CI Foo - Day 1

Being at CI Foo is a kind of an out-of-body experience for me. As a "Wall Street guy," my experiences, my mind-set and my DNA is a world away from most of these dudes. That said, given my experiences as an entrepreneur and as an investor over the past 3.5 years, I am in tune with the pulse of the conversation. It just feels odd being part of it and contributing. But I am and it is very, very cool. It is always a thrill being around so many super smart people whose intelligence is so different than mine. I love filling in gaps in my knowledge by being near people who are smarter than me. And there are plenty of those people here at CI Foo.

I sat in on three sessions today, presented at one, got to spend some quality friend with my friend Paul Kedrosky and had numerous conversations with people from Google, Microsoft, Yahoo, Amazon, MIT, Carnegie-Mellon and other equally impressive institutions of commerce and higher learning. Being at Google is cool as well. Larry Page stopped by to say hi and the environment at Google Ground Zero is simply terrific. One of the amazing things about being in such a setting is that people are very disclosing, far more disclosing than they'd be in "normal" life. And it is particularly impressive that top professionals from Microsoft, Yahoo and News Corp. can co-exist at this conference, participate in conversations, grab a beer, hang out, etc., without any strange feelings arising from the goings-on in M&A land. It is a tribute to Tim O'Reilly and his team for creating an environment for sharing, openness and safety. This is how great ideas are exchanged and foundations laid for future work. It is so different than Wall Street. It truly blows my mind.

10am - participated in a session led by Gary Flake of Microsoft Live Labs on the topic of editorial gatekeepers at the head-end of the distribution vs. long-tail recommenders, their dynamics and movement across the distribution. Paul Gould of Fox Interactive, Joshua Schachter of Yahoo, Dave McClure of 500 Hats and Paul Resnick of University of Michigan were all key participants in the discussion. Peter Bloom of General Atlantic, one of my Wall Street-to-venture investor heroes also shared his incredibly valuable two cents. We spent a lot of time discussing issues of reputation and how reputation is established, how to reduce transaction costs associated with reputation and how to quantify subjective/fuzzy recommendations. Search, music and job boards were all used as examples to tease out the ability of long-tail media and services to move towards the head end. I wouldn't say that the group emerged with any definitive conclusions, but he discussion was spirited and fascinating.

130pm - gave a presentation on Making Money from Collective Intelligence (CI). I dove into the issue of expert networks and proposed a third way of making money from CI (with prediction markets and conventional expert networks being the first two). I will work to put my slides up sometime soon. I was followed by Dave Leinweber of the Haas School of Business at Berkeley and Rob Passarella of Bear Stearns, with Dave speaking about applications of computational finance to different data sets and Rob proposing a model for making money from alternative data. Mike Kearns of University of Pennsylvania and Bank of America and Peter Bloom both had interesting perspectives on the issues. Mike, who is a brilliant guy and a hard-core quant focused on machine-learning and automated processes for parsing news, found our bottoms-up, research-oriented approach anathema to his data-heavy, readily quantifiable model. We agreed to disagree on the value of context-specific, non-quant data. I personally think both approaches can generate alpha; they just reflect two different ways of managing money. It was fun to so actively contribute to a discussion. I felt like I was now beginning to pull my weight as a CI Foo participant.

330pm - sat in on a session chaired by Paul Martino of Aggregate Knowledge. We spent most of our time discussing which data sets are most predictive, as well as the relative value of explicit versus implicit data. Super interesting, highly cerebral yet very pragmatic discussion. Andreas Weigand of Stanford/Berkeley/INSEAD, Peter Norvig of Google and Eric Zitzewitz of Dartmouth were big contributors to this dialog. There was a big battle as to whether explicit data was tainted due to Heisenberg's Uncertainty Principle and therefore less predictive than implicit data, or if the diversity of perspectives offered by explicit data far outweighed its potential corruption. I think the group ultimately agreed that both implicit and explicit data are extremely valuable and have their place, and that it is difficult to put the value of one above the other without the benefit of context.

5pm - participated in a discussion run by Paul Resnick of Michigan on the application of CI to solve big social problems. Many suggestions were tossed out ranging from collecting more granular data on our health, our diet and our environment in order to data mine and draw conclusions about ways in which people can live better and live longer; to prioritizing needs within communities involving government support; to bringing greater transparency to government. There were far more big thoughts tossed out than I am reflecting here but given my lack of sleep last night it is hard for me to pull anything more out of RAM.

In sum, it is great to be here. I know that I am getting a lot out of my fellow campers and the discussions we are having. I hope they feel similarly. I will report back after Day 2. It should prove to be a barn-burner!

Comments

[nba]

Awesome! Thanks for taking the time to post the details from such a great experience. I recently discovered your blog, super interesting! Keep it up and best wishes.