Funny I made a TV show for public access titled "The Real Debate" where I spliced Ralph Nader into a Bush-Gore debate. I had one sequence where I spliced the two mainstream candidates to make them breathlessly advocate militarism and then segue into a discussion of schools. The timing was brilliant for most of it except for the part where Gore says forcefully "IF IT'S A FAILING SCHOOL, SHUT IT DOWN!" at which point a smart bomb was supposed to hit a bunker but because of the weirdness of the S-VHS editor I was using I could never get that synced perfectly with the sound track.
So far as flagging, this is my understanding.
A small number of people need to flag something for it to appear [flagged]. It may be as little as 2 or 3 but it is certainly more than 1. Some people do watch the new queue and flag a lot, other things are likely to get flagged when they get more visibility.
I made a model which could predict the fate of HN submissions and found my model seemed to perform unreasonably well at predicting flagging (a ROC around 0.96 as compared to the 0.72 or so of my "does it get more than 1/2 a comment per vote?" model and the 0.63 of my "does it get more than 10 votes?")
Turned out it was too good to be true because HN automatically kills a lot of submissions that you never see unless you turn on showdead. Some people spam the same headline over and over again hundreds of times and if you don't design your test-train split such that the same headline can never appear on both sides you will get the wrong idea of the performance of the model. (This affected my other models too but not so catastrophically)
The #1 way people get their posts marked [dead] all the time is they keep posting links over and over from their own blog or other site but never post anything else. You can certainly do some self-promotion as a community member of HN but if you're not a participant in the community it's not welcome.