https://blog.ycombinator.com/two-hn-announcements/

> HN’s approach to moderation has always been three-pronged: software automation where possible, human intervention where necessary, and as much community moderation as users can provide. Our long-term vision for HN is to have the site be as self-regulating as possible, and we’ve been working on a project code-named Modnesty (for ‘moderation amnesty’) to develop new ways to do that...

I did a quick search and hadn't seen any mention of "Modnesty" (maybe it changed names?) recently. This question is sparked by something I saw regarding a new commenting system that the Washington Post and other big news outlets hope to try out:

http://www.niemanlab.org/2017/05/how-the-washington-post-pla...

The quality of HN's community seems to heavily rely on quality of moderation, which implies any new commenting system will be reliant on effective moderation. The New York Times famously expends a lot of resources for manual curation [0], though was said to have partnered with Google to figure out a more automated way.

As an end user on HN, we only see the end result of moderation -- things getting flagged, downmodded, rescued, and dang/sctb intervening (detaching threads, banning folks). But compared to 2 years ago, how much of that process has been streamlined? That is, has trusting users to vouch/flag worked out, or do human mods have to jump in just as much as before? How much has flagging/detection been moved to automation even before users act on it?

Obviously not asking to give up the secret sauce. Just interested in knowing how much progress has been made in moving more of the judgment/detection to algorithms.

[0] https://www.nytimes.com/interactive/2016/09/20/insider/appro...