I had so many great experiences this PLDI; for the benefit of the reader, I will condense them down into a few choice vignettes...
1) About the people I finally managed to track down after spending countless hours reading code and being told "hmm.. I don't know, Brian wrote that..." when asking how it worked.
2) About the time I accidentally crashed the ECOOP banquet by happening to be touristing in the same district at about the same time, and ending up in a conversation about research with an attendee as it was starting.
3) Apologies to the deep-fried scorpions that Alan Leung and I ate the night after his talk at Wanfujing market. Your lives were regrettably short, but delicious.
4) About learning to haggle. I am of chinese ancestry by genealogy, but very much californian by any other metric. I am now convinced that there is some hereditary component to the ability and enjoyment of haggling with street vendors.
By attending talks and talking extensively with other researchers, both in and out of my specific static analysis niche, I learned quite a bit about both the research community, the work that we are doing, and the process by which it is leaking out into the industrial community. In particular, in the niche of static analysis, my feeling was that the biggest holes in our research are not expressiveness, as we (ie, I, in particular) are wont to pursue, but first, frameworks for handling the features and scale used in real industrial code, and second, error and warning reporting. (On the latter note, special attention should be given to the Dilligs' excellent paper and talk, which pushed this problem forward in a particularly important way.)
The overall feeling I got was that the common case in static verification is not one in which code is correct, and a programmer simply needs some assurance that this is the case, but that code is inevitably incorrect in some way, and the programmer simply needs some feedback on how robust his or her code will be when used at scale prior to deployment, and more particularly, how that robustness will change as he or she evolves that code.
Finally, I want to extend my sincere thanks to all who were involved in organizing the conference, everyone who presented, the NSF for funding my trip, everyone who deigned to listen to me babbling on about my work in the hall or at lunch, everyone who came to my talk and asked such excellent questions, all the student volunteers, the city of Beijing, and anyone else I'm forgetting.