Archive for July, 2011

Google Faculty Summit recap

July 29, 2011 Leave a comment

In July 2011 I attended the Google Faculty Summit at Google NYC. The Google offices contained everything that I’d heard about: open offices, tons of meeting rooms, high-quality free food and drink every 50 feet or so, scooters to take you from place to place, old video game console library where you can “test out” old games, a giant Lego play area where people create Google logos and Sergey pictures and such, a rec room with foosball and ping pong and medicine balls and yoga classes, the by-appointment massage area, great meeting rooms with views of the city, and all the other perks I’m forgetting. It was a smart decision to have the meeting at the headquarters–definitely a positive experience.

The meeting started with a series of presentations, with Vint Cerf (yes, that Vint Cerf, now a Google employee) leading things off. Other speakers included John Wilkes (Omega), Andrew Chatham (self-driving car), Johan Schalkwyk (mobile voice search), Stuart Feldman (Google around the world, especially the East Coast sites). Most of these things you can learn about online–though it was good to put faces with the projects and be able to connect with the speakers afterward.

Seems like meetings of this sort have one big “half-baked idea”–for the Google Faculty Summit it was the Library Wall–a wall-sized display of all of the Google Books. You could swipe around with it, open books, read parts of them, add them to your personal collection (no wonder Borders went under). It was great fun…not sure what will become of it.

The afternoon of the meeting was pretty free-form, with lots of breakout groups, which was a bit frustrating. I ended up in a meeting that started off great but went in some directions not at all interesting to me–and it wasn’t really possible to switch to anything else. (Other rooms were full, or discussions were well under way.) It was good to get to talk to Johan about mobile voice search–though 2/3 of the people in there were algorithms folks trying to get him and his team to adopt their latest algorithm. The rest of us tried to push the Google team to broaden what the API can do (e.g., use more limited vocabularies, return correctness probabilities) but no promises!

Breaks and dinner provided a great opportunity to connect with people (my UC Boulder Department Head Jim Miller, NCSU/UVA colleague Mark Sherriff) and re-connect with people (Jason Hong and his pending baby and tenure case, Ben Bederson and the new UMD professional MS program). More time for that sort of interaction would have been great.

One interesting aspect of the second day was a rehashing of Google Health by its creator, Alfred Spector. Google had recently announced that they were shuttering Google Health at the end of the year. Most people assumed that it was the security and privacy issues that led to the closure, but Alfred’s take was that they never had enough data to really run into those issues (kind of scary that they were collecting data without a model of how it would be used/shared). Alfred pointed to two issues that were more relevant to GH’s failure: buy-in and objectivity. First, nobody was joining–not enough people to make it worthwhile for the types of tools Google wanted to develop. Second, it was difficult to establish objectivity; e.g., if liver function is poor but improving, it’s hard to say whether it should be classified as “good” or “improving” or “disastrous”…all of which may be true (or false).

Alfred also explained the Google research model–to focus on things that are useful now, to have researchers working side-by-side with developers, to seek out flexible knowledge dissemination paths (i.e., not papers). Not sure how ideas for the long-term will emerge from that, but it seems to be working for the short term.

Alfred also pitched all of the university programs–grants, focused awards, named fellowships, internships, visiting faculty, postdocs, summer of code, CS4HS–all great. In talking with the Google folks afterward, it’s clearly valuable (essential?) to have a high-up contact at Google to make this happen. Guess that should be my next step…Google Boulder, here I come!