Joho the Blogyoutube Archives - Joho the Blog

January 18, 2014

Replacing YouTube ads from outside the filter bubble

Like most people (I assume), when a YouTube I want to watch begins with a 30-second, I switch to another tab until it’s safe to come back.

So, could I please have a plug-in that will instead show me an ad-free video that I wouldn’t have come upon otherwise? Preferably something that will stretch my imagination, cast doubt on my assumptions, and enlarge my sympathy? Or at least a cat doing something awwwwwwwesome.

(Hat tip to MakeMarketingHistory.)


December 29, 2012

Excellent PSA. Bad Algorithm.

This is a terrific public service announcement about the Special Olympics.


Unfortunately, take a look at the upper right at what YouTube thinks is a related video you might enjoy:

Retarded Elephant Running

Comments Off on Excellent PSA. Bad Algorithm.

April 20, 2010

YouTube’s automated copyright filter

YouTube uses a Content Identification tool that enables copyright holders to find instances of copyrighted material so that they can then issue a takedown notice, “moneytize” it, or track it. This includes identifying copyrighted material on a video’s audio track and automatically muting it. There’s no mention in Google’s discussion of Fair Use exemptions.

Mark Smitelli has poked around at the system, uploading copies of the copyrighted song “I Know What Boys Like,” sonically altered in various ways: compressing or expanding the time, lowering or raising the pitch, adding noise, etc. Mark runs the complete results, but to roughly summarize: Altering parameters more than 5% often seems to fool the Identifier, and using less than 30 seconds also seems to let the clip slip through the rule-bound robot’s shiny little nets. Playing clips in reverse confused the Identifier, but stripping out everything except the vocals did not.

Using a clip for as satire [later: I probably got this wrong] or political commentary undoubtedly wouldn’t keep it from the Identifier’s snares, although such use is likely protected and non-infringing. The Identifier, unsurprisingly, seems to be a poor reader of human intention. [Thanks to David Abrams for the tip.]


April 6, 2010

[berkman] Christian Sandvig on the future of TV

Christian Sandvig is giving a Berkman lunchtime talk called “The Television Cannot Be Revolutionized.” [NOTE: I am live-blogging, making mistakes, getting things wrong, leaving things out, not spellpchecking. READ AT YOUR OWN RISK and do not assume this is an accurate reflection of Christian’s talk.]

He begins by crediting Gil Scott Heron for the title. He says he’s looking for a research agenda for studying TV, especially three bottlenecks: distribution, search, and genre.

He talks about a 1995 effort to create a cable channel (The Puppy Channel, then Channemals) that was all cute animals all the time. The creator’s market research showed it would capture a respectable 0.1% of the US TV market. So, he went to Rupert Murdoch, Barry Diller, and Ted Turner, but they thought it was “too weird” an idea for cable. He found it’d cost $17M to distribute it himself.

So, says Christian, the creator launched no one is watching it, but people are watching cute animals on YouTube, etc. The bottleneck has been broken, but it still looks like 1995. E.g., YouTube has a Rentals beta and puts ads everywhere. “YouTube is behaving like a television network and not like a tube for you.” The person behind the recent redesign (Margaret Stewart) of YouTube said “We want you to go into passive mode, sit back, and watch.” We used to think that the Net and Net TV were about interactivity.

TV is important, Christian says. A Ball State U study in 2009 looked at what Americans spend their time doing with media. 100% of people use the phone every day. People use video about 6 hours a day in the US. (This includes video on any device.) About 2/3 of views were on YouTube last year. Analysts say YouTube loses money, but Hulu is profitable. One possible conclusion: Dump amateur content.

In an essay, Christian and a colleague compared Life Magazine and You Tube. Life launched as pioneering photo-journalism. Is YouTube becoming like Life magazine?

Christian shows the power law distribution. What would be the ideal distribution, he asks. We don’t want a single producer of media. We don’t want only a long tail because then none of us share any single media content; that would be complete fragmentation. We don’t want only the head, because that’s media concentration, although it would at least give us enough shared experience to have a culture. In fact, says Christian, we shouldn’t be thinking simply about the shape of the curve. If we were talking about monetary income curves, we’d want to do a mobility analysis: How hard is it to move up the curve? It’s not so much the shape as whether you can move up it, he says.

“We seem to be in the process of building two Internets, Christian says, “which is worrying.” Getting your baby video up on YouTube is easy, and you don’t even need YouTube to do it. But, if you want to show the Olympics, NBC has to do a deal with a Net intermediary; if you tried to do it from the server under your desk, it wouldn’t work. History break: Adorno was exercised by the making industrial of culture via things like expensive, complex broadcast studios; the same is happening with the Net because we need expensive, complex hosting/edge-caching services.

A difference between broadcast and Net media: The head of the curve was purposefully built for TV, but emerged for the Net. TV started out with only local audiences because there were no national networks. After a lot of investment and lobbying, after 1962 we have a national TV system, built to satisfy advertisers.

The Net is supposed to be a cheaper form of distribution than TV. Is this true? If you use Amazon Web Services to distribute a video to a million people, is it cheaper than TV? It’s a hard comparison because TV bundles in the costs of building a market. So compare putting on a late-night one-minute infomercial. The costs are surprisingly similar.

Christian asks why more people aren’t doing research on this and on mobility. In part, he thinks, this is because of the way university departments are structured; they don’t always have people combine expertise in media and Net infrastructure.

Second bottleneck: Search. Videos become popular through being featured on distribution home pages, and on recommender systems. Chris Anderson says recommender systems help you find unpopular results. But there’s no reason for a system to design their algorithms that way, instead of promoting more popular systems. These systems look at things like featured videos (which can be paid placements). Steven Wittens in 2009 found that the algorithm tends to match view counts: If you look at a video, it will recommend other videos with the same or greater number of views.

Finally, the third bottleneck: Genre. How do unpopular things become popular? So much on YouTube apes the conventions of broadcast genres. Parodies of television sitcoms. Parodies of newscasts. Does mobility depend on adopting broadcast genres? Is YouTube just the A&R of the tv industry, externalizing the development of new talent?

Christian says he’s surprised that television seems to be going backwards. Distribution: He’s looking for ways to study this. And, he says, policy issues depend on this as well. E.g., Princeton has a distributed, p2p edge-caching system, sans Akamai. Search: See Frank Pasquale, Christian says. Search: Maybe vlogging, video game commentary, or animal videos are the new genre.

Q: Google says it puts user experience first. Are they betraying that at YouTube?Are they serving users by driving them to popular videos?
A: It’s hard, because people’s wants can be trained. Their search algorithms may reinforce popularity.

Q: You set up a dichotomy. TV is still transforming. There are more channels than content. TV is heading to the Net. Why not see this as a convergence?
A: I admit that in the framing I’ve emphasized a question of Internet exceptionalism. But, this, the Berkman Center, is Ground Zero for Internet exceptionalism.
Q: Maybe the conclusion is that people want mass hits and a long tail.
A: We don’t have good tools for arguing about what we want the shape of the curve to be, but we could about mobility.

Q: What about marketing? That makes a big diff about where you are in the curve. Networks are the biggest advertisers of themselves.
A: Yes, marketing is crucial. We’ve also unintentionally made capital requirements for distribution. I’d like a way to host a video that isn’t hugely expensive.,br>
Q: Why not be in the tail?
A: Chris Anderson says it’s finding your niche, but you could also call it total irrelevance. You want everyone to be able to construct culture.

Q: Isn’t BitTorrent the tool you’re looking for?
A: Any p2p performance tracks to page rank. The ones that will perform best are the ones that are popular.
Q: We want more bandwidth for the popular, and less for not popular.
A: We want a route to move from unpopular to popular. If I put it on my host, I’ll get hit if popular zooms.

Q: Could we be going back to the CompuServ model where we paid but the content was good?
A: Normatively, if we paid for it, there’d be a lot of advantages. Right now we have a sender-mostly-pays system, which is why we have edge-caching.

Q: How much does the culture of the users matter, vs. the culture of the distributors? What happens when users start to game the system?
A: One of the best way to find out about the algorithms is look to people who use these services a lot. E.g., the search engine optimizers. And the new users of new media tend to bring forward assumptions and behaviors from the old culture. Cf. Claude Fischer. Amanda Lotts in The Television Will Be Revolutionized focuses on the old industry. It’d be interesting to see how the old TV folks think about the new one.

Q: Content will flow uphill to money. TV has moved toward the Net. We watch when we want. We’ll see a merge of the two technologies. Understanding why some videos go viral would be valuable, because that will attract money.
A: It’s such a challenging research challenge. There are so many factors involved. But, I do want to say that my talk does recognize the influence of money.

Q: [yochai] You are proposing three distinct kinds of research projects: 1. Look at video creators and musicians. Mimi Ito on anime movie videos: they’re not primarily on YouTube. Look at how hard it is for a producer community to reach the relevant audience. And how important is that, normatively. 2. How much does a user see, read, or hear that is not news and that does not come through capital-intensive media? What is the flow of streams that people watch? We don’t know the answer to that. 3. What is the ability to set the agenda for what is broadly viewed as culture? MediaCloud looks at agenda-setting over time. Start with case studies. E.g., find 50 of the 895,000 viewers of the autistic kids YouTube.
A: This talk is a reaction against looking at how artists reach their desired communities. Lots of research is being done on this. How virtuous communities arise is not the most pressing question. More important: Are there new cultural gatekeepers arising?
A: Imagine that the answer of #2 is the 40% spend 50% of their time on content that doesn’t reach more than 100 people…

Q: Anita Elbersee [sp] says that the Web is magnifying the impact of blockbuster media. This gives the mainstream little incentive to be revolutionized. What’s the impetus to change?
A: Chris Anderson’s book doesn’t hold together because what drives things down the tail you could argue the other way, e.g., the recommender algorithms I talked about. So, what’s the incentive? There’s no “charlie bit my finger” lobby. There are beginning discussions about open video standards. I’d like to see more of a counterpoint to the mainstream approaches. The institutional impetus isn’t there. In most user-produced content, you don’t see revolution, nor would you expect to. Maybe I was being naive.
Q: If you kill the edge-caching business, why wouldn’t everyone get what they want?
A: The fact that you have to rely on third party hosting has effects broadly.


April 3, 2010

The television will not be revolutionized

I’m looking forward to Christian Sandvig’s Tuesday talk at the Berkman Center:


Tuesday, April 6, 12:30 pm

Berkman Center, 23 Everett Street, second floor

RSVP required for those attending in person ([email protected])

This event will be webcast live at 12:30 pm ET and archived on our site shortly after.


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 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.

All part of what one might see as the Berkman Center’s recent (past couple of years) theme that the Internet isn’t working out the way some of us hoped. (Note that the Berkman Center doesn’t really have a theme. I’m just pointing to a trend that may be more of a result of selective perception on my part, reflecting changes in my own thinking.)

Note the info about RSVPing if you want to attend, and the fact that it will be webcast live…


January 19, 2010

[berkman] Tarleton Gillespie: The Politics of Online Media Platforms

Tarleton Gillespie of Cornell is giving a Berkman lunchtime talk on the politics of online media. He’s been interested in how we are shaping cultural discourse through the confluence of tech, policy, economics, etc. Today he wants to look at how social platforms are shaping social discourse.

NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people.

He begins with YouTube’s announcement in Dec. 2008 that they’re going to become more conservative in blocking offensive videos: removing some, moving some behind an age firewall, and algorithmically demoting some so they won’t appear on the most popular lists. This combines traditional tactics with newfangled technical management of where things appear. We don’t really have a language for how these sorts of innovations work.

He asks: How do we take the tradition of asking questions about how commercial providers shape the public discourse … with the basis that these providers, especially the most prominent ones, are playing a role in determining what ends up online, viewed and possible? How do we apply this to new media? Three differences in how online media work: 1. Emphasis on user-generated content. 2. Gatekeeping or comprehensiveness? E.g., Google wants comprehensiveness for Google Books. That changes why they would include or exclude. 3. They cater “to active niche communities, trying to produce a coherent site, consistent brand, and commodifiable audience.”

“How do these sites promise to be everything and not everything at the same time?” How have they cultivated the notion that they provide everything in a neutral manner? How do they intervene in what they provide? “What obligations are we willing to impose, to protect free speech and ensure a healthy public discourse?”

What about the promises they make that makes them appear neutral? How do they articulate their services and sell themselves to the various stakeholders, setting the terms for how they’re judged? Part of the answer: They use the term “platform.” “The role this term plays is indicative of the type of positioning a youtube, facebook or flickr would like to establish.” These terms are carefully chosen and are carefully massaged. Why has this term fit so comfortable in these sites’ characterizations and why have we accepted it? E.g., before being bought by Google, Youtube referred to itself as a service and a community. Afterward, it became a “platform.” The term draws on the computational meme: an infrastructure on which tools can be built. Marc Andreesen disagrees because you can’t build tools for it. [This is the original geeky meaning, but its meaning has shifted, IMO – dw] It also has a political meaning. And architectural. All these meanings help the term resonate. There are a series of connotations that are powerful in this tool: An open space, egalitarian, wide, limitless, facilitating something of value.

The term “platform” manages the conflicts among stakeholders for youtube. For users, it’s a platform from which to be heard. For advertisers, it’s a platform of opportunity. For media partners, it’s a distribution platform. For lawmakers, it’s a fragile, valuable platform that enables free speech. When they are talking about liability, they are merely a platform. Structurally, “platform” is not unlike “conduit.” [Hmm. I think that for advertisers, YT is a platform in that it’s an open space where millions of users come together. – dw]

So, how do you begin to find a language for the technique and justifications online media make about what belongs on their site and what doesn’t. Facebook, youtube, and flickr adopt different strategies. Youtube maintains that it doesn’t look at content proactively but only when their users flag it. But they do look for spam and are obliged to look for child porn [actually, I think they are not required to proactively search out child porn — dw]. Youtube has a figure 8 model of community governance: Users flag content. Users can comment on the guidelines. Users can game the system, but Youtube can decide which flags to ignore. Users can complain about being flagged. So, while Youtube positions itself as non-interventionist, it actually isn’t. It says it’s defending the community according to the community’s norms, but those norms have been crafted by YT in accordance with its legal and economic interests.

YT’s algorithmic demotion of videos manages their front page. They don’t want it to look like a soft core porn site; those videos are there, but it’s not their image. Flickr does this carefully as well. Their front page tells you that this is a site for landscape photos, and birds, and arty shots. Amazon’s best seller list excludes “adult” literature. Not to mention (he says) Amazon’s removing from the Kindle a copy of 1984 that was posted in violation of copyright; what seems like ours isn’t really.

[I’m doing a terrible job capturing the questions. Basically, I just couldn’t hear the first couple. Sorry!]

Q: [wendy] Platforms vs. intermediaries. “The lawyers tend to talk about intermediary liability or immunity, whereas economists talk more about platforms.”
A: Intermediaries such as ISPs have a different set of protections. YouTube wants the protections but doesn’t fit neatly into that definition. Viacom calls YTY a “distributor.”

Q: Couldn’t these platforms get out of the dilemma by providing curated and uncurated versions?
A: Flickr comes closer to that. It tries to have it all but not be visible about having it all. They have a “Porn is in the back” approach.

[me] We’re in a confusing time. We’ve invented new things that don’t fit the old vocabulary perfectly. What should we do about the lack of a vocab. Invent a new one? Be vigilant about understand how people use terms?
A: All vocabularies are strategic. We should unpack the terms and recognize that they’re doing work, and that the connotations matter. E.g., issues of liability depends on whether we see them as intermediaries or distributors. Is it about imposing a new vocab? Or maintaining vigilance? I’m torn about the impulses in those directions.

Q: We had a system that had user ratings. We a “text jockey” looking at msgs 24/7. We call them global ratings vs. contextual ratings. It’s gotten very very complex. E.g., cleavage photos have to have a head included.

Q: [jodi] How has the near real time feedback influenced accountability/exposure of algorithms and decisions? The Twitter/#amazonfail incident, for instance. Amazon was faced with a decision to respond or not, and then further faced with a decision of what to do about the allegation.
A: The Amazon FAIL revealed what was going on all along. Now the reaction can be faster, is more public.

[Missed some more questions because my hearing is getting worse.]

Comments Off on [berkman] Tarleton Gillespie: The Politics of Online Media Platforms

May 30, 2009

The YouTube election … in Iran

Hamid Tehrani at GlobalVoices posts about how Iranian candidates for the presidency are using YouTube…including controversies about jokes and ad hoc footage…

[Tags: ]

Comments Off on The YouTube election … in Iran

April 9, 2009

Happy Birthday to You isn’t copyrighted???

In a comment to a distressing post about YouTube automatically taking down any video that contains any copyrighted material, even if it’s covered by Fair Use, a commenter posts a seemingly learned post explaining why “Happy Birthday” may indeed not be under copyright.

And to think of all the years I spent singing “For She’s a Jolly Good Birthday” instead!

[Tags: ]

1 Comment »

December 22, 2008

Four hands one guitar

Two of the hands are especially good. This is a fun video.

[Tags: ]

1 Comment »

October 31, 2008

YouTube 1985

This video pretends to be from YouTube’s origins in 1985. Cute.

[Tags: ]

Comments Off on YouTube 1985

Next Page »