Related: I generate an "Ask Hacker News Weekly" listing by scraping /ask the same way as my "Hacker News Daily" scrapes /news: http://www.daemonology.net/hn-weekly-ask/

A slightly more utilitarian approach is to elide the "Who's Hiring" results: http://www.hnsearch.com/search#request/submissions&q=%22...

I'm impressed they are indexing two letter words and a typical stop-word "ask".

Must be an interesting database size.

While I think they actually are doing minimal stemming, those are key terms that an index for this site absolutely should include in their index: even if every other two an three letter word is thrown out, "ask" and "hn" should not be.

Actually it looks like they are doing something special when phrases are searched for.

as it was returns zero results, while "as it was" has plenty

I am able to pick up pieces of relevant discussions using the search. A useful addition would be 'trending searches' ;

awesome Meta Tool, how do you guys make this, where is HN's api documented?