I go through 2-10 papers a day, nearly on PL research, semantics, type-theory and implementation lore. There are a bunch of us on HN, some I correspond with via email, others twitter.
EDIT: 2-10 papers a day is an awesome rate!
There is a huge community of PL enthusiasts.
We can setup a site, but the field is too diverse. Right now it mostly makes sense as each of use tweeting a summary of a paper :-)
Email's in my profile.
Hi there partition,
I can't just send papers your way, but you're free to ask me questions and I will relay to my social circle of PL snobs (most of whom are in my twitter follows anyway)
I rarely read the abstracts. I find interesting papers in the references of other interesting papers.
I usually skip the first 15% of the introductory prose, go to the meat, jump to the conclusions and "future work", and if I think the paper covers enough ground, I fast forward to its references and see if the authors are aware of the seminal works. Only then do I actually read it.
For the truly important ones, you're already aware for their findings from texts, since they're cited often. For the useless ones (the majority) you're just interested in one idea, technique, finding or implementation method, and it's easy to zoom in. But from time to time you will come across a diamond in the rough that isn't as well known as it should.