Wait, my message was obviously intended as a bit sarcastic (which isn't very smart, I'll admit). But are you actually saying that I'm now allowed two racist comments without risking a ban? (three, counting this guideline-abiding comment?)
It's not the case that "for each post that doesn't break the guidelines, you're allowed one that does", and that's not what I was doing. When I said HN moderation has worked the same way for over a decade, I didn't mean that the description you gave was accurate—it isn't. (Nor, I assume, did you mean it to be, since you were being sarcastic.)
I meant that what I was doing in the GP comment (https://news.ycombinator.com/item?id=44403362) was standard practice. As you can see from https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que..., it goes back a long time.
We try to persuade users to follow the site guidelines, and tend to give warnings and make requests before banning accounts, especially if they are active participants who have been around for a while. We don't rush to banning such users; we try to explain the intended use of the site and convince them to honor it. Sometimes that works, sometimes it doesn't.
I am absolutely no one, but I'd like to highlight that this kind of policy is (indirectly) why I don't use HN. Tolerating intolerance to the extent you do (which isn't 100% but still a lot) allows people like the one you responded to originally to drive hackers like me, my loved ones, my colleagues and my students away, while attracting other hateful people, as they see that they are tolerated here. In a possibly too extreme comparison, this the same dynamic as the "nazi bar problem" (https://en.wiktionary.org/wiki/Nazi_bar). I hope you know what kind of community these policies has made of HN.
It's easy to invoke strong pejoratives like "hateful" when describing people who have opposing viewpoints and passions to one's own—in fact, it's hard not to. But it leads to a rapid escalation. A bad comment turns into a "hateful view", "hateful view" turns into "a hateful person", and soon that leaps to "how can you tolerate hateful people on your site". (The next logical step would be to suspect the mods of being "hateful people" themselves.) This escalation is, in my view, bad for community. It leads to uniformity within one's own group and rage and enmity towards difference.
Having banned countless accounts for breaking the site guidelines over the years, I can't accept that "hateful people" are tolerated here for very long. When accounts are posting abusively, we may give them more warnings than you (or a lot of other users) would prefer, but we ban them in the end. A good example is this very subthread. I ended up banning that account (https://news.ycombinator.com/item?id=44403629). (Not, I should probably add, because of this or any other conversation about moderation, but just out of standard practice.)
p.s. You are not no one! I appreciate your comments and I wish I could write a better reply—I know a better one is possible, that expresses more precisely how I think about this. Alas it would take me hours, so I'm making do with one I don't much care for.
https://news.ycombinator.com/item?id=23308098 is one time that I got closer to it, and maybe https://news.ycombinator.com/item?id=31812293. I still like the phrase "supported communication across differences". Unsupported communication across differences just leads to Hobbesian flamewar.
But anyway, this is only one case and we should not base our thinking just on it. The problem is the policy (or the way it's systematically enforced) and its broader results. I don't know the details of how the moderation works here nor have I any statistics. I only know that I saw too much racism and hate towards whole groups of people because of their identity here in the past, and that when I occasionally stumble across a HN link, I usually can still see that hate being a lot more represented than in other spaces I frequent, and that the kind of policy you described to me has never worked at building diverse and interesting communities.