You know you're tired when it takes you more than two tries to hit an elevator button. I'm back in town and back at work, having arrived in Pgh this morning around 9pm.
Last travel log bit: As the conference ended, I got that "oh no, life has no structure again" feeling coupled with my usual lonliness.. I decided to try to make the two last tourist-y things on my trip - Amber India and Birite Creamery. Amber India was a thorn in my side for the whole go - a wildly incorrect and inconsistent address between all the places that referred to it. I did spot it while heading back from the Stanford area once, so I went off in that direction, drove for about 45 minutes, turned around, and eventually managed to find it. Definitely worthwhile! The Saag Paneer was absolutely fantastic, as good as my memory of the best Saag Paneer I've ever had (in Portland, Oregon). The samosa were pretty good, and were accompanied by chickpeas that were spiced a bit like how Srees does his, but much more richly (answering the question I've always had on whether that was something he came up with here or actually part of Indian cuisine). I then tried to go to Birite, but I hit rush hour and didn't want to risk missing my flight (also wasn't particularly in the mood for Ice Cream). I returned the broom, and had a pretty uneventful-but-long set of overnight flights back - I kinda-sorta managed to sleep through most of both of them. Went home, dropped off the drawing tablet and laptop number two, swung by Taza21 to get lunch, and came to work.
Impressions of the Bay Area -- Newark is to New York what the Bay Area is to San Francisco. This is broadly true, with some heavy nuances. The land in the Bay Area is incredibly beautiful - the preserves I went to were fantastic (made me glad I decided on that rather than the city-touristy kind of things). The Bay Area city planners did an amazing job at making an awful, horrid stream of towns/cities out of this beautiful land - the whole area needs to be razed and replaced (not quite as badly as Newark does, though). SF is okay as a city, bordering on nice, but the neighbours.. ugh. Endless sprawl is generally pretty bad - DC is the only city I know of that manages somewhat-less-than-awful in this area (although there are a number of cities where I don't know what their sprawl looks like, and some areas, like Brecksville, where their sprawl-status is unclear). Overlooking the city was a good way to aesthetically survive.
Networking stuff is not naturally a particularly deep interest of mine - it's not uninteresting but little in it makes me go wow. One of the things that would've still interested me were I not paid to be interested was a specification called OpenFlow for an interface on high-end managed switches and other network hardware that can do powerful network manipulation at line rates. The demo was really neat - they have a white paper here. Other than that, I got a feel for where Emulab (the software project my job is associated with) fits into the GENI project/proposal (a government programme which came around later to fund and specify networking testbeds and pulled in 4 other existing projects, pushing them to work together in some ways). One of the competing projects, Planetlab, looks like it has a lot of momentum behind it, probably more than Emulab. There was also some interesting discussion in a break-away group on the topic of IRB approval in testbeds, both on which parties need it when a researcher from university A and one from university B use a testbed affiliated with university C, and how different research groups can learn from each others IRB experiences. This is particularly relevant to where we are in one of the projects here I'm working on. I was a bit worried when they repeatedly abused the term "user opt-in" at the conference though, and was very bothered by one of the Planetlab engineers describing a "avoid talking to IRBs when at all possible" perspective. One interesting thing that was brought up by one of the rebuttals to that statement was that IRB-approved experiments can grant some legal immunities (making some data unavailable by subpoena, for example).
I think a lot of geeks would love to treat legal matters as an externality which we can ignore. That makes a lot of sense, and we often do practically ignore these matters (who hasn't ripped a CD and shared it with their friends?). Taking that attitude does not always make sense though - ignoring externalities often has dire consequences. My general attitude has been to use the private life/employment as a dividing line - I use my notion of the public good and ideal legal code combined with a certain attitude towards risk/consequence when encountering the actual law (as I understand it) and other people, while when I'm at work acting on behalf of that institution, I tend to play things almost strictly by the book - I'm not generally happy lying or omitting things to internal institutions at work, and see it as my duty to be forthright in such an organisation, because I believe this generally helps the organisation function better (even if shortcuts can be quite appealing). It's interesting seeing which people have that kind of split behaviour versus the all of one kind or all of another (or some other attitude entirely).
The conference overall wasn't too bad - seeing HP's HQ was kind of neat, and I did get a feel for big-picture stuff at work. I don't know if it was worth the money to send me there, but I generally don't understand how value and money relate in universities (nor do I feel any particular need to know). I suspect very few people actually could solidly say whether things like this are worth it based on more than a gut feeling, this holding for universities, corporations, small companies, etc. On some level, apart from cultural differences all these things are the same on small scales.. a people's committee to build a bridge, a committee of university engineers, government engineers, private engineers, etc.
It's intersting how many plebiscites California has.