Hi all,
Thought I could contribute some useful info, with some help from my colleague and fair use expert Patricia Aufderheide. Given the amount of misinformation large media corporations and software associations have purveyed over the years, it's no wonder there's a lot of confusion about fair use. Some things to keep in mind:
1) Fair use is a right, as the Supreme Court has recognized affirmatively, and its technical designation as a defense doesn't make it any less of a right. Your right to self-defense is a right, although it is a defense.
2) Fair use isn't complicated to interpret, in most common cases. In fact, you probably used fair use today without thinking about. Every time we quote another scholar in our work, that's fair use. Every time search engines temporarily copy documents to scan them for relevance, that's fair use.
3) The Codes of Best Practices in Fair Use, which are available at csmsimpact.org/fair-use, are extremely valuable for providing user-friendly logic for fair use in specific disciplines. They exist for communications research (might be the most pertinent for the original question), documentary film, media literacy, visual arts (as noted in this thread), journalism, film studies teaching, open courseware, nonfiction writing, museums and archives facing orphan works problems, and film studies research. Forthcoming is a best practices code on software.
4) To echo SamR's comments, of course judges can find that a use is not fair, just as they can find that a use is obscene, or defamatory, or treasonous. That doesn't make the First Amendment any less real as a result. This is why Codes of best Practices are so helpful--they show what is generally accepted in a field already, and therefore unlikely to be found outside the acceptable in court, and therefore unlikely to draw any legal action from a copyright holder. Risk is a known feature of expression and of life in general. So the question is not whether you are risk-free (unless you don't want to express yourself at all), but what the risk is. Having consensus codes of practice is a way of vastly lowering risk. Documentary filmmakers found that it lowered risk so dramatically that insurers changed policy 180 degrees from "no" to "yes" on insuring against fair use claims, once they knew the expectations of the field, through the filmmakers' code.
All of this and more is covered in a very handy book by Aufderheide & Jaszi, Reclaiming Fair Use: How to Put Balance Back in Copyright, 2d ed. (University of Chicago, 2018), which I highly recommend for anyone with questions about the details of how this works in practice.