When it comes to making a concept making a post like common knowledge well read I’m not sure that curating a list is the best way to go about it.
The alternative is to rely more heavily on hyperlinks. Hyperlinks provide a good way for new people to discover content that’s important to know about because it’s referenced by other discussions.
Hyperlinks seem to be the main way LW goes about this (both now and previously), and they definitely seem quite important. (This is indeed how I read most of the sequences). But they’re fairly bad for being confident in my “completionism” (I have no idea what percentage of the sequences I’ve read), which is in turn bad for knowledge of what everyone (or at least most people) have read. It also means that people end up reading things in pretty random order, which means someone who’s read, say, half the sequences and a bunch of random posts may get a less than half-solid foundation.
Having said that, I think reading via random hyperlink crawl was more _fun_ than reading R:AZ in order. I’m glad it’s an option, but I don’t think it should be the only option.
Heavliy hyperlinking a post doesn’t force the post the actually undergo periodic revision.
Hyperlinks seem analogous to “citations” in science, as a reasonable proxy metric for how important an idea was (i.e, if a lot of people are linking to you while writing other posts, that means you post was a useful building block for other ideas). This is obviously easy to goodhart on but I actually think is pretty reasonable for a rough approximation.
It’s occurring to me as I write this that I’d actually like to see more mechanics tying in hyperlinking within LessWrong.
if an author retracts a post, then all posts on LW linking to that post could get a little red “this post is linking to a retracted
you could plot out a graph of which essays get most linked within LW articles that are in turn further linked, to generate an approximate list of which ideas are most load-bearing, which could be useful both as a “most important to read for context” list as well as a “most important to actually vet the claims, in case The Replication Crisis happened or something”. Depending on use case this’d probably want to be a first pass list that is then manually curated
In science you don’t have any journal where all papers are read by everyone and it still works efficiently to build on the work of other people.
The main way this is achieved is by putting the responsibility of reading whatever is referenced when the reader feels like they need more information.
I think that even if you would have a curated list of very important posts not everybody is going to read them. Having such a resource could be useful as a resource for some people who desire to have more curation to guide their reading but that won’t be everybody.
When it comes to additional features to encourage linking, we might add a list of backlinks within LessWrong to a post. On the top of a post the amount of backlinks could be stated and when a user clicks on the counter it expands and all the backlinks are shown.
When it comes to making a concept making a post like common knowledge well read I’m not sure that curating a list is the best way to go about it.
The alternative is to rely more heavily on hyperlinks. Hyperlinks provide a good way for new people to discover content that’s important to know about because it’s referenced by other discussions.
Hyperlinks seem to be the main way LW goes about this (both now and previously), and they definitely seem quite important. (This is indeed how I read most of the sequences). But they’re fairly bad for being confident in my “completionism” (I have no idea what percentage of the sequences I’ve read), which is in turn bad for knowledge of what everyone (or at least most people) have read. It also means that people end up reading things in pretty random order, which means someone who’s read, say, half the sequences and a bunch of random posts may get a less than half-solid foundation.
Having said that, I think reading via random hyperlink crawl was more _fun_ than reading R:AZ in order. I’m glad it’s an option, but I don’t think it should be the only option.
Heavliy hyperlinking a post doesn’t force the post the actually undergo periodic revision.
Hyperlinks seem analogous to “citations” in science, as a reasonable proxy metric for how important an idea was (i.e, if a lot of people are linking to you while writing other posts, that means you post was a useful building block for other ideas). This is obviously easy to goodhart on but I actually think is pretty reasonable for a rough approximation.
It’s occurring to me as I write this that I’d actually like to see more mechanics tying in hyperlinking within LessWrong.
if an author retracts a post, then all posts on LW linking to that post could get a little red “this post is linking to a retracted
you could plot out a graph of which essays get most linked within LW articles that are in turn further linked, to generate an approximate list of which ideas are most load-bearing, which could be useful both as a “most important to read for context” list as well as a “most important to actually vet the claims, in case The Replication Crisis happened or something”. Depending on use case this’d probably want to be a first pass list that is then manually curated
In science you don’t have any journal where all papers are read by everyone and it still works efficiently to build on the work of other people.
The main way this is achieved is by putting the responsibility of reading whatever is referenced when the reader feels like they need more information.
I think that even if you would have a curated list of very important posts not everybody is going to read them. Having such a resource could be useful as a resource for some people who desire to have more curation to guide their reading but that won’t be everybody.
When it comes to additional features to encourage linking, we might add a list of backlinks within LessWrong to a post. On the top of a post the amount of backlinks could be stated and when a user clicks on the counter it expands and all the backlinks are shown.