Sunday, July 29, 2012

On Facebook

I am not an expert, but people ask me to talk about Facebook's business. Here is a silly answer. Say Facebook has 1B users and each is able and inclined to buy a single share of Facebook for USD 40, a onetime transaction, no annual fees.  That is 40B dollars that Facebook can invest, and using what private hedge funds "guarantee" --- 3% of free spending money after maintaining capital, adjusted for inflation --- have 1.2B dollars of spending money per year, for perpetuity! You can maintain a decent Engineering and Infrastructure operation for that budget. Make those shares non-voting. 

What people like about the Bay area

I often have a question in mind, a question for which it is not important what is my answer, but I need to know the answers of the human diaspora, not an unique answer but the variations. More than a decade ago, I asked people, "What does the phrase Safe Love mean to you?" and the experience filled a short story I didn't get to publish yet. These days I ask transplants to the Bay Area (not the natives), "What do you like about the Bay area?", and the answers worth remembering have been:
  • Fresh fruits
  • Humming birds
  • Near-zero humidity 

Saturday, July 21, 2012

Oldtime Tweets

I had a conversation with someone who reminded me there were oneliners, before there were tweets; the oneliners didnt need hyperlinks, only a lot of thought from the audience to find the hidden links.
  • What  any research area needs: A Howard Roark.
  • The reason I hate Apple: The Genius Bar.  
  • Terence Tao is the Kevin Bacon of Google+
  • What I look for in the Bay area: NY expats.

Summer School on Markets, Trento

Summer schools, like the one on Market Design at Trento, are mental islands: you get out of your context of every day, meet new people, in particular students,  you learn about some topic and world seems full of possibilities.

The summer school was run by Dan Friedman and David Parkes, who make a powerful Venn Diagram covering CS and Econ among their research interests, with help from Axel Leijonhufvud and Enrico Zaninotto. The Trento schools have a dagstuhlesque history. There were 20+ students from Econ -- CS -- Applied Math spectrum. They formed groups, did research projects during the week, and presented their work on the last day.  Example projects included improved matching markets and auction design with information revelation costs. I didnt get as much time as I wanted to interact with the students on specific research projects, but enjoyed talking to several. It was two weeks long, but I missed much of the first week due to travel to Edinburgh for ICML, so my notes are spotty.

Lectures: Dan F and David P anchored the lectures, providing intro and glue lectures to solder together the lectures of guest lecturers.  Peter Cramton spoke about electricity markets. It is reassuring to have bona fide economists so deeply embedded. Estille Cantillon spoke about mechanisms for matching students to schools in Belgium, this was a great example of Economist in the Field! Tuomas Sandholm spoke about kidney exchanges and  expressive auctions. I lectured on problems in Ad Exchanges.  I liked Dan's definition of recommendation in a utility (and hence, for me, repeated games framework); David's lecture on combinatorial exchanges in which he posed many open problems; and, a talk with Peter Cramton on what were the a handful of fundamental concepts of Economics. Thinking about the marginal effects emerged as a central tenet, and it came up again in a conversation with Zhenyu Lai when he proposed a target for carbon trading that all countries could adopt. 
  • It is so difficult to travel within Europe by air. There are few direct flights and when they are by EasyJet/RyanAir, they run in the middle of the day and the entire day is consumed traveling. My proposal to EU: start a pan-EU airline or two, kill off the struggling ones supported by individual countries that are flailing.
  • The summer school had sparse Internet access. As a result, no, I didnt get more done, I spent more time fretting there was no Internet access.

Tuesday, July 03, 2012

My Blog

The other day someone asked, paraphrased, "when you write a blog entry, what  do you optimize for the audience"?  Google and other search engines optimize so its visitors who pose a query get quickly or instantly to what they want. Yahoo and other portals optimize so visitors instantly view all they care about on a single page. I have a vision I optimize for: I wish the visitors will scan my blog entry coarsely, get a nugget or two of interest, chuckle or frown or perk up, maybe retrace over a portion of the entry to resolve the nugget better, and soon pivot, go on with their every day lives. But the nugget should linger beyond those few seconds they spend at the blog entry, should resurface later in their psyche, and have some long term utility. 

ps: I couldnt think of a metaphor for this effect I want to induce --- slingshot, catapult, pivot --- nothing captures the whirr.

ICML 2012

When I gave a plenary talk at SODA in 03, I had to give the talk a day earlier because the slotted speaker could not make it to the time slot. At ICML, where I was slotted to give the plenary talk at 8:40AM on the first day, I could not make it to the location and the hosts (Andrew McCallum and Joelle Pineau and Charles Sutton), very graciously, scheduled me at 7 PM the same day by carving out a new slot. This tale of two slots worked out well for many: the North Americans who had flown in and had sheepishly slept thru my early AM slot could make it to my evening slot; the attendees had the poster session just before my talk and had a chance to grab some beer and snacks; and, I could go over the standard 50 min --1 hour slot because I was *only* standing between the audience and their dinner.

I spoke about streaming, summarizing the by-now standard material on sketches, l_0 samples and their applications, in particular to compressed sensing and graph algorithms resp. For the latter, I highlighted the work of Ahn, Guha and McGregor on graph connectivity problems using l_0 sketches. Then I discussed new directions: (1) new models of learning algorithms in distributed world where the goal is to minimize the communication (I highlighted recent work by Daume+Phillips+Venkatsubramanian, Blum+Balcan+Fine+Mansour, see here). (2) Continual distributed streaming (highlighted Cormode+M+Yi) and (3) Pan-privacy (I highlighted this work of Dwork, Naor, Pitassi, Rothblum, and Yekhanin). Besides gratuituous jokes, puzzles and comments on recommendations, that was the talk. From the questions afterwards, it seemed like the audience took in the hashing part of it and discussed ways to use it, I hope the other parts will percolate in later. Over dinner:
  • Some in ML learn about developments in TCS world (solely?) by reading Michael Mitzenmacher's blog. Michael: you are carrying us, mate!
  • Like all communities, ICML is tinkering with the conf. One of the suggestions concerned making reviews (think reviewoverflow) public. A central suggestion will be to separate/delink the identities of authors from reviewers (which you can do on the web) but make identity persistent. 
  • Someone asked me how many pairs of shoes (2) and glasses (6) I had while traveling. I said only 2 shoes because very few extroverted mathematicians at ICML to look at my shoes, the ref to the joke was lost in some cases.
  • Finally, I saw printed NY Times in French. 

Collages from Baerden to Bradford

I am a huge fan of collages. Romare Baerden is collages then, pieces of debris creating incredible 3D worlds of neighborhoods, music bands and reality; Collages now is Mark Bradford, an incredibly talented artists who works with wax paper used in hair salons and layered posters and claws out deep, nuanced scapes. Thanks to Whitney for helping world discover Mark.