Yes, this is why advertising funding is an "interesting" case and falls between the extremes. One solution is "firewalls" between the department selling advertising space and the editorial team, so that explicit threats of blackmail can't get through. The paper might need to show evidence of such firewalls to claim protection for pieces which are labelled as comment but look suspiciously-like paid-for advertising.

What is most difficult here is "self-censorship" whereby the editor knows that if he runs a particular story, then the advertising will dry up, and the paper risks going out of business. But this is not in principle different from dilemmas on readership such as "If I run this shocking story about what our troops are up to abroad, then I'll sound unpatriotic, lose readers, and go out of business".

There is self-censorship in almost all speech contexts ("If I say that, my friends will think I'm an idiot", "If I post that, it will get down voted"). But the important point is that what emerges through the self-censorship filter is protected. The intuition here is that we don't want to impose even more filters.