Sunday, July 12, 2009

EC 09 Travel

Travel to confs is when you do a little bit of many things, not a lot of any one thing. I traveled to EC to be part of the gestalt of Electronic Commerce (note to organizers: must change the name!).
  • EC started on Monday with the Workshop on Ad Auctions, a feast for specialists, papers flying off on themes of externalities, expressive bidding languages, pricing and equilibria; all papers are online.
  • Next (for me) was the Tuesday of tutorials. Jon and I did a tutorial on sponsored search and took a risky route, not talking about auction design. Jon splendidly laid out the world of many knobs that advertisers face (as one researcher exclaimed later, "And we work on what auctions to run?"). I spoke about how to optimize ad campaigns with a given budget. Researchers angst about whether advertisers care about clicks, profit, ROI, conversions, or whatever; do they consider interaction between keywords, broad match or stochastics, etc. My main point was these goals --- notwithstanding our passion to apply LPs or discover hardness of problems --- can all be compiled down to the knapsack problem with suitable weights and values, to some approximation. Then, simple bidding methods work. I ran out of energy some, and did not wrap up this argument like I wanted. I hoped to give an effective way for advertisers to think about their ad campaigns.
  • EC Conf started on Wed, banquet-ed on Thurs and sputtered on Fri. I liked the two papers (1, 2) that tried to understand the price of truthfulness in learning click-thru-rates for revenue/efficiency in ad auctions. At the banquet, we learned Nicolas Lambert continued his run and won the best paper award at EC. On Friday, Michael Moritz, a journalist, venture capitalist, and surely other things too, gave a keynote talk.
  • Nikolay Archak (data analysis), Tanmoy Chakraborty (cs theory) and Mallesh Pai (econ) are all interns in NY who were at EC as well, and I got to see the conf from many different perspectives through them. Nitish Korula, who was at ICALP, was missed.
When in confs, you discuss issues: on impact and problem formulation (Mukund Sundarajan), role of knapsack a la plastics in The Graduate (Rakesh Agarwal), estimating sparse queries (Bill Chang of Baidu), expressive allocations for large publishers (Tuomas Sandholm), etc. You catch up with friends as their lives evolve and some --- Arpita, shoutout to you --- let out their creativity. Finally, this is Palo Alto, you take an evening off and catch up on startup companies and VCs.



Anonymous Anonymous said...

Coud you post your/Jon's slides from the EC

1:34 AM  

Post a Comment

<< Home