Dennett on the Evolution of Religions

In Wild and Domesticated Religions: How the Machinery of Religion Evolved, Daniel Dennett at the Santa Fe Institute discusses how religions evolve and how they should be studied as a natural phenomenon.  Here’s the video.



I was intrigued by Nicholas Christakis’ recent TED talk on social networks.  After explaining how people’s social networks influence behavior such as smoking and showing some striking statistics on it, he goes into a discussion on the advantages and disadvantages of social networks. He suggests that, overall, social networks must be good for us (“social networks are fundamentally related to goodness”) because otherwise they would not be so pervasive, and concludes that “what the world needs now is more connections”.

I disagree (though with the current obsession on rather shallow online social networks, the conclusion is not surprising). More connections can be a good thing, but not necessarily. Would you rather have 10 good, supporting friends, or 1000 “facebook friends” that you never talk to?  The quality of the social network and the kind of links seem to me to be even more important than how large the network is, or the level connectivity.

Some online social networks seem to provide mostly virtual connections (Facebook), whereas others (such as CouchSurfing) seem to be more effective at bridging the gap between the virtual world and the real one.  What causes the difference? I think there’s two main factors:

  • The kind of people in the network, and their expectations of the network.  The network will be more valuable with more participants and less spectators
  • The kind of tools: if there are tools that help build trust, or tools for collaboration, then the network will be far more valuable

There’s a point at which having more contacts is meaningless. Online social networks extend the number of people we have instant access to, but unless there is some way to harness the power of the unnatural number of connections, what good does it really do us?

I do think the world needs more real connections, and that the available technology has a huge potential for making these connections do something useful. Here is one example: Kiva Microfunds connects people to those in need of micro-lending.  Here is another: No Shortage of Work helps connect people in projects that are mutually beneficial (the idea being that it is better for an unemployed person to work for free and gain experience than to do nothing).

What is often missing is the right space or right tools for the right people to interact.  I think social networks, if well designed, could provide the space and the tools. A social network, rather than being something you just belong to, could be something you can act in, and that can act for you.

Museum of Playing Cards

(May 14, 2010)
A few days ago I visited the Museum of Cards in Vitoria-Gasteiz. It had examples not only of Heraclio Fournier’s playing cards, but also of cards from different countries and time periods: Spain, France, England, Japan, China, from the 10th century (when playing cards where invented in China!) to the present day.

But the museum was most interesting because of the variety of kinds of cards, a variety that no longer seems to exist.  There were circular cards, playing cards commemorating historical events, cards with maps to help people learn geography, Tarot cards, and my favorite: musical cards.  Each of these contained part of a musical score, and so by combining them differently different melodies could be played.

I think creative playing cards could be used effectively in schools, to aid the learning process in a fun way. They could be used for anything with a somewhat modular structure or lots of connected facts: like music, language, history or math. It would be fun to think up a game involving playing cards with axioms and theorems.

Electric howl

I saw an interesting sight yesterday: Bayou, the cute three-legged dog at the farm, was imitating the sound of what was probably a chainsaw in the distance.  Each time he heard the sound he would howl after it, but it was an unnatural, electric howl.  I ran off to grab my camera to film it, but it was too late. I wonder what he was thinking.  Did he think it was it some kind of dog, or another animal, calling to him from afar?  If so, what was it saying?

Did something in my past create a hole?

The title comes from Porcupine Tree lyrics, and reflects how I feel as I try to restart this blog.  I am sitting at a cybercafe in southern France, and I decided it was time to revive my blog.  It is odd to read the blog since I have not used it in months now: it is as if I am reading about another person.  I don’t have time at the moment to summarize what I’ve done or seen or read or why I stopped blogging… but I expect it will happen slowly over the next few posts.  I am heading to Spain next week, and so am trying to use my time in France as effectively as possible.  Practicing French will be much harder from now on.

Mumbai, India

I arrived in India 3 weeks ago.  To write about my experiences and thoughts of Mumbai and India was difficult, as I was thrown right in the middle of it and needed time to figure things out.  Talking Heads lyrics come to mind when I look back on how I felt: “How did I get here?”, and “Where does that highway go?”  I had no map and really had no idea where I was for the first week and a half.

I am writing from my small room in the guesthouse, drinking tea.  The tea here is so good, I have always enjoyed it, but now I can have it every day!  The guesthouse is for company employees, and while I knew I would be put up in one, I did not expect there to be caretakers who are attentive to every single detail and who also cook for us.

Mumbai struck me as a frenzied, chaotic city of movement, color, and noise.  People tell me that it is the “New York” of India.  Once important difference, however, is that unlike New York, Mumbai is lacking in infrastructure.  The city desperately needs a subway to help with the traffic problem (and one is being built, but it will take years to finish).  I think the government should invest money in improving the roads as well (the roads here are some of the worst I have seen anywhere, and that includes Mexico and Ecuador).

This is an interesting thought experiment: what would happen if the governments of India and China were switched?  How much would things change?  One of my colleagues remarked that things are more orderly and efficient in China because it is a dictatorship, though I don’t think it is really so simple.

Continue reading

Time Series and Forecasting

“What follows is ever closely linked to what precedes; it is not a procession of isolated events, merely obeying the laws of sequence, but a rational continuity.” — Marcus Aurelius, Meditations, Book IV,  45

I am now working with TCS on a project on forecasting demand (I will write more details later).  It is a great opportunity to learn (and apply) more statistics and time series analysis, and this also has connections with lots of interesting areas from dynamical systems to machine learning techniques.  I suspect I will learn more statistics than if I were in a course, and it will be more fun too.  (I think that college statistics is generally taught extremely poorly, and can be improved in many many ways, but to go into details of this will require another post).

The problem: given we know the past history of some variable(s) , what is the best way to predict the future value(s)?  What “best” means depends on the application.  Usually it means low expected out-of-sampled error (measured as RMSE for instance), though in a business context I think it is better to use loss functions that are better suited to the area of application.

I am particularly interested now learning more about the problem of model selection and how to prevent over-fitting.  How do you find a model that will give the lowest out-of-sample error?  Validation or cross-validation is the most obvious and least sophisticated way and is very commonly used, but there are some issues about it that worry me (example: if partitioning a data set into training set and a test set, how does one decide how much data to include in the test set, and how much to include in a test set? Surely this is an important decision!  If anyone knows anything about this please let me know.)  There are lots of other things I should mention: how prediction markets work, exponential smoothing…