My friend and colleague Christian Schneider forwards this interesting excerpt from an essay by David Foster Wallace on talk radio... which captures something that most critics miss.

All I would add is the difficulty of talking without the normal feedback; the human reactions that are so much a part of conversation... eye contact, head nods, looks of puzzlement or amusement, even annoyance. Try having a regular conversation with a blank wall instead of a person and you get the idea. That's one of the reasons you will hear hosts interacting with their producers so much; even though there may be thousands of people listening, there is still the need to have a single human being present and responsive.

"To appreciate these skills and some of the difficulties involved, you might wish to do an experiment. Try sitting alone in a room with a clock, turning on a tape recorder, and starting to speak into it. Speak about anything you want-with the proviso that your topic, and your opinions on it, must be of interest to some group of strangers who you imagine will be listening to the tape. Naturally, in order to be even minimally interesting, your remarks should be intelligible and their reasoning sequential -- a listener will have to be able to follow the logic of what you're saying-which means that you will have to know enough about your topic to organize your statements in a coherent way. (But you cannot do much of this organizing beforehand; it has to occur at the same time you're speaking.)

"Plus, ideally, what you're saying should be not just comprehensible and interesting but compelling, stimulating, which means that your remarks have to provoke and sustain some kind of emotional reaction in the listeners, which in turn will require you to construct some kind of identifiable persona for yourself-your comments will need to strike the listener as coming from an actual human being, someone with a real personality and real feelings about whatever it is you're discussing. And it gets even trickier: You're trying to communicate in real time with someone you cannot see or hear responses from; and though you're communicating in speech, your remarks cannot have any of the fragmentary, repetitive, garbled qualities of real interhuman speech, or speech's ticcy unconscious "umm"s or "you know"s, or false starts or stutters or long pauses while you try to think of how to phrase what you want to say next.

"You're also, of course, denied the physical inflections that are so much a part of spoken English -- the facial expressions, changes in posture, and symphony of little gestures that accompany and buttress real talking. Everything unspoken about you, your topic, and how you feel about it has to be conveyed through pitch, volume, tone, and pacing. The pacing is
especially important: it can't be too slow, since that's low-energy and dull, but it can't be too rushed or it will sound like babbling. And so you have somehow to keep all these different imperatives and structures in mind at the same time, while also filling exactly, say, eleven minutes, with no dead air and no going over, such that at 10:46 you have wound things up
neatly and are in a position to say, "KFI is the station with the most frequent traffic reports. Alan LaGreen is in the KFI Traffic Center" (which, to be honest, Mr. Z. sometimes leaves himself only three or even two seconds for and has to say extremely fast, which he can always do without a flub).

"So then, ready: go."