Loop Like a Native by Ned Batchelder was a talk on "fundamentals" or "primitives" -- I like that he pitched it that way, not as "for beginners". Experts need refreshers in basics too. "The ability to loop tall buildings in a single bound -- that joke is the whole reason I'm giving this talk." (applause). The talk was in Python 2.x instead of 3.x. Iterators come into their own in 3.x. "Abstract the iterations more"; good advice.
I also attended Anna Ravencroft's talk on what to do if your talk was not accepted -- not the boat I was in (my lightning talk sailed through) but I like to see how PSF members role model welcoming behavior. She did a pretty good job I thought.
At the PSF lunch, we all applauded the positive outcome of the trademark dispute in the UK, which had cost us. I sat at the Texas table (unofficially that), and put a good word in for Austin in 2016 (we're booked for Montreal the next two years, no US Pycons planned or expected as of this writing, other than the smaller state ones like PyOhio).
I don't have any quarrel with people marketing computer services using Python in their name, a practice PSF should encourage. However, if the trademarking rules in some country are such that there's some winner-take-all model regarding who uses the token "Python" for their services, well then of course PSF should fight to keep its channels free and clear.
We can't afford to have some fly-by-night operation call the dogs on us just because some crazy rule book says one and only one computer company is allowed to hold that token (most rule books aren't that crazy). We fight to keep the tent big, not the monopoly of any one group. Even nonprofits can get pretty dictatorial. Our table at least seemed to agree the OCLC was not a role model in how they took a bat to that New York Library Hotel that spoof-emulated / celebrated library filing, with books pegged to room number.
The New Relic guy did a great job profiling profilers, a world class expert in the subject.