Complexity and Social Networks Blog

17 November 2009

twitter? Hell no! It's time for twitteleh!

It's time for twitteleh ---- twitter for your Jewish Mother .......

And then PLEASE answer all the questions!

Why can't you be more like your brother?

Now THIS is something you can really do network research on ........ think of all the wonderful homophily/social influence hypotheses you could test on twitteleh networks!

By Stan Wasserman | 5:14 PM | Comments (0)

16 November 2009

Cambridge Colloquium on Complexity and Social Networks: Michael Macy on "The Length of Weak Ties"

A short schedule for CCCSN this term, with Michael Macy on November 23 and Carter Butts on December 10. The details on Macy:

Michael Macy (Cornell)

The Length of Weak Ties

November 23, 12-1:30pm
440 Huntington Ave. ~ 366 West Village H

Granovetter's theory of "the strength of weak ties" is one of the most-cited social theories in network science and an important precursor for Watts' and Strogatz' discovery of small world networks. The "strength" is that information about economic opportunities is more likely to come from socially distant acquaintances than close friends. Until recently, however, the structure and strength of social ties has been almost impossible to measure at the societal level. We test predictions of this theory by combining the most complete record of a national communication network studied to date with data on the socio-economic well-being of communities and the purchasing behavior of individuals. We show for the first time that the diversity of individuals' ties has a strong positive correlation with the social and economic rank of the community (R2=.78). Moreover, contrary to theoretical predictions, clustered relations were no stronger than ties that bridge between clusters and yet were more effective as conduits of peer influence. We speculate that ties within clusters are mainly friendship-oriented, while those between clusters are mainly task-oriented.

Three other talks of interest this coming week:

The College of Computer and Information Science presents a Colloquium by:

Madhav Marathe (Virginia Tech)
Computational Network and Social Science: Implications for Public Policy

Date: Wednesday, November 18
Time: 2:00pm
Place: 366 West Village H

Title:

Abstract:
Complex Networks are pervasive in our society. Realistic biological, information, social and technical networks share a number of unique features that distinguish them from physical networks. Examples of such features include: irregularity, time-varying structure, heterogeneity among individual components and selfish/cooperative
game-like behavior by individual components. Furthermore, the network structure, the dynamical process on the network and the behavior of constituent agents co-evolve over time. The size and heterogeneity of these networks, their co-evolving nature and the technical difficulties in applying dimension reduction techniques commonly used to analyze physical systems makes the task to understanding and reasoning about these networks even more challenging.

Recent quantitative changes in high performance and pervasive computing including faster machines, distributed sensors and service-oriented software have created new opportunities for collecting, integrating, analyzing and accessing information related to such large complex networks. The advances in network and information science that build on this new capability provide entirely new ways for reasoning and controlling these networks. Together, they enhance our ability to formulate, analyze and realize novel public policies pertaining to these complex networks.

Over the last 15 years, our group has established a theory based program for modeling, simulation and associated decision support tools for understanding large complex network. Complementing this is a program to develop a scalable service delivery framework, that provides policy analysts and scientists seamless access to the modeling
environment. After a brief overview, I will describe our approach within the context of a specific application: development of modeling and decision support environments to study epidemics in co-evolving social and wireless networks. Understanding these
epidemiological processes is of immense societal importance. Additionally they serve as excellent "model organisms" for developing a theory of co-evolving complex networks. Individual and collective behavioral adaptation is critical in these systems and will be highlighted via illustrative case studies.

Nov 18th, 3:30-5:00 pm
Harvard Jason Greenberg, MIT
Lifeblood or Liability?: Schumpeter, Stinchombe, and the Double-edged Sword of Strangers or Strong ties in the Startup Process
1550 William James Hall.

Who should you start a business with, individuals you trust such as family members or friends or strangers who are more likely able to provide access to distinct resources? In one of the most influential arguments in organizational sociology Arthur Stinchcombe argued that new as opposed to old organizations are more likely to die because of a "liability of newness." The general thesis has received empirical support. However, Stinchcombe identified four mechanisms that individually and collectively compose the liability. One of the liabilities he identified holds that new organizations are more likely to die because they must rely upon relations among strangers, and strangers are less likely to trust each other. Trust, in turn, is an essential ingredient in economic transactions that entail risk and uncertainty, which are central elements of any startup. This "liability of strangers" mechanism has not been evaluated empirically in startups. On the other hand, research suggests that strangers are particularly well suited to act as bridging ties that afford advantages in the startup process by offering access to diverse information about market opportunities and distinct resources. This social structural mechanism is consistent with Schumpeter's view of entrepreneurship as novel combination and Simmel's theorization of the social position of stranger. This paper assesses whether including strangers or strong ties (e.g., family, friends) on a founding team is net positive or negative in terms of predicting the fraction of business-relevant milestones achieved, dissolving the venture, or achieving viability. Results consistently show that starting a business with friends from outside work is associated with negative outcomes, whereas starting a business with friends from work is associated with positive outcomes. There also does not appear to be a liability of strangers. In point of fact, founding teams in knowledge or research intensive firms that include stranger dyads in the founding team are more likely to accomplish business-relevant milestones.

Nathan Eagle, (Santa Fe Institute & MIT)
Big Data, Global Development, and Complex Social Systems
Friday, November 20th at 2:30 PM
440 Huntington Ave. ~ 366 West Village H
Boston, MA 02115

Petabytes of data about human movements, transactions, and communication patterns are continuously being generated by everyday technologies such as mobile phones and credit cards. This unprecedented volume of information facilitates a novel set of research questions applicable to a wide range of development issues. In collaboration with the mobile phone, internet, and credit card industries, my colleagues and I are aggregating and analyzing behavioral data from over 250 million people from North and South America, Europe, Asia and Africa. I will discuss a selection of projects arising from these collaborations that involve inferring behavioral dynamics on a broad spectrum of scales; from risky behavior in a group of MIT freshman to population-level behavioral signatures, including cholera outbreaks in Rwanda and wealth in the UK. Access to the movement patterns of the majority of mobile phones in East Africa also facilitates realistic models of disease transmission as well as slum formations. This vast volume of data requires new analytical tools - we are developing a range of large-scale network analysis and machine learning algorithms that we hope will provide deeper insight into human behavior. However, ultimately our goal is to determine how we can use these insights to actively improve the lives of the billions of people who generate this data and the societies in which they live.

By David Lazer | 5:26 PM | Comments (0)

15 November 2009

A Call for Papers on Twitter networks at the 2010 Sunbelt Social Networks Conference

At the 2010 International Sunbelt Social Networks Conference, to take place at Riva del Garda, Trento, Italy, from June 29 to July 4, 2010, we intend to organize a session on Twitter networks. We are looking for papers with empirical evidence using the social networking and micro-blogging service Twitter.com. Twitter has become one of the most prominent publishing channels of short messages during the last year and caught our attention as network researchers. We would like to initiate a session on how network researchers use the public conversations observed on Twitter and how they capture, analyze and interpret Twitter networks.

Potential contributors should also note that at the Sunbelt conferences series, final (written) papers are not requested to be submitted: just abstracts suffice. However, all submitted abstracts of the proposed papers will be reviewed by the INSNA organizers (and not by the session organizers), who may also decide on the format of the paper presentation (as lectures of 20 minutes or posters of 60 minutes). In any case, since we would like to coordinate the Twitter networks session, potential contributors in this session are free to contact us before the end of November 2009, as the deadline for submission of abstracts of contributed papers is December 1, 2009. Nonetheless, the abstract submission should be done by the contributors themselves who should complete the Submission Form here. Since the name of the session "Twitter networks" is not listed in the Session field of this Form, applicants should insert it in the "New Session" field.

Session initiators:
Ines Mergel, Maxwell School at Syracuse University, NY
Moses Boudourides, Department of Mathematics, University of Patras, Greece
Lothar Krempel, Max Planck Institute for the Study of Societies, Germany
Marc Smith, Connected Action Consulting, Belmont, CA

By Ines Mergel | 7:59 AM | Comments (0)

13 November 2009

David is a Hero

Bravo to David L, for standing up, intellectually and otherwise, to the senseless and
politically motivated attack on not only his research, but on Political Science research
financed by NSF.

His account of the last two weeks, presented here, is a great read. He is my choice for "Player of the Week".

By Stan Wasserman | 11:09 AM | Comments (1)

12 November 2009

Endorse the Open Declaration on European Public Services

A while ago, I blogged about the complexity of government 2.0. I described the efforts by a group of people to create an open declaration on eGovernment in Europe in the coming years alongside the official declaration which will be presented during the 5th EU Ministerial conference in Malmoe, Sweden next week. To learn more about participation in the age of 2.0 I joined the group. It was neither easy to fit quality contribution time into ones schedule nor getting people to participate in the various stages of the creation of the declaration.

There are 6 days left to get another 800 [or more] supporters of the open declaration on European public services. If you like it, please endorse it by visiting this website and/or spreading the word through the channels available to you.

I will send updates directly from the conference in Malmoe next week.

Update: The official declaration's scope and structure might be as follows:

Common vision
"We, the ministers, agree to create an e-enabled European Union where citizens and businesses can easily access public services and public information all over the Union, enjoy true freedom of movement on the Single Market, and where administrations easily collaborate across boundaries in order to fulfill this goal."

Priority 1: E-government Empowering Citizens and Businesses
"We, the Ministers, agree to work pro-actively in order to make it easy for citizens and businesses to engage with government administrations at anytime and anywhere in the EU in order reap the benefits of the Union, have easy access to public information and feel secure that their integrity and information is protected when engaging with administrations over the Net."

Priority 2: E-government in Support of the Single Market
"We, the Ministers, agree to make it easier for European citizen to travel, study, work, retire and reside in all countries of the Union and make it easier for businesses to provide and procure services everywhere in the Union. Consequently appropriate legal and technical preconditions for cross-border e-services needs to be created and joint demand-driven e-government projects needs to be supported. Such projects should be centrally monitored in order to seek synergies and prevent overlap with other ongoing projects and to avoid the creation of new barriers in the internal market."

E-government Enabling Administrative Efficiency and Effectiveness
"We, the Ministers, agree to make it easier for public administrations in the EU to collaborate across boundaries by creating a European Common Area of e-Government. This new area of partnership and joint action between Member States should enable European public administrations to interact easily and connect intelligently with each other as well as with private actors in order to deliver personal and overall public value to society and facilitate the implementation of European Directives. Within this area experiences should also be shared on how to reduce the administrative burden, reduce the carbon footprint and facilitate organizational change."

The details would be described in an action plan.

By Alexander Schellong | 10:45 PM | Comments (0)

5 November 2009

Final thoughts for now on the Coburn attacks on the online townhall report

This is hopefully my last post for a while on the online townhalls. I do think there is a value in dialog and discourse, and I wanted to excerpt the critiques from some of the more credible online sources and provide my responses, more for posterity than anything else.

I would first note that none of the posts responded to the findings that the online townhalls reached people who do not show up to regular townhalls (based on demographics), where, notably, people who were frustrated with the political system were more likely to show up. Further, none of the criticisms deal with the findings that the people who participated were subsequently more engaged in politics, more likely to vote, increased their knowledge of the policy, more likely to follow the election. These are the parts of the report that actually have normative bite; clearly, approval of Members of Congress by itself is neither here nor there normatively. These may be things that the bloggers below do not value (in which case they should explain why) or actively ignored.

In any case, here is what they did say.

First, the Heritage post, in its words:

[referring to the fact that the research methodology called for offensive questions to be culled] CMF does not say what qualifies as offensive, but if this summer is any indication that definition would include anything that the Congressman did not want to talk about. In other words, this report urges Congressmen not to actually interact with their constituents, but to avoid them altogether by holding safe townhalls they can completely control. And what did CMF find where the results of these Potemkin townhalls?

The online town halls increased constituents' approval of the Member. Every Member involved experienced an increase in approval by the constituents who participated. The average net approval rating (approve minus disapprove) jumped from +29 before the session to +47 after. There were also similar increases in trust and perceptions of personal qualities - such as whether they were compassionate, hardworking, accessible, etc. - of the Member.

The lesson: avoid your constituents' inconvenient questions and your approval ratings will rise. And this is a taxpayer funded study. Here is the grant from the National Science Foundation.

Congress is actually using your tax dollars to pay social scientists to find ways they can avoid actually talking to their constituents while improving their chances of reelection.

Response: as noted in the report, the possibility of screening anything as "offensive" was theoretical. We did not actually exclude any questions for this reason (we did state this in a footnote rather in the text). We had pretty high (low?) standards for offensive--e.g., we would not post questions that included expletives. It was part of our research protocols, and thus our instinct was to mention it.

That said, it is worth noting that the medium is potentially manipulable, and there is nothing to stop someone who is doing an online townhall from excluding difficult questions. (Of course, all communication media are manipulable in some way, so it is not obvious that this is an advantage or disadvantage of online townhalls.) We had a neutral moderator, and included all questions that time would allow, in the order that were posted. This included some that were pretty hostile to the Member. Our assessment (and recommendation) was that these very confrontations made the events more effective, because they reflected the authenticity of the event. In short, the Members approval ratings increased because they had done the right thing.

The Wall Street Journal:

The National Science Foundation prides itself on making research grants that lead to path-breaking discoveries. So it seemed odd to Coburn, a physician known as the Senate's 'Dr. No,' that science foundation money was being used to show legislators how to exile angry town-hall mobs to cyberspace.

Response: It is unclear whether the blogger is speaking in her voice, or Senator Coburn's. In any case, the criticism that we were trying to eliminate traditional townhalls came up repeatedly through the blogs. We do not advocate this in the report (or elsewhere). But the online and tele-townhalls do allow Members to reach many more people than they can via traditional townhalls, and people they could not otherwise reach. Indeed, Senator Coburn himself has participated in many tele-townhalls, presumably without crowding out traditional townhalls.

From Fox (John Stossel's blog):

This summer's town hall meetings made many congressmen and senators uncomfortable. No worries. The sycophants they fund have used your tax money to fund a study that advises politicians how they can avoid seeing you altogether.

[The rest is just derivative from the Heritage blog]

Response: All of the NSF funded data collection was conducted in the summer of 2006, three years before the town hall meetings of this past summer. Further, the participating members included some conservative Republicans, i.e., there was no partisan or ideological tilt in this research.

Otherwise, I'd note that name calling is a poor substitute for a persuasive argument. 'Nuff said.

----------------------------
The more serious argument is the resource constraint argument that Coburn himself raises in the flyer below. Taxpayer dollars are wrested, involuntarily, from individuals who have worked hard for their money. Should these dollars be used to fund political science research, especially in these resource constrained times? Could these funds be used, for example, to help cure cancer (something that the Coburn amendment did not mandate, I should note). (NB: a useful fact, we currently spend more than 2000 times as much money on NIH than on the NSF political science program.) Does this research provide public value equal or greater than its cost?

This requires a more thorough answer than I can give here and now. The broad intellectual argument for giving even one penny for research from the government is that certain types of knowledge, for which there is no mechanism for intellectual property right protection, will be underproduced in our market system. And, to the extent it is produced, it might remain proprietary in a fashion that reduces its possible positive impact. For example, in the political arena, I am sure Senator Coburn (and every other Senator) has spent tons of money on politically related research by consultants, etc. (across all politicians, vastly more than is spent on the NSF Political Science Program). But little of that money is aimed at producing knowledge that might improve democracy, nor would any insights along those lines be made publicly available.

To take the boundary case for supporting political science research, say the authors of the Federalist papers, the inspiration for our research, had been drawing government salaries on the time that they were writing those foundational papers (if any readers of the blog can speak to this historical fact, I would be interested in knowing). Would Senator Coburn say that this had been a waste of taxpayer money?

So, in the case of our more modest research, we produced something of a model for how democracy could be done, with the Internet facilitating direct interactions between Members of Congress and citizens that they had, before, been unable to have a conversation with. We applied cutting edge scientific methods to test the effects--field experiments, with control groups, the whole nine yards. Any criteria you might throw out there regarding "scientific methods" we match-- rigor, research design, replicability, inferential power, etc etc. And we got some really compelling results from a normative point of view. I am not sure what the dollar value was, but I am happy for our report to be held out as a poster child for the public value that federally funded research for political science can produce. And on the more general issue, presumably enhanced understanding of the inner workings of democracy, the causes of war and peace, of the difficulty/challenges of institution building in damaged states, of the causes of the growth and decline of terrorist networks (all things that have been federally funded political science research) are all things that plausibly produce public value, and are worth more rigorous treatment than pundits and talking heads on Fox, CNN, and MSNBC might provide.

By David Lazer | 7:50 AM | Comments (1)

3 November 2009

Following the digital breadcrumbs...

A reader of the blog found an antecedent for the photograph on the Coburn flyer:

can be found at Progressive States:

Putting aside the small irony here, there is an interesting lesson here about being able to track the linkages among objects in digitized information, and, in turn, what those linkages reveal. More on that another day.

By David Lazer | 1:39 PM | Comments (1)

1 November 2009

Network picture of attacks on online townhall report

Readers of this blog will know that last week I summarized an NSF and Harvard funded report on which I was lead author on online townhalls. Our findings included: if a Member of Congress reaches out to constituents in an online forum, it will reach people who tend not to participate in politics, and especially individuals who are frustrated with the political system, it will affect those constituents' views of the Member and increase their policy knowledge and political engagement. Well, this was picked up on by Senator Tom Coburn (who is trying to defund the Political Science program in NSF) last Wednesday who hammered the report as a waste of taxpayer money. A flyer his staff passed out at a briefing we conducted to Congressional staff on Friday captures the spirit of the critique:

(would be interested in the provenance of the picture, in case any readers know)

At the moment I will not address the merits of the criticisms, but focus instead on the interesting diffusion process that followed from the initial criticism from Coburn. Each day it was picked up by another few blogs. A quote from John Stossel provides a sense of the tone of the postings: "This summer's town hall meetings made many congressmen and senators uncomfortable. No worries. The sycophants they fund have used your tax money to fund a study that advises politicians how they can avoid seeing you altogether." Initially, I would infer, the first few blogs must have been on some distribution list from Coburn's office (i.e., they weren't just watching his website) because there were quotations from materials from Coburn that were not on his website. Thereafter you could see how different blogs picked up on the story, typically quoting or copying from another blog. So what one sees is a signal propagation process through the blogs. And as the signal propagates it evolves. Thus, for example, Stossel quotes from the Heritage blog, but then adds his distinct emphasis. The link and copying structure reflects the attention each blogger is paying to other blogs, however one would guess that each blog has a different but overlapping audience. Here is a picture of the diffusion process to date:

Node colors correspond to dates (28: white, 29: light gray, 30: dark gray). Time flows left to right, where the variations within each day reflect publication time of day, but only in an "eyeballing" sense. Link weights are encoded white: explicit mention, black: shared text, grey: both. Arrows point from destination node to source

What one cannot see, but is certainly reflected here somehow is that there is an interpersonal network among these bloggers (and relevant nonbloggers). However, this micro case study does suggest ways that one might disentangle the informational ecosystem among blogs, by looking at shared text, link structure, and to look at it dynamically and by content. If you know of interesting papers along these lines, please feel free to post in comments. One relevant, freshly minted paper, is:

Carter Butts and B. Remy Cross, "Change and External Events in Computer-Mediated Citation Networks: English Language Weblogs and the 2004 U.S. Electoral Cycle", Journal of Social Structure 10, 2009.

By David Lazer | 7:02 PM | Comments (10)

26 October 2009

Papers on online deliberative field experiments

There might be some interest in the scholarly papers undergirding some of the research in the aforementioned report. Below we list some of the papers from the online deliberative field experiments that we posted on SSRN.

Who Wants to Deliberate - and Why?

Michael A. Neblo
Ohio State University - Department of Political Science

Kevin M. Esterling
University of California, Riverside - Department of Political Science

Ryan Kennedy
University of Houston - Department of Political Science

David Lazer
Northeastern University - Department of Political Science; Harvard University - John F. Kennedy School of Government

Anand E. Sokhey
University of Colorado at Boulder - Department of Political Science

Interest in deliberative theories of democracy has grown tremendously among political theorists over the last twenty years. Many scholars in political behavior, however, are skeptical that it is a practically viable theory, even on its own terms. They argue (inter alia) that most people dislike politics, and that deliberative initiatives would amount to a paternalistic imposition. Using two large, representative samples investigating people's hypothetical willingness to deliberate and their actual behavior in response to a real invitation to deliberate with their member of Congress, we find: 1) that willingness to deliberate in the U.S. is much more widespread than expected; and 2) that it is precisely people who are less likely to participate in traditional partisan politics who are most interested in deliberative participation. They are attracted to such participation as a partial alternative to "politics as usual."

Means, Motive, & Opportunity in Becoming Informed About Politics: A Deliberative Field Experiment with Members of Congress and Their Constituents

Kevin M. Esterling
University of California, Riverside - Department of Political Science

Michael A. Neblo
Ohio State University - Department of Political Science

David Lazer
Northeastern University - Department of Political Science; Harvard University - John F. Kennedy School of Government

Survey research on political knowledge typically measures citizens' ability to recall political information on the spot, and in these surveys most citizens appear appallingly ignorant. Deliberative theorists emphasize, however, that citizens' capacity to become informed when given a motive and opportunity to participate in politics is equally important for democratic accountability. We assess this capacity among citizens using two deliberative field experiments. In the summer of 2006 we conducted a field experiment in which we recruited twelve current members of the U.S. Congress to discuss immigration policy with randomly drawn small groups of their constituents. In the summer of 2008, we conducted a similar experiment using a large group of constituents interacting with Senator Carl Levin of Michigan on detainee policy. Using an innovative statistical method to identify average treatment effects from field experiments, we find that constituents demonstrate a strong capacity to become informed in response to this opportunity. The primary mechanism for knowledge gains is subjects' increased attention to policy outside the context of the experiment. This capacity to become informed seems to be spread widely throughout the population, in that it is unrelated to prior political knowledge.

Estimating Treatment Effects in the Presence of Noncompliance and Nonresponse: The Generalized Endogenous Treatment Model

Kevin M. Esterling
University of California, Riverside - Department of Political Science

Michael A. Neblo
Ohio State University - Department of Political Science

David Lazer
Northeastern University - Department of Political Science; Harvard University - John F. Kennedy School of Government

If ignored, non-compliance with a treatment and nonresponse on outcome measures can bias estimates of treatment effects in a randomized experiment. To identify treatment effects in the case where compliance and response are conditioned on subjects' unobserved compliance type, we propose the parametric generalized endogenous treatment (GET) model. GET incorporates behavioral responses within an experiment to measure each subjects' latent compliance type, and identifies causal effects via principal stratification. We use Monte Carlo methods to show GET has a lower MSE for treatment effect estimates than existing approaches to principal stratification that impute, rather than measure, compliance type for subjects assigned to the control. In an application, we use data from a recent field experiment to assess whether exposure to a deliberative session with their member of Congress changes constituents' levels of internal and external efficacy. Since it conditions on subjects' latent compliance type, GET is able to test whether exposure to the treatment is ignorable after balancing on observed covariates via matching methods. We show that internally efficacious subjects disproportionately select into the deliberative sessions, and that matching does not break the latent dependence between treatment compliance and outcome. The results suggest that exposure to the deliberative sessions improves external, but not internal, efficacy.

By David Lazer | 7:32 PM | Comments (0)

25 October 2009

Online Townhall Meetings: Exploring Democracy in the 21st Century.

I am pleased to announce the release of the report, Online Townhall Meetings: Exploring Democracy in the 21st Century. As noted below, this report summarizes the results of a series of randomized experiments, involving 13 Members of Congress meeting with constituents in small groups online. What were the quality of these sessions? What impact did these sessions have on participants? The results were quite heartening. Some of the key findings of the report:

• The meetings increased engagement in politics. Participants in the sessions were more likely to vote and were dramatically more likely to follow the election and to attempt to persuade other citizens how to vote.

• The discussions in the town hall meetings were of high quality. By standards of deliberative quality (use of accurate facts to support arguments, respect for alternative points of view, etc.) the discussions were of a very high quality.

• The town hall meetings attracted a diverse array of people. These sessions were more likely than traditional venues to attract people from demographics not traditionally engaged in politics and people frustrated with the political system.

• The sessions were extremely popular with constituents. A remarkable 96% of participants said they would like to be included in similar events in the future.

• The online town hall meetings increased constituents' approval of the Member of Congress. Members experienced an average net approval rating jump of 18 points. There were similar increases in trust and perceptions of personal qualities such as hardworking and accessible. The sessions also increased constituents' approval of the Member's position on the issue discussed.

• The online sessions increased the probability of voting for the Member. The probability of voting for the Member was 49% for control subjects and 56% for people who participated in a session, with a particularly dramatic impact on swing voters.

• The positive results were seen in small and large sessions. Most of the sessions were conducted by Representatives with small groups of 15-25 constituents. To test the scalability, the team conducted one session with a Senator and nearly 200 people. All of the major results were replicated with this larger group.

By David Lazer | 10:45 PM | Comments (1)