One example was the self-help blog of Phillip Eby (pjeby), where each new post seemed to bring new amazing insights, and after a while you became jaded.
Er, you do realize I stopped most of my blogging for more or less that reason, right?
Around that time, I started pushing for a (partly LW-inspired) greater focus on empirical improvement in my work, because there was just too much randomness in how long the effects of my then-current techniques would last. Some things were permanent or nearly-so, and others might only last a few days or weeks... and I had no reliable way to predict what the outcome of a particular instance of application would be.
It was a tough shift, because at the time I also had no way to know for sure that anything more reliable or predictable in fact existed, but unlike the more "faith-based" self-help folks, I couldn't just keep ignoring the anomalies in my results.
The good news is I got over that hump and developed more reliable methods. The bad news is that it didn't involve brilliant simple epiphanies, but lots and lots of little hard-won insights and the correlation of tons of practical knowledge.
(And one of those bits of practical knowledge is how to avoid stopping at the "epiphany" phase of a given insight.)
Anyway, I quit blogging about it (at least to the general public) because once you're no longer dealing in simple epiphanies, there starts to be too much inferential distance to be able to talk about anything meaningful, short of creating my own Sequences to reconstruct the inferential chains... one mini-epiphany at a time.
creating my own Sequences
Please do!
LW doesn't seem to have a discussion of the article Epiphany Addiction, by Chris at succeedsocially. First paragraph:
I like that article because it describes a dangerous failure mode of smart people. One example was the self-help blog of Phillip Eby (pjeby), where each new post seemed to bring new amazing insights, and after a while you became jaded. An even better, though controversial, example could be Eliezer's Sequences, if you view them as a series of epiphanies about AI research that didn't lead to much tangible progress. (Please don't make that statement the sole focus of discussion!)
The underlying problem seems to be that people get a rush of power from neat-sounding realizations, and mistake that feeling for actual power. I don't know any good remedy for that, but being aware of the problem could help.