Take a look at something like Chomsky's political speech attacking NAFTA and neoliberalism in the 1990s, which Graham briefly alludes to. NAFTA was and remains a fairly popular initiative for the liberal establishment, but the principles under which Chomsky's justification stands can be imagined to hold up under scrutiny decades in the future, whereas exploitation of cheap Mexican labor -- probably not so much, I'd wager.
Another recent example -- Raghuram Rajan's 2005 paper [0] criticizing the financial innovations that would go on to destroy the world economy in the 2008 crash. This example satisfies the two "conspiracy-theory-esque" conditions that you cite from Graham, as he certainly made Kohn, Greenspan, Summers, and other major figures in the economic establishment angry enough to call him a "Luddite" with views that would qualify under Graham's "unsayable" criteria. At least, before he would be proven right just a few years later.
So you're certainly correct in pointing out that not every instance of taboo carries ideas worth considering -- and for some ideas, this is indeed a dangerous understatement -- but I don't agree with you when you imply that Graham is wrongheaded in every instance, and no taboos should ever be considered. To give another example, Joyce's Ulysses was banned as obscenity across the world in the 1920s [1]. As far as I can tell, Graham is more concerned with scientific innovations rather than political speech, anyway.
[0]: https://www.nber.org/papers/w11728
[1]: https://www.theguardian.com/books/2014/jul/04/most-dangerous...