For the record, I agree with your points in your original post regarding the nature of free speech and with regard to the Overton window for tolerable speech (if there is such a thing).
I disagree with the notion that Yishan made a mistake in how he wrote about spam. You used that as a basis for disclaiming his conclusions.
Yishan was not making a point about free speech, he was making the point that effective moderation is not about free speech at all.
A) saying moderation is not about free speech is, I think, making a point about free speech. Saying one thing is unrelated to another is making a point about both things.
B) Even framed this way, I think Yishan is either wrong or is missing the point. If you want to do content moderation that better supports free speech, what does that look like? I think Yishan either doesn't answer that question at all, or else implies that it's not solvable by saying the two are unrelated. I don't think that's the case, and I also think his approach of focusing less on the content and more on the supposed user impact just gives more power to activists who know how to use outrage as a weapon. If you want your platform to better support free speech, then I think the content itself should matter as much or more than peoples' reaction to it, even if moderating by content is more difficult. Otherwise, content moderation can just be gamed by generating the appropriate reaction to content you want censored.
I disagree with the notion that Yishan made a mistake in how he wrote about spam. You used that as a basis for disclaiming his conclusions.
Yishan was not making a point about free speech, he was making the point that effective moderation is not about free speech at all.