Paul Graham gave a keynote at PyCon 2012. He was heavily criticised, because he consistently says ‘um’ ~7 times per minute.
In response, he wrote an essay, where he rates talks as "certainly inferior to the written word as a source of ideas". Though he also admits that "talks are also good at motivating me to do things". This captures the difference between writing and speaking. Speaking is not about information transmission. Speaking is to make people do something.
- If you want to inform people: write!
- If you want to mobilize people: talk!
Look at exceptional speakers. For example, Steve Jobs keynotes made you go to the Apple Online Store and preorder the latest products. Bret Victor in his recent Inventing on Principle talk makes you rant about the current state of IDEs. Martin Luther King's I have a dream made lots of people stand up against racism. Ronald Reagan made Mikhail Gorbachev tear down this wall. Randy Pausch in his Last Lecture remembers you to achieve your childhood dreams. And there are many more.
As an academic, I often perform and witness a strange abomination: The academic talk. Persuasion is disdained in science, but lots of talks are given nonetheless. The convincing is makeing the audience look closer into the research of the talker. Though nobody publicly admits that this is the sole reason. Of course, there are exceptions like this:
The difference between writing and speaking also holds for one on one interaction. That is why managers like to come by and interrupt you. That is why programmers prefer email.
As a corollary, this means oral status reports are ineffective, because the idea is just to inform people. Also, written rants are ineffective; Hence, Steve Yegge should be more effective by speaking instead of writing.