In a talk or seminar, the audience is crucial to success. Will they ask incisive interesting questions? Are they prepared? In some seminars there is a tradition of very vigorous questions, often not letting the speaker go for more than a few minutes. That is, there is no waiting until the end of the talk.
In the law school workshop I attend, people have the papers ahead of time, read them carefully, and the session begins with a commentator, then a response from the author, and then questions from the audience. The commentator is often incisive in figuring out what is crucial in the paper. The audience is ready with questions that they have noted while reading the paper, and some members have particular kinds of questions they always ask. I find it interesting and fun, even though my knowledge of law is epsilontic. In general, the audience is trying to help the author make for a stronger paper. It is in the family, so to speak.
I tend to be impatient, and want to know what's up well before most speakers (if no paper ahead of time) tell you. So often in the first twenty minutes I finally figure out what is going on, and ask if my figuring out is correct. Ideally the speaker would give away the main ideas and findings or argument in the first few minutes.
My advisor would seem to sleep through a talk, and then ask a crucial questions. One of my other teachers, used to ask innocent-sounding questions, likely to sink the speaker since my teacher had discerned a deep problem with the work.
My other questions are usually about analogies to the current subject or situation.
There are questions about statistics, reliability, etc, characteristic of social science, chipping away at the work. Rarely are they interesting or able to much decrease the credibility of the speaker. Some of the time, they can be devastating, but a good scholar has already anticipated those problems, largely because they have colleagues read their work before going out and talking about it.
The best questions come from understanding the problem in the writer's or speaker's terms, and then try to deepen or question the endeavor. One is reading to find out what is really going on, and your goal is not to chip away at the paper. Rather, you want to engage in a conversation with the author/speaker that enables the work to be seen in a more significant light.
Minor advice for speakers: If you quote numbers in your talk, be sure to give comparisons. Telling me that 532 schools were closed is useless unless I know how many schools there are. If you are quoting numbers that are statistical or simulation-model based, be sure to give the error bars or estimates or the range from sensitivity analyses. Be sure that you can tell the main point in plain English. And if it is a complicated argument, sketch out the argument before you go through it--and perhaps have that sketch available throughout the argument (as a slide or handout) so that people can follow what is going on.
Also, give away the findings that are important in the first few minutes.
