I think this question is framed slightly off. I agree with the general observation that "some authors get status effects that results in their stuff getting more upvoted", and/or "karma is awarded for writing style more than insight."
But, the question here is "what are the large structures within the rationalsphere, and who gets to establish those structures." Who has constructed a paradigm that others worked within? Who has attempted to construct a paradigm but failed to garner attention?
- LW began because the existing academic/industry structure didn't accommodate the paradigms Eliezer was thinking in. Eliezer was writing on Overcoming Bias, which had something of its own paradigm that gave him some space to talk about it. I wasn't actually there at the time so I don't know how that felt. But eventually Eliezer broke off to create LW, and by dint of having created it, having a lot of mostly coherent ideas, and being a good writer, established a new structure, that included substructures relating to rationality, meta-ethics, AI, etc.
- CFAR eventually was born. It mostly created its own internal structure, and mostly did not publish. The current state is that CFAR-style-rationality is notably different from traditional LW-style rationality, and only some of this content has been back-ported over to LW. (However, I think there is some recent work, esp. by Kaj Sotala, that is starting to bridge this backwards)
- Some cluster of people found that they either weren't getting traction on LW (or maybe already existed outside LW?) and started calling themselves postrationalists.
- Scott started writing a bunch of stuff. His original stuff fit within the default LW paradigm. Eventually he started talking more politics-y, which didn't fit as well within the LW structure, and he went off to his own space. I'm not sure how much "structure" or "paradigm-ness" Scott has, but there's at least some things like "the archipelago model" that I think of as his.
- EA solidified as a concept. It began in a few different places, including LW, Givewell and academia. It eventually carved out it's own infrastructure where it could focus on its thing.
- Within LW, especially recently, there are a couple different paradigms relating to AI. Perhaps tellingly, I think of them as "the MIRI paradigm" and "the Paul paradigm", and maybe the "vaguely mainstream ML paradigm".
- Zvi/Benquo sort of (accidentally?) co-created the "Slack and Sabbath" paradigm and some clusters of things surrounding that.
- Zvi/Benquo/SarahConstantin often reference each other and have something of a structure relating to organizations, integrity and out-to-get-you-ness.
- Duncan established something of a paradigm, and then withdrew to his own blog for reasons.
- By now, after a year+ of LW2.0, there's a structure here that's subtly different from the structure when Eliezer was the primary curator of LW.
So this points at a few key questions/points:
1. Some people seem to have an easier time establishing new structures. Is the active ingredient here "status", "writing skill?", or, perhaps most optimistically, "insight?".
2. I think the minimum you need to establish a structure is at least a couple other people to start linking to your stuff. I think this requires some minimum threshold of insight and writing skill, but can be bootstrapped by a couple people who think the same way. Early Givewell seems like a prime example of "had a minimum threshold of writing skill, and mostly gained status by virtue of legibly good insights."
3. Empirically, even highish status people tend to go off and establish their own fiefdom where they don't need to chafe slightly against the default culture on LW. This doesn't seem intrinsically bad. While a bit sad, it makes sense that people in different structures want the freedom to run with the ideas in their structure, and not obviously wrong that the way new major "rooms in the sistine chapel" get added is by someone going off and creating a new wing on a different website.
(Incidentally, this empirical observation is why I currently still hold out hope for LW eventually moving towards an internal Archipelago model, so that people can more easily fork off into their own subfield while still having the best ideas bubble up into a central location. I don't think we ended up doing enough work to fully support that, and the team is currently focusing on the Questions structure, but do think we'll come back to it eventually)