My experience with transcripts is that they are often far from perfect
unless someone who knows the material first-hand puts in _much_ time
editing the transcript against the actual recording. This is never
done in mass-transcription settings. Court transcribers are amazing
but not always right -- I've been in courtrooms where a transcript is
read back and it's not accurate. I had a time once where I gave a
deposition in a case as a 3rd party on behalf of one of the parties. I
am so glad I asked for a copy of the transcript and tape and said I
wouldn't certify the deposition until I checked it against the tape.
There were several key transcribing errors that messed up the intent
of my sentences. How many people do this, I asked the lawyer? He said,
you're the first!
The problem that Mike Beil was describing, though, is more shaded.
YouTube is even less relaible than Whacky-Packia. Every PC -- Mac or
Windows -- comes with video-editing software. So nothing on YouTube
can be taken as a complete, unedited source. So O'Reilly's blowup
could have been heavily or lightly edited before showing up on YouTube
-- or it could have been a selective portion showing O'Reilly in the
worst light (or the best light, for that matter). I agree that it's an
unreliable source. If the producer or owner of the show were to
produce a complete video that they would certify is unedited (not sure
how this certification would work), then that's a different matter. As
for using newscasts as an original source -- a newscast is by nature a
secondary source, highly edited, highly interpretive, highly filtered.
I agree the original audio is more legitimate than a written
transcript of that audio, as long as both are used and checked against
each other (for instance, if there is a paragraph in the written
transcript but not present in the copy of the audio you have, there is
a problem, Houston). For scholarly research, unless you are writing a
critique of something like a newscast, or using the newscast to show
the editing and filtering at work, it's much better to go back before
the filter and try to find genuine first-person sources. I always
prefer to find as many original-actors as possible, because you need
to triangulate their stories to get a clearer picture of the event
(memories are imperfect and even the first person has a heavy
filter/interpretation on everything). Bottom line, it's nearly
impossible to know beyond question the micro-details of anything you
don't experience yourself. "History" below a pretty high macro-level
is usually just a glorified mythology, a compromise of competing
filters and interpretations.