NETWORK EFFECTS

Clay Shirky (clay.shirky@nyu.edu)

H79.2299

http://stage.itp.tsoa.nyu.edu/~cs97/networkeffects/

Syllabus Version 1.0 / August 31, 2005

OVERVIEW (Jump to Classes)

Networks exist in the spaces between things; they are made up not of concrete entities (computers, people, businesses) but of abstract relations (protocols, friendships, contracts). We understand intuitively that networked computers differ from standalone computers, or that a group of friends differs from an aggregation of anonymous individuals, but describing that difference is hard.

 

In particular, networks exhibit emergent characteristics that cannot be explained by simply examining their constituent parts. People are not like computers, but networks of people and networks of computers behave alike in many ways. It is the organization of the respective networks that creates those similarities. Network Effects is a class about the structure of networks, and about how that structure affects human experience. We will focus in particular on two "invisible" networks -- communications networks (especially the internet and phone networks), and social networks.

 

Communications networks are invisible in the traditional sense; their inner workings are hidden inside devices, behind walls and underground, or pass silently through the air. We will examine a variety of electronic networks -- telegraph, telephones, internet -- and design philosophies -- client-server, lattice, peer-to-peer -- and explore the ways different networks alter the experiences that are and are not possible within them.

 

Social networks are invisible in a different way; because we are immersed in myriad social networks -- friends, family, work school -- and because humans are so natively good at understanding and working in such networks, we don't see them. We will examine some of the structural elements of social networks, such as strong and weak ties, clustering, and small worlds networks, to understand some of the ways that the shape of social networks affects us.

 

The goal of the class is to synthesize our observations about these two types of networks. Technological choices embodied in electronic networks profoundly affect their social dimensions: Why can we CC people on email but not on phone calls? How does the one-way network of television differ from the two-way network of the internet? What effect does bittorrent's architectural decentralization have on its users? Social choices also affect the design of technology; resistance to spamming or attempts to hide from the RIAA have led to several deep technological changes in the design of weblogs and file-sharing networks respectively, changes that alter the social relations among the users. This interplay between social and technical adjustment creates complex feedback loops.

 

The class meetings will center on discussion of readings and lectures. Outside class, you will complete two short papers, a mid-term project, and a final paper or project. In the the first paper you will compare and contrast real-world networks. The second will discuss a social situation of your choice. The mid-term project will involve designing and implementing a small network (though not necessarily a computer- mediated one), and describing its effects. The final will be a project or research interest of the your choice, and will involve designing, building and describing a network, a visual and descriptive analysis of an existing network, or a research paper.

GOALS

The course has three broad goals: at the end of the course, you should be able to understand the ways in which both technological and social networks differ from the sum of their parts; be able to make educated guesses about the kinds of emergent properties a given network design might exhibit; and be able to design a network that will exhibit some of these properties.

 

CONCEPTUAL MODEL

The course will progress through 4 phases:

1. Introduction to Networks

We often use the word network without defining the term. FedEx, T-Mobile, and Earthlink all operate networks of one sort or another. What do these different sorts of networks have in common? What characteristics are common to various sort of networks? What desirable effects come from organizing things into a network, as opposed to either more or less rigid forms? What properties do our various communication networks exhibit? How do they differ from one another? What strengths and weaknesses arise from those differences?

2. Architecture and Constraints

Why do networks work? Put another way, what is special about a network, as different from a mere collection of nodes? Can we even talk about networks as a distinct thing, since they are, by definition, interconnected collections of other things?

 

Networks exhibit emergent properties, properties that cannot be predicted from merely examining their component parts, and these properties vary in surprising ways depending on the size and construction of the networks. What are the rules and concepts governing these emergent properties? How can we describe and discuss them?

3. Networks and Culture

Humans both shape and are shaped by networks. We live in them every day, and they become so completely woven into the fabric of our lives that the technology becomes invisible, and our primary experience of them becomes social. "Who said what to whom when" is more important than whether the messages traveled by email or carrier pigeon.

 

Yet the structure of networks does affect the culture that uses them. The kinds of conversations people have via snail mail differ significantly from the conversations they have in email; talking on the phone is very different than "talking" via IM; group conversations that take place in communities like Metafilter are very different from those that take place on irc and different again from mailing lists, in large part because the technology shapes the culture.

 

What is the feedback loop involved here? How do networks affect the social lives of people that use them? How do the users affect the design and deployment of those networks?

4. Open Problems

Over the centuries, networks have consistently exhibited effects their designers didn't anticipate. The internet in particular has been and continues to be a rich source of these unpredicted effects. This quality, coupled with the rapidly growing size and complexity of the world's networks, means these unexpected effects are growing in social and economic importance.

 

What are the open issues in network design today? Given the explosive interest in social networks, from LiveJournal to Friendster to MySpace, what new effects can we predict? As the arms race continues between the RIAA and MPAA vs. designers of tools like bittorrent and allpeers, what new network designs can we expect to see deployed? With an increasingly globalized work environment, what new tools can we expect to see deployed to aid in distributed collaboration? As mobile networks untether labor from location, and make replace planning with coordination in social life, what cultural and economic effects can we expect to see?

 

These are open questions: they don't have easy answers, and are worth thinking about precisely because we don't know what will happen as the world's networks are adapted to new uses.

 

Creating Networks

In addition, although this is a seminar, we will be concerned with creating simple networks, because the actual experience of designing and deploying a network is too slippery to describe without experiencing it.

 

To a first approximation, the networks we are interested in can be defined by 3 aspects: nodes, connections, and contents. The Web and email, for example, use the same nodes (users computers), but have very different ways of connecting (real time versus delayed delivery) and very different sorts of contents (request and reply -- "pull" -- for a specific URL versus sending for later delivery -- "push" -- of text messages), which make using the Web so different from using email.

 

Students will design networks with different properties of the nodes, connections, and contents. These networks can be simulated on a single computer, implemented on the internet, or even executed in the offline world -- projects using networks of friends are as valid as projects as those using networks of computers.

GRADING

Class participation 20%

Papers 30% (15% each)

Mid-term Project 20%

Final Project 30%

SECTION ONE: Intro to Networks

 

Week 1: What is a network

 

We will answer the basic question "What is a network?", building a simple but viable network in class, then examining its constituent parts, and observing its emergent properties.

 

Readings:

 

The Victorian Internet (Handout); Standage, Tom; ISBN: 0425171698

Death and Life of Great American Cities (Handout); Jacobs, Jane; ISBN: 067974195X

Wired / Unwired: The Urban Geography of Digital Networks (handout), Townsend, Anthony

 

Week 2: Elements of communications networks

 

We will decompose networks into a set of constituent elements -- nodes, connections, contents -- then use those elements to describe various types of networks at work today. We will pay particular attention to the operation of the telephone system, and the design challenges it's construction and operation present.

 

Readings:

The Computer As A Communications Device, (Handout); Licklider, J.R.

Inventing the Internet (Handout); Abbate, Janet; ISBN: 0262511150

 

PAPER #1 ASSIGNED

 

Week 3: Rise of the Internet

 

Using our basic elements, and adding three more -- addresses, protocols, and topology -- we will examine the internet. We will contrast it with the telephone network, and how applications are in turn built on top of the internet as a platform.

 

Readings:

 

Smart Mobs (Handout); Rheingold, Howard; ISBN: 0738206083

File Sharing Goes Social; Shirky, Clay http://www.shirky.com/writings/file-sharing_social.html

 

PAPER #1 DUE

SECTION TWO: Architecture and Constraints

 

Week 4: Peer-to-peer and Wirelessness: Flexibility and Mobility

 

We will look at two departures from "classic" internet design: peer-to-peer and wirelessness. Peer-to-peer technologies, best known from file-sharing, but covering a range of architectures and uses, provide tools for weaving a large number of small devices into a fabric of cooperation. What makes peer-to-peer special? How does it differ form classic Web architecture?

 

The label "wireless" is a hopeless grab-bag of technologies, from the short-hop connections of Bluetooth and RFID tags, to communications with geosynchronous satellites 23,000 miles away. What are the various effects of doing without wires on portability, flexibility, mobility. What are some of the long term changes on patterns of living being wrought by those effects?

 

Readings:

That Sneaky Exponential; Reed, David; http://www.contextmag.com/archives/199903/DigitalStrategyReedsLaw.asp

Linked (Handout) Barabasi, Albert-Lazlo; Perseus Publishing 2002; ISBN: 0738206679

Six Degrees (Handout); Watts, Duncan; ISBN: 0434009083

 

Week 5: Mysteries of Scale

 

Our communications are not general-purpose tools; they are fit to particular human purposes (though the fit is not always perfect.) Even without any technological mediation, humans live in complex communications networks; the import of multi-hop sentences like "Jamal told Hermione that Anthony was off with Ernesto's girlfriend Sebla" is obvious even without knowing the people involved.

 

What are human social networks like? What characteristics do they have, and how do they differ from the communications networks we have examined? What are some of the ways the two networks interact?

 

Readings:

The Social Brain: Mind, Language, and Society in Evolutionary Perspective (Handout); Dunbar, Robin

Divided We Stand? Krebs, Valdis; http://www.orgnet.com/divided.html

Vizster; Heer, Jeff; http://jheer.org/vizster/

 

PROPOSAL FOR MIDTERM NETWORK ASSIGNED

 

Week 6: Social networks

 

More is different; networks get complex faster than they get big. A dinner party for 60 is a different kind of thing than a dinner party for 6. As the traffic on the FDR doubles from 4am to 6 am, it hardly slows; another doubling, from 6am to 8am, and it grinds to a standstill. It is more than twice as hard to get a group of four to agree on a movie as a group of two. What effects does growth in size have on the design and operation of networks? What topologies are especially good at handling large scale? Especially bad?

 

Readings:

"Lessons from Lucasfilm's Habitat"; Morningstar, Chip and Randall Farmer; http://www.fudco.com/chip/lessons.html

 

PROPOSAL FOR MIDTERM NETWORK DUE; BUILD AND OBSERVE NETWORK

 

Week 7: In-class Discussion of Midterm Networks

 

Discuss the design and observation of your mid-term network project, with particular emphasis on any surprises, and on what you would do differently next time.

 

Readings:

Open Source Television; Pesce, Mark http://www.disinfo.com/site/displayarticle4565.html

We the Media; Gillmor, Dan; http://www.oreilly.com/catalog/wemedia/book/ch01.pdf

SECTION THREE: Networks and Culture

 

Week 8: Future of Media

 

One of the nexus points of social and technological networks is the media industry; no industry has been more challenged by the spread of novel forms of networking. What challenges have we seen from the spread of flexible networks, and what can we expect to see in the near future?

 

Field Observations: Groups observe Digg, Last.fm, Pandora, YouTube, Second Life.

 

Week 9: Return of the Real 

 

For much of the last two decades, we've been in thrall to the idea of cyberspace, a data space separate from the physical world. Most of the driving force behind cyberspace, though, turns out to have been based not on technological capacities but on technological incapacities. Personal computers forced us to access the internet from fixed locations, while the small online population meant that there was a sharp distinction between the people we knew online and offline.

 

With internet use approaching saturation among an increasingly mobile population in the industrialized world (itself increasingly an anachronism), both those constraints are vanishing, leading to a world where application design is increasingly integrated with what we used to call real life, instead of requiring a separation from it.

 

Each group will briefly discuss their observations of social applications.

 

Readings:

"The Cathedral and the Bazaar"; Raymond, Eric; http://www.firstmonday.dk/issues/issue3_3/raymond/index.html

Here Comes Everybody (Handout); Shirky, Clay (forthcoming)

 

Week 10: Open Source and Simple Collaboration

 

All groups are hampered by coordination costs, the costs incurred in managing communications and transactions in the group, and these costs rise disproportionately quickly as group size grows. One of the principal effects of the internet on group communications is to reduce those costs, thus allowing large groups to create material together. What changes have we seen in the way groups get things done as a result of these lowered costs? What further changes might we see?

 

Readings:

"Social Capital"; Smith, Mark K.; http://www.infed.org/biblio/social_capital.htm

"Autistic Social Software"; boyd, danah; http://www.danah.org/papers/Supernova2004.html

 

Proposal Paper for Final Paper or Project Assigned

 

Week 11: Social Capital and Trust

 

Humans rely on one another in ways far more complex than simple transactions can account for. We built roads and schools, we donate to charities, we help the old and frail across the street without any direct transactional benefit, but the larger community benefits from such actions. What leads people to engage in creating public goods? To engage in reciprocal altruism? What do our communications networks do to help or hamper this kind of behavior?

 

Readings:

TBA, for Guest Speaker

 

Proposal Paper Due

SECTION FOUR: Open Problems

 

Week 12: Guest Speaker, TBA

 

Invited speaker, working on problems relevant to the overlap of communications and social networks.

 

Readings:

The Wealth of Networks (Handout); Benkler, Yochai; ISBN: 0300110561

 

Week 13: Presentations; Politics of Peer Production

 

Theoretical and practical issues in networking abound, from technical issues like namespace and mesh routing to social ones like support for collaborative groups or mobile social application design. We will discuss some of the open issues in networks today.

 

Half the class will present a brief overview of their final.

 

Readings:

The Wisdom of Crowds (Handout); Surowiecki, James; ISBN: 0385503865

 

Week 14: Presentations; Open Issues

 

Continue discussion of open issues in networking; finish discussion of final papers or projects.

 

Half the class will present a brief overview of their final.