposium, a meeting, or at your own departmental seminar. There are never enough chairs, people stand all around the place, and all too often the talk is quite boring. It is often the obscure scientist, working on a problem nobody seems to be interested in, who gives the best talks. Over the last few years, the most scintillating talk I heard was by a colleague who works on cestodes parasites of sharks. Of course, no one wants to hear about shark parasites, and all of five people came to the talk. They missed a glorious presentation, organized, concise, and exciting. Her enthusiasm was palpable and infectious. It handily beat talks by superstars presenting endless genomics seminars, which is hard to beat in inducing terminal ennui.
I have a feeling that this point is related to the fact that science has become so fragmented that no one understands anyone else's data anymore. This is probably true even if one actually opens a journal, peruses the figures and tables and tries to understand them with unlimited amounts of time at one's disposal. But in a seminar setting, where slide after slide flashes before your glazed eyes, it is quite impossible to understand the data, especially when the field is peripheral to your own. Early in the seminar, as people walk in full of enthusiasm, burning with desire to learn something new, almost everyone is awake. They stay awake through the obligate feeble jokes from the host and the guest. Some people begin to doze off during the introduction, though most stay awake because almost any scientific phenomenon is quite interesting, as long as you don't get into the gory details.
Things get grotty during the data section. There are always eager beavers, people who apparently suffer from chronic insomnia and ask penetrating questions. The rest of the audience yawns and can be heard muttering under their breath, "I wish the damn fool would shut up so that the speaker can finish and I can get back to my lab."