Archive for August, 2012



August 30, 2012

In a comment to the previous post, Rich talks about bozos.  I’ll have to think about the bozo phenomenon a little bit, but it seems eminently mathematizable.  In particular it seems the concepts of homophily and preferential attachment might come into play.

On a totally separate note, I went to a talk today in Yorktown that mentioned the H_2^* distribution, which is a mixture of a delta function at zero and an exponential distribution.  I didn’t know that there was a name for this distribution, but as noted by Whitt in a paper on queuing theory:

Because the special service-time distribution is an extremal distribution among the class of hyperexponential  (H_2, mixtures of two exponentials) distributions, we denote this class by H_2^*.

I’ve seen this distribution before as the capacity-achieving input distribution for an additive exponential noise channel.  In a paper with Mitya and Jesper, this distributions is also the predicted distribution of synaptic volumes in the mammalian brain.  I’ve been thinking about both of these facts recently, so watch out for that.

And of course, I’ll soon be moving to Yorktown, so watch out for a bozo like me cruising the hall.


The Dark Side of Fraternity

August 28, 2012

I just read a very interesting and clever paper about the downsides of social capital.  It is by James Oldroyd and Shad Morris from The Ohio State University and is titled, “Catching Falling Stars: A Human Resource Response to Social Capital’s Detrimental Effect of Information Overload on Star Employees“.  Using a simple degree-distribution argument for scale-free graphs, the authors argue that although there are significant informational benefits to having a large amount of social capital, people with too much social capital may suffer from information overload.  People with a great deal of social capital are referred to as stars, and e.g. “stars in science fields find it easier to acquire the resources necessary to facilitate research, such as colleagues seeking collaboration, cadres of highly capable students, and access to databases.”  As they go on to say, “in a state of overload, cognitive limitations may constrain the value of a star’s social capital; if the information load goes unmanaged for long periods of time, the star may stumble and, ultimately, fall.”  In order to prevent this, several control actions are suggested, some at the individual level, some at the organizational level, and some at the structural level.  In some sense, the control policies are reminiscent of the kind of things discussed in The Attention Economy.

So there is at least one potential downside to too much social capital, but are there others?  I had finished reading Bowling Alone at Terminal 3 and Putnam was concerned with the same question: what are the potential dark sides to social capital.  In a previous post, I had alluded to the tradeoff between liberty and equality, but he does even better and brings fraternity (social capital) into the picture, somewhat analogous to the notion of duty that I had tried to bring in.  As he says on p. 351:

On the banners of the French Revolution was inscribed a triad of ideals—liberty, equality, and fraternity.  Fraternity, as the French democrats intended it, was another name for what I term “social capital.”  The question not resolved on those banners, or in subsequent philosophical debates, is whether those three good things always go together.  Much of Western political debate for two hundred years revolved around the trade-offs between liberty and equality.  Too much liberty, or at least too much liberty in certain forms, may undermine equality.  Too much equality, or at least too much equality in certain forms, may undermine liberty.  Less familiar but no less portentous are the trade-offs involving the third value of the triad: Is too much fraternity bad for liberty and equality?  All good things don’t necessarily go together, so perhaps a single-minded pursuit of social capital might unacceptably infringe on freedom and justice.

The general conclusion reached in the book is that fraternity does not have deleterious effects on either liberty or equality.  I’m glad.

Incidentally, some of the most interesting statistical evidence marshaled by Putnam is freely available in case you want to go at it.  Some of the data is on a state-by-state basis, but what I think would be particularly interesting is looking at that kind of data on either a city-by-city or MSA-by-MSA basis.  It would be really awesome if you could find it.


go, go, no go

August 14, 2012

I am back from India, which is now renewing its tryst with destiny for the 66th time after coming off perhaps its best Olympics ever.  As you alluded to, I did go to the Dravid and the Himachal.  In Bangalore, I had the good fortune to visit the Tata Institute of Fundamental Research Centre for Applicable Mathematics, the Tata Institute of Fundamental Research International Centre for Theoretical Sciences, the Indian Institute of Science, the IBM India Research Laboratory, and LimberLink Technologies, as well as meet up with Krishna Jagannathan from IIT-Madras and Manoj Gopalkrishnan from TIFR’s campus in Mumbai.  So pretty busy overall, not even counting the tourism within Bangalore, and to Mysore and Srirangapatna.  After a quick stop in Delhi (and a meeting with Aspiring Minds), we were off to Amarnath, which was simply breathtaking in its natural beauty, and also in its spirituality.  The end of my trip was in the familiar comforts of Aligarh, though not long enough.

Although everyone is big on virtual interaction these days, there is nothing that can replace going in person.  I’m sure there are all kinds of studies on this point, including the informational aspects, but let me not worry about linkage. 

What you brought up about incentives, motivation, and whether to use outcomes-based or behavior-based methods is quite interesting.  It seems there may be some interesting connections between that line of thinking and the kinds of results that were presented by my former intern Gireeja Ranade at HCOMP 2012 in July, in a paper  To Crowdsource or not to Crowdsource? that she also recently blogged about.  Speaking of crowdsourcing, while I was away, Wes Gifford was kind enough to present a paper for me on privacy and reliability in crowdsourcing at the SRII Global Conference.  It was really too bad that I could not go.

Your other post about pinterest and all was also quite interesting.  You really should do something along the lines you suggested.  Seriously!

I went ahead and read the article you had linked to and found the following rather interesting.

The site’s name combines the words “interest” and “pin,” in reference to “pin boards,” which are also known in various creative professions as inspiration boards or mood boards — basically a large board onto which appropriated images (torn from magazines!) are juxtaposed to evoke in the viewer a certain feeling, atmosphere or mood. Once the exclusive province of advertising art directors, designers and teenage girls in boarding-school dormitories, mood boards and their electronic equivalents have exploded online.

Perhaps something pinteresty in the area of computational creativity is also warranted?