April 6, 2010
Notes from "The Television Will Not Be Revolutionized" by Christian Sandvig

Only a few hours ago, I had the opportunity to attend a Berkman Center for Internet & Society luncheon with a lecture by Christian Sandvig on the future of "television" distribution online.

The description from the event reads:

Video on the Internet briefly promised us a cultural future of decentralized production and daring changes in form--even beyond dancing kittens and laughing babies. Yet recent developments on sites like YouTube, Hulu, and Fancast as well as research about how audiences watch online video both suggest a retrenchment of structures from the old "mass media" system rather than anything daring. In this talk I'll argue that choices about the distribution infrastructure for video will determine whether all our future screens will be the same.

Christian argued that online video is more and more resembling old models of television networks, and he talked about everything from the YouTube redesign to a new approach to Chris Anderson's "long tail" model of distribution. He delivered some engaging thoughts on bandwidth monetization and asked critical research questions into how television and online video researchers can go about tackling issues of network algorithms. My liveblogged notes provide some textual takeaways from his talk, but the full lecture will eventually be available on the Berkman website here.

Christian Sandvig is a Fellow of the Berkman Center and Associate Professor in Communication, Media, and at Coordinated Science Laboratory at the University of Illinois at Urbana-Champaign. He holds the Ph.D. in communication from Stanford University. In 2006 he received the Faculty Early Career Development Award from the National Science Foundation in the area of Human-Centered Computing. He blogs at multicast.

My notes follow (unedited) after the jump!

The Television Cannot Be Revolutionized: Christian Sandvig


Amanda Lotz: The Television Will Be Revolutionized

distribution: defines mass medium of television
tv would change if distribution bottleneck went away

other bottle necks: search, genre

research agenda: for studying this empirically

1995: Puppy Love: about her father who tried to launch cable television channel: The Puppy Channel, then Channemals

after research: viable to create the channel: getting enough audience members

result after pitch: "too weird"


2010: distribution bottleneck: broken

The Kitten Channel, I can Has Cheezburger, YouTube

distribution over the past decade: looks more like 1995 than 2010



- Rentals Beta
- streaming sports online
behaving like television network and not 'tube for you'

The Internet: more wingman than TV killer: Brian Stelter (NYT)

YouTube redesign: motivation: emulate TV more closely: want audience to go into 'passive mode'
Margaret Stewart, Head of User Experience @ YouTube


interactivity in television: Raymond Williams: "flow"
TV works by preempting the user to not change the channel

Sandvig's worry: same thing is happening to YouTube

multiscreening is increasingly the norm

CRE/Nielsen 2009, Ball State Study
audience tracking: tried to see what audiences did not w/ static categories (eg., TV v. Internet); instead: followed daily lives by watching minutes per person

video, per person: 6 hours per day

people aren't watching less video, but watching more YouTube (maybe)

YouTube: 2/3 video views of all video on Internet (estimated)
but Hulu: reported profitable (2010)

Life Magazine compared to YouTube:
Life failed as weekly magazine in 1972, was pioneering as photojournalism source
audience: not demographically appealing to advertisers
non-traditional content on YouTube might be popular, but it's not what advertisers are not happy w/ it

Graph: Audience size (linking, hits, watching reading) v. number of sources (web sites, TV channels, books)
past 10 years: corresponds to power log distribution (Webster 2006, Benkler 2006, Adamic & Huberman 2003, Barabasi & Albert 1999, and many others)
graph: small number of sources w/ large audience, large number of sources w/ small audience (shape varies by industry & what distribution you're using)

Chris Anderson: popularized this idea with concept of "long tail"

NOT a good way to think about distribution
normative line = ?

diversity --> seen as good: the longer the graph goes to the right the better
media studies: see this as fragmentation
if you look at the left: concentration: so few things, everyone's watching them: terrible or important: shared experience, can have conversations & form community

Stratification: Distribution & Mobility
not the shape of the curve but if you can move up it
(this is research that largely isn't done)

maybe: in the process of building 2 internets
distribution bottleneck is reappearing

home videos versus major television content (eg., Olympics)
NBC has to partner w/ distribution network (eg., Limelight Networks)
eg., issue of bandwidth & source congestion

factories of culture emerging (Adorno's time)
need cultural critique

ecology of companies guaranteeing video for large audiences
by and large: can't use them: won't accept video unless you're a large producer
need intermediary like YouTube that will use one of these services or be one

television stations were the edge caches of the 70s 80s and 90s
we are creating a new television network

stuff to the left: 1962 (Sterne 1999: Television Under Construction)
television: defining characteristic: huge mass audience
but: wasn't national audience for television: some people thought it was the way to go, others thought local was the way to organize the media system

after 1962: national television system: largely to satisfy advertisers

Internet: different from TV: supposed to be cheaper: not clear that it actually does
experiment w/ Amazon web services: has CDN that individuals can actually subscribe to (Akamai: won't, need minimum amount)
generally speaking: not hugely different in price
when you pay for TV: you're including the cost of the audience (bundled)
when you pay for content distribution network like AWS: not generally cheaper

point in raising distribution bottleneck: surprised there's not research on mobility, or awareness of content distribution networks (no expertise in departments where there's media but also understanding guts of industry)


other emerging bottlenecks: search & genre

how do you drive traffic to your video?
YouTube: featured placement on homepages, recommender systems
Chris Anderson: recommender systems: can democratize distribution of content, drive traffic to smaller videos
but: will these algorithms do this? can easily design it to push toward videos that will make more money
but: we don't know the algorithm Google's using

academic computer science research on algorithms for recommender systems: Video Suggestion and Discovery for YouTube: Taking Random Walks Through the View Graph

algorithm: tends to match view counts: never recommends videos w/ lower views (YouTube enthusiast: Steven Wittens) --> can optimize the algorithm to hit videos from favored media companies
we don't know if the algorithm is democratizing in itself

so much that's happening on YouTube: apes what is happening in mass media
YouTube: parodies television sitcoms: 18 minute w/ ad -type show
why would you make something that looks like a newcast or television sitcom? besides it would be recognizable to audience that knows the conventions

it matters whether videos that can move up are different or same
mobility for things that look like television formats that already exist: externalizing development of new content to amateurs (Andrejevic)

Adorno: the products of mass culture are interchangeable: hard to argue why one should succeed over the other
but if you're writing about news & public affairs: can take example of something that may not have gotten out on its own (politically relevant programming), but not sure if it's the same for something like YouTube

is there something you can identify on YouTube that mirrors photojournalism of Time Magazine


surprised that television is evolving backwards: resembling past rather than adapting to future

now: "research" at the level of anecdote & suspicion

there are ways around edge caching
don't NEED a CDS that would be required to distribute video, can't distribute other ways
eg., P2P edge caching: can donate bandwidth to help other people

in search: debate over (law prof: Frank Pasquale) need to expose algorithms

on genre: maybe what we need vlogging or video game commentary IS the new genre; in a few years: YouTube will be the pioneer of "weird" things that couldn't be on TV
maybe in a year: it will be all laughing babies and Modern Warfare 2 commentary


Chrisian Sandvig




Q: Is Google making the algorithm for users or to drive ads?

A: people can be taught what to want; so, hard to distinguish between the two
featured videos on YouTube's front page

Q: TV's convergence?

A: media concentration
emphasized: Internet exceptionalism

Q: marketing

A: capital is required to gain popularity
way that we find ways to watch (others avenues: word of mouth, eg.)
unintentionally made requirements for distribution?
would love avenue for video that no advertiser would support that wouldn't be hugely expensive to support

Q: BitTorrent, for distributing videos

A: any P2P algorithm: problem: performance characteristics that mirror PageRank: video w/ best performance: the most popular one w/ most seeders, etc.

Q: two internets: 2.5 million views by gaming algorithms: creating content to get views through the systems (anti-Wikipedia?)

A: HBO model:

Q: (my question) user culture: how does that affect how videos connect

research method: look at people who use YouTube alot: they're trying to figure out the algorithm: communities that probably aren't large

Amanda Lotz: focuses excessively on the old culture of the TV industry; what does modern industry culture look like, and how do they approach these issues?

Q: focus w/ money: merging of two technologies?

A: talking about distribution money versus advertising money versus subscriber pay money; socialist YouTube?

Q: Yochai: primacy of distribution: AMV research by Mimi Ito: users don't care about where they're hosting
communities: that create cultural contexts for themselves

what is the flow of streams that people catch, how many require capitalization and how many don't?

question of ability to set the agenda of what is broadly conceived of as culture? eg., what Meme Tracker is going w/ text
creating some objective element of culture & capture it to see how it evolves and works

A: talk is a reaction against how communities reach desired audiences
YouTube Reader: but it tends to act as if scholarship needs examples of virtuous small communities
but interesting question: is there a new gatekeeper or cultural industry configuration that's arising?
we agree that these communities already exist: eg., fans & remix culture

Q: Web: magnifying the value of these blockbuster media
people in mainstream media: little impetus to be revolutionized, because things drift toward blockbuster size

A: Chris Anderson's ideas: not useful: arguments don't hold: because you can easily argue opposite way
incentive: writers about television's transformation: have forgotten institutions: there are huge places that have investment where there's not lobbyists for Charlie Bit My Finger
open video standards: still at nascent state
want to see more of a counterpoint

edge cache: if it disappeared: fine