[Meta] Finding free energy in karma

by [anonymous]1 min read24th May 20204 comments


Site Meta
Personal Blog

As I've been posting here on LessWrong for a few months now, I wanted to share two things I've noticed about the karma system. These are ad-hoc observations, not rigorous empirical results, and while they might feel like ways to "cheat" the system, I hope by making them public and known, it serves to make the "market" for karma more efficient instead.

1. [Fairly confident] Posts appear on the front page unevenly throughout the week. As best I can tell there's a lull on Sunday and a spike on Monday/Tuesday. I'm not sure if this is due to unevenness in when people write, or unevenness in when mods promote posts to the front page. Regardless, since your post's karma depends in part on how long it remains on the front page (which provides more eyeballs), and that depends in part on what other posts it's competing against, there is opportunity here. Consider delaying publishing your post to a less busy time of the week.

2. [Less confident] Automatic cross-posting is a self-fulfilling karma signal. Having automatic cross-posting set up requires both sufficient author interest and sufficient moderator trust, since it's still a fairly manual process. This means that auto-x-posting is a fairly significant indicator of status/karma in the community, and such posts tend to do better in part as a self-fulfilling prophecy (of course they often do better because they're unusually good too - the karma system isn't broken, just slightly unbalanced). There's no way to replicate the automatic-x-post indicator in a manual post, but even just putting "[Cross-posted from MyBlog]" at the top of the text seems to provide a small boost.

4 comments, sorted by Highlighting new comments since Today at 2:22 AM
New Comment

Automatic cross-posting is a self-fulfilling karma signal.

I think there's something true about "crossposting is a signal of high trust", but I'd guess that it's not necessary to postulate an additional benefit from crossposting. The people with automatic crossposts are respected authors with an established readership, so they get more karma right away, and probably would still do so even if it weren't a crosspost.

If you have some actual numbers saying that less-high-profile-authors who crosspost also get a karma boost that'd be interested to learn, but I'd bet against it.

[-][anonymous]1y 1

Funny cross-thread coincidence, but I now think that maybe what I really noticed for point #2 is what you described here: https://www.lesswrong.com/posts/4Gbv7tmJs3ADbhJX6/reflective-complaints?commentId=KtekbKMi6SSzsaw4H. Crossposts just do better than linkposts, not necessarily better than "original" posts.

Speculative things that could reverse conclusions/cause and effect around these factors (if they are the case):


  • More posts on Monday/Tuesday indicate more people are using the site at those times, readers and authors.
  • There are more posts Monday/Tuesday because there are less posts on Sunday. (Readers become authors in the absence of content.)


  • People like reading on blogs, or reading blog material which didn't get cross posted, and that's where the boost comes from.

Also people in non-US timezones are at a disadvantage since they either have to sacrifice visibility if they post it at a time that's convenient for them, or comment engagement if they deliberately post it late at night.