InfoTech & Public Policy

It isn’t exactly the internet-television that most people imagine, but YouTube and Google Video do enable videos to be easily propagated. Anyone can upload a video and these sites will provide the bandwidth to serve up the file. These are extremely popular because they remove the sophistication needed to publish a video on the internet.

Videos can be can uploaded in a variety of formats. For example, Google Video currently accepts

AVI, ASF, QuickTime, Windows Media and MPEG formats … Specific video codecs we accept include H.264, H.263, MPEG 1/2/4 and motion JPEG. [cite]

YouTube is a bit less specific, but does accept the same file types. Once uploaded, the video is converted into a Macromedia Flash video. This conversion, I believe, was a key component of the popularity of these sites. Most web browsers have the Flash plug-in and providing the video in this format decreases the hurdles to watching a video. They could have streamed the video via a Windows Media player or Quicktime Media player, but the Flash version allows for a wider audience base. The quality of the video is not very good, but it seems like a nice compromise of bandwidth and quality since the costs are all paid for by the companies (it does not cost anything to the user to upload a video). For those who have never used YouTube or Google Video, here is a video I found of the Princeton University marching band on Google Video.

Now, as I’m sure you can imagine, this type of service attracts a large number copyright violations. Google Video has all of the DMCA (Digital Millennium Copyright Act) rules on their FAQ site and requires uploaded video to be “verified” before it posted. YouTube also has a copyright and DMCA section in their help, but they do not require a verification process.

Google says the verification is for technical and policy reasons, but I imagine it is more for the first. There are a large number of uploaded videos each day so it seems infeasible that a human would go through and watch each video for copyright violations. There are a number of pirated videos on these sites and that also suggests that each video is not watched before posted. These sites have been quick to obey DMCA takedown notification for copyright violations and remove the offending content.

YouTube recently put a limitation on the size of the videos that can be uploaded. They began a 10-minute limitation (unless you have a Director Account) to curb the copyright violations. The thinking was that real user-created video is usually small, but professionally created (where the owner would not want the video distributed for free) was large. I think they were specifically targeting television episodes (which are about 20 minutes for a half-hour show without commercials). YouTube hasn’t posted evidence either way, but I imagine this restraint did not do much to stop the actual violations, because now the shows are split into multiple uploads. It may have been more of a public image motivation. As long as these companies continue to remove the offending content upon notification from the copyright owner, they should be safe from lawsuits.

I want to talk about the new Sprint Nextel cell phone tracking and alert system. Just recently Sprint began offering their new Family Locator Service for a certain set of Motorola/Samsung/Sanyo phones. If you have one of these phone and a Sprint service (or you can just sign up for a new contract), for an additional $9.99 parents can get detailed location information about where their children are.

The service allows one “parent” phone to make requests for four “children” phones. To setup tracking on the children phones, a request is sent to the phone notifying the user that this phone will be setup with this tracking service. The user has to confirm the request (with a confirmation number). I’m guessing that most parents who subscribe for this service would do this setup by themselves before handing over the cell phone to the child.

After the initial setup, the parent account can make a request to the service and view on a map the location of all four cell phones. This can either be viewed via a regular computer the Sprint website, or for certain phones, can be viewed directly on the parent cell phone.

Sprint made an interesting design choice to reduce some of the intra-family privacy issues when a request is made. If the Family Locator service can successful locate a child’s cell phone, that phone is sent a message notifying it that the parent phone has requested its location. This seems like it will at least prevent parents from secretly monitoring their children (I don’t have the service, but it does not look like it can be disabled). I imagine this kind of technology can cause a whole host of trust problems between the parents and children, but at least the tracking will be done openly.

Sprint also offers a Saftey Check feature. From the Sprint website, the parent account can setup a periodic tracking request to the children phones. If the phone is not within a specific area when the request is made, an alert is sent to the parent’s phone. This Safety Check is setup for situations like arriving at school safely or being home before curfew. I find it funny that Sprint FAQ recommend keeping the child’s phone on vibrate during the Safety Check times so the constant tracking notification does not interrupt classroom activities.

The Safety Check works with positive logic – ensuring that a child is within a certain area at a certain time. I think it would be easy to implement a feature that would check and only alert a parent if the phone is within one of the forbidden zones. I bet some parents would pay for the ability to setup a registered-sex offender or bad-influence-friend alert.

Sprint does specifically remind parents that the tracking only works when the phone is on. I think Sprint should also remind parents that this is only tracking the cell phone, not the child. It is very easy for the child to leave the tracking cell phone in school when he/she decides to cut class. The underlying technology is GPS when available and then cell phone tower locations. It is only as accurate as the technology it is using (from a few yards to a few hundred yards). This type of variation may make the houses nearest to the school the hottest party spots.

During the in-class discussion of the future of television, Professor Felten briefly mentioned the lawsuit brought against ClearPlay by the Directors Guild of America. ClearPlay enables certain DVD players to be customized with specific rating criteria (like nudity or violence levels) by the users so the playback of a DVD will skip or mute certain segments to conform to the rules setup by the users. The Directors Guild charged ClearPlay with copyright infringement for creating an unauthorized derivative work. I was looking for more details about the case and found it to be pretty interesting.

As I stated before, ClearPlay allows for users to setup their DVD player to filter certain content during the playback of the DVD. Users must purchase a DVD player with filtering capabilities (such as this one from amazon.com) and subscribe to ClearPlay to receive new filters. The information about how to filter each particular DVD can be obtained from ClearPlay (via various subscription models). They do a pretty good job of offering filtering for new releases and have a list of over 1900 movies. Once the specific filtering information is obtained by the DVD player, users can configure a filtering menu to specify the type of content that can and cannot be seen. There are four main categories: violence, sex and nudity, language, and other (which, for example, include certain types of drug use). Each category can also be further configured:

Under language, for instance, viewers can filter for six levels, including “vain reference to the deity” or “strong profanity.” [1]

I have never watched a ClearPlay filtered DVD, but here is a nice excerpt from a PC Magazine review :

Spiderman has, I’d say, one truly worrisome or suggestive scene for kids under 13. Mary Jane Watson is walking home alone in the rain when she’s accosted by a bunch of hooligans who, as the scene progresses, appear to want to rob or rape her. Spiderman arrives in the nick of time to save the day in pretty violent fashion. There’s punching, kicking, and more, but no blood. With the bad guys vanquished, Spiderman steals away into an alley. Mary follows him. She’s dripping wet. So wet in fact, that her top is virtually see-through. She and Spiderman engage in a steamy, prolonged kiss, and then he takes off.

Here’s how the same scene played out with the filter on. The guys still chase Mary, but the intensity of the scene is muted because it’s been cut down a bit. Spiderman arrives and beats the guys up, but this scene is also somewhat shorter, with some of the most bone-crunching blows removed—oddly, both versions did include Mary delivering a kick to the crotch to one of the attackers. The scene in the alley receives some serious tightening, and any glimpse of Mary’s see-through blouse has been excised.

The nice thing about all this is that you could never tell when the cuts were happening. There was no delay, no hiccup in playback. I would imagine that trying to maintain some semblance of continuity in a profanity-laced scene could prove more challenging.

This technology seems like a natural replacement for manually pausing or fast-forwarding through certain scenes that parents do not wish their children to view. The Directors Guild of America, however, does not believe that and in 2002 brought a lawsuit against ClearPlay. They claimed that the ClearPlay technology was creating an unauthorized derivative work of the content and this was in violation of copyright law.

“ClearPlay software edits movies to conform to ClearPlay’s vision of a movie instead of letting audiences see, and judge for themselves, what writers wrote, what actors said and what directors envisioned,” The Directors Guild of America said in a statement.

“Ultimately, it is a violation of law and just wrong to profit from selling software that changes the intent of movies you didn’t create and don’t own,” the statement said. [1]

The lawsuit was filed against ClearPlay and other companies that offer similar filtering and is still pending in U.S. District Court for Colorado. Here are the EFF links to the case documentation.

Since the case has been filed, new legislation has been enacted to specifically address some of these issues. Now this is where things get interesting. In 2005, Congress passed S. 167: The Family Entertainment and Copyright Act of 2005. The relevant section of this bill specifically enables technology that allows :

the making imperceptible, by or at the direction of a member of a private household, of limited portions of audio or video content of a motion picture, during a performance in or transmitted to that household for private home viewing, from an authorized copy of the motion picture

This bill specifically allows filtering technology such as ClearPlay. It is currently unclear if the lawsuit will continue given this new legislation. S.167 answers the question about this specific type of filtering technology, but did not give a conclusion of the scope of the derivative work restrictions on copyright.

One of the reasons, I believe, the lawsuit may still continue is CleanFlicks. I mentioned before that the lawsuit was brought against ClearPlay and other filtering companies; CleanFlicks is one of the others. The CleanFlicks model is to make a one-time filtering of the DVD content from an authorized (rental) copy and create a new DVD. This filtered DVD is then rented or sold to subscribers. The filtered content is not removed during playback, but rather permanently removed from the media. I don’t believe that this scenario is specifically allowed from S.167, so we are now back to the question of derivative work. I think it will be very interesting if this case continues through the courts.

            Many networks have begun offering their television shows online for a small fee.  For $1.99 most of today’s most popular shows can be downloaded through a service like iTunes.  This approach has been the most common for all of the major networks.  Recently however ABC has launched a new service which allows users to watch four of their current shows for free with limited commercial interruption.  These shows include two of the most popular shows on television: Lost and Desperate Housewives.  It also includes Commander and Chief, a show which has struggled to take off.  From this selection of shows, it would seem clear that ABC is attempting to test the waters for a service like this and see how popular the service is and what it does for ratings of different shows.

            I have had the opportunity to play around with this new service a bit.  Its construction is very clever for a few reasons.  First, they do not post a new episode of a show until the next morning after the show has aired.  This means the fastest way to gain access to the show is still to watch it live when it is aired.  However, for many viewers this is not possible or not convenient for one reason or another.  This service allows them to view the episode later on without having to remember to set their VCR or Tivo to record.  In effect, this service allows ABC to take some control over time shifting. 

            For ABC it would seem there would be a number of advantages to offering this service.  First, many of their popular shows build on an ongoing plotline from week to week.  New viewers will often feel lost if they begin watching the series in the middle of the season.  This service allows those who discover the show midway through the season to catch up.  Other viewers who might have to just miss a single episode would also be able to stay caught up in the series without having to remember to record the missed episode.  All of these different considerations can only lead to increased overall ratings.  Even if some users switch from watching the show live to watching the show online the overall number of viewers should still increase.  If one is a fan of a particular show, one will not watch it less because it is available more.

            The one potential way for the number of viewers to decline is if less people are exposed to new shows through commercials.  However, this new system can expose people to even more shows.  For example, when going to ABC’s homepage to find an episode of one show advertisements for all of the other shows can be placed throughout the website or even as commercials during the show. 

            ABC’s method for including commercials during internet viewing is very clever.  There are three or four sections in each show that are briefly interrupted by commercials.  Each commercial spot lasts for only 30 seconds.  By making the commercial interruptions so brief, ABC accomplishes two things.  First there is a better chance people will actually pay attention to the commercials because their brevity provides less incentive for a person to find some other activity to occupy themselves during a commercial. The commercials are also short enough that people will not view at as too much of a nuisance compared to recording a show themselves and then fast forwarding or downloading a commercial free copy illegally. 

            Time shifting has been around for a long time and now with products such as Sling Box place shifting is likely here to stay as well.  With this in mind ABC has found a way to embrace these two technologies and take control of them.  The one major hindrance to the service thus far when I have used it is connection problems.  There are times when the network slows down to the point where an episode can no longer play.  It is unknown where the bottlenecks that hinder playback are occurring.  However, these bottlenecks could potentially add to the debate surrounding network neutrality as networks might prefer to pay extra to give their shows priority over the network to ensure quality. 

In class last week, we talked about the use of geo-location technologies with respect to directing emergency calls of someone using VoIP. A little looking into how these technologies are available and what they’re used for, and it seems that there are applications to many of the topics we’ve talked about this semester. When so much of the problem of internet governance stems from the anonymity of internet users and the ease with which internet traffic can transcend borders, knowing the location of an internet user changes the game.

First, let’s take a look at the geo-location technologies currently in use. A large part of the software used now is available from proprietary providers such as Quova and Digital Element. This proprietary software is based on determining the location of a particular IP address. This information has to be gathered by analyzing the locations of IP addresses all over the world. Note that while there are ways to conceal location (such as using a proxy server), the software can flag internet users who are doing so. Interestingly, another mechanism for determining physical geo-location without using IP addresses was patented in September, 2005. The patent holder: the NSA. The NSA website briefly profiles its “Network Geo-Location Technology” which as described here measures latency to build up a network latency topology map. This map can then be used to look up a computer based on the time it takes to connect to that computer. (This system can also be avoided using some kind of proxy service). The patent on this method of geo-location is one which the NSA has made available to be licensed for industry use.

Consider some of the possible uses for geo-location technologies (some taken from this article and the websites of Quova and Digital Element): Credit card companies or banks can use this software to detect fraud. Gaming operators can adopt this technology to comply with laws regarding trade practices over international and state borders. Sites such as eBay can use this to restrict which products are available in which locations (think Nazi paraphernalia and France). Hosts of digital worlds could keep track of the international transference of real money through in-game assets, and online providers of digital content can use it to comply with contracts that require them not to broadcast shows in certain areas (according to this article, major league baseball is a Quova customer that uses the product to make sure locally-broadcast games don’t lose their exclusivity by being unconditionally available online).

Aside from all of these applications for law enforcement, the companies that offer geo-location software market it for its ability to improve the user experience of the web. Namely, they cite the advantages of being able to give a user local search results, for example to guide a user to a store closest to them, and to target online advertising (ad-serving is currently the most common use of geo-location technology). Digital Element’s website claims that 25% of all Internet searches are local in nature, and that targeted ads have been shown to sell up to 30-40% more than general ads. Even if these numbers are exaggerated, it’s hard to argue that such location technologies have value to advertisers, search engines, and others.

The problem with geo-location technologies is that there’s a fine line between using general geographic data add desirable features or to make sure border-dependent laws are enforced on the web, and infringing on privacy and collecting personal information. Depending on the edition of Quova’s software, information obtained can be anywhere from simply geographic information in the basic edition up to much more in the security edition (“geographic information and confidence factors, demographic information, connection type, connection speed, IP routing type, AOL flag, ASN, carrier name, top-level domain, second-level domain, registering organization, a list of anonymizing proxies, hostnames, and routers”), and the services offered by Quova includes a variety of audit services for data collection (marketed as a means to help online business owners manage their business). We have to ask how much information should be available, and to whom.

While the current capabilities of both systems do raise questions about privacy and information availability, the question really comes up when we consider potential improvements on these technologies. While for marketing uses (among others) being able to locate some user / IP address to a general geographic area is sufficient, there are certainly other uses that would benefit from more exact geo-location technology such as emergency service for VoIP users, or the ability to more exactly locate someone committing fraud. The downside is that in different hands (or even in the hands of law enforcement officials depending on your point of view), the ability to accurately locate anyone using the internet is a dangerous tool, certainly one that you would not want to be commercially available. Using proxy servers may be an effective method of avoiding being located, it would also means the loss of all the positives of geo-locating technology.

The applications and benefits of geo-location technologies are quite substantial, and potentially provide a means to deal with many of the policy-related issues of internet regulation we’ve discussed this semester. As the technologies develop however, care needs to be taken with respect to the capabilities of such software available through general distribution.

I want to talk a bit about the television advertisements. It seems to be a very popular in posts over the past few weeks – debating if the 30-second advertisement is dead or if in-show placements are the future. I want to focus on actually measuring viewer and the predominant company behind making those measurements: Nielsen Media Research.

Nielsen was founded in 1926 by Author Nielsen, first doing product testing and later moving into market research as a way to determine how products were selling. In 1936, it purchased the technology for an “Audimeter:”

the machine was capable of making a minute-by-minute record of when a radio was on and where the dial was set.

Does that device sound familiar? In 1942, Nielsen launched the Nielsen Radio Index based on data collected from Audiometers placed nationwide in 800 homes. Flash forward three decades, and in 1973, Nielsen Media Research launched

a new metering technology called “Storage Instantaneous Audimeter” for nationwide service. The new Audimeter automatically recorded and stored minute-by-minute tuning records for channel, time of day, and duration of tuning.

Currently, Nielsen uses a combination of user diaries (where the members of a household record what and when they watch television) and these home set top boxes. Recently, there have been a few new methods to monitor the way in which people are watching television, and I’ll explain those in a bit. I was amazed that until a few years ago, the technology and methodology used to measure television viewers, which in turn is used to determine advertising costs (a $70 billion industry), was essentially developed in 1926.

It was just last year (in 2005) that Nielsen began taking DVRs into account when tracking television viewers. They began to break down the viewers into “Live” and “Live Plus Same Day” and “Live Plus Seven Days.” As Andrew pointed out in his post, 70% of viewers are still viewing live TV, but what does that really mean in terms of advertisement viewers?

Also in 2005, Nielsen began releasing the minute-by-minute viewing habits of its samples which allows for analysis of specific commercials in specific programs. The minute-by-minute analysis has been collected since the first Audimeter in 1926, yet it was only released (or able to be purchased) beginning last year.

Nielsen also has been reluctant to offer “commercial ratings.” These ratings would show the effectiveness of commercials and have a huge impact on advertisement pricings (they would take into affect the DVR aspects and aggregated minute-by-minute viewings). The current thought is that these commercial ratings will be released by for the beginning of next season’s television.

Nielsen is also looking at other ways of monitoring viewers, including programming on cellphones, iPods, or via internet broadcasting. They are also looking at a measurement of “engagement.” This would account for how much attention a viewer paid to the show/commercial rather than solely if the television was on. Although an advertisement may be on television, if nobody is actually watching it or retaining information about it, it is not as useful.

It seems that Nielson, so essentially the entire industry, is way behind in being able to measure these new methods of television viewing. Although the DVR technology is only a decade old, minute-by-minute analysis and the corresponding “commercial ratings” could have been released much earlier than last year. Nielsen essentially has a monopoly on the television ratings (when was the last time you heard something other than the Nielsen ratings related to TV?) and that seems to be slowing innovation. In late 2005, erinMedia brought an anti-trust case again Nielsen citing anticompetitive practices to impede innovation, although the case has not gotten very far. Nielsen is under the threat of government scrutiny (see Senate FAIR Ratings Act), although nothing has come from that either. It seems that the introduction of new ways to view television may be threatening the Nielsen monopoly as advertisers are becoming more unhappy (and vocal) about the lack of ratings information, but so far, there have been no significant challengers to the Nielsen monopoly.

A recent article in the Washington Post describes how Google and other ad networks use typosquatting to generate revenue. In brief, somebody buys up a domain name that’s a likely misspelling of another website’s URL, relying on typograhical errors for site visits. It’s a win-win situation for Google’s booming ad network, which provides the ads for these sites, and the cybersquatters themselves, who leverage the low fixed cost of individual domain names to make a handsome profit.

As with any flow of easy money, this makes some people very angry. Trademark lawyers insist the typosquatters are eating into advertising profits that should belong to the victim site. Many consumer advocacy minded figures, such as Harvard researcher Ben Edelman, indicate the practice of profiting from user error is just plain evil. I want to examine whether current policy applies to this case, and whether there’s a need for new cybersquatting laws.

Legal action is established in three major cases:
(1) The content is of an adult nature, in a domain name that gives no indication of such content. The site’s owner is in violation of the The Protect Act and liable to fines or 2-4 years if imprisonment. The Google AdSense Program Policies explicitly prohibit this as well, so we can rule out this case - most of the sites are blank except for ads and links.

(2) The content is harmful to users, containing viruses or spyware that downloads itself onto the host computer. Both Yahoo and Google’s ad programs forbid this. Nevertheless Yahoo was recently accused of trying to use its relationship with ad-ware sites to generate extra revenue around earnings time. Let’s assume Google sticks to its own policies, but leaves harmless ad-filled pages alone.

(3) The domain name and content violate trademark law. Existing law has developed through two major allegations, trademark infringement and dilution. The Berkman Center for Internet and Society provides a summary:

Elements required for trademark infringement:
1. Prior rights in the trademark - through use or registration.
2. Commercial use
3. Likelihood of confusion

Elements required for trademark dilution (added by Congress in 1995 and reinforced in 1999)
1. Marks must be famous
2. Commercial use
3. Protects against blurring or tarnishment
* Blurring: blurring occurs when the defendant’s use of the plaintiff’s mark causes the public to no longer think only of the plaintiff’s product upon seeing the famous mark, but rather to associate both the plaintiff and defendant with the mark.
* Tarnishment: tarnishment occurs when an association of a famous mark with inferior quality or unsavory products tarnishes the mark.

I agree with Google’s trademark lawyers that “obvious” misspellings, such as “BistBuy” instead of BestBuy and “blgospot” instead of BlogSpot, are confusing to nobody, ruling out trademark infringement claims. Given the uniform apperance of these ad pages, it’s usually obvious you have come to the wrong place, so there’s no question of trademark dilution - although I personally have wondered whether a truly “professional” company wouldn’t be sure to buy up any domain names that are common misspellings, and have them redirect to the company’s website (try orbits.com, gooogle.com).

Google likes to claim it has no control over possible trademark violators using the Ad-Sense network, which I find hard to believe:

Hagan, Google’s trademark lawyer, said that software formulas aren’t smart enough to identify trademark infringements.
“It’s subjective when you look at domain names to decide how many letters off does it have to be to form a trademark or conjure up that trademark,” she said.

(The Web’s Million-Dollar Typos)

My fellow students who are more familiar with the inner workings of Google should check me here, but I think Google’s algorithms have this one down, particularly since the Google’s service Oingo.com specializes in domain parking and generates lists of common misspellings automatically. Additionally, I’ve noticed the efficiency with which Google regulates click fraud, and closes Ad-Sense accounts when traffic to the site has suspicious patterns. I think the company could easily control this practice if necessary, but its current practice of simply removing the sites if a trademark owner complains generates a lot more revenue in what is, if Google truly follows its policy of removing infringing sites, a victimless transaction.

The argument that Google’s domain parking and partnership with typosquatters generates profits “unfairly,” ignores the nature of “websurfing.” In my mind a user who misspells a domain is like someone who has gone down the wrong alley on his way to a store. Though the victim site’s owner may have a claim to all the revenue once a web user reaches his site, he has no claim to revenues generated along the way. Of course this all changes if the site knowingly imitates the victim website, of which phishing is an extreme example, emphasizing the need for Google to police such actions.

Furthermore, there’s no legitimate claim that typosquatting takes revenue away from the trademark owner. Because no confusion or trademark dilution has occurred, it’s unlikely that the appearance of a misspelled site will stop the user from continuing to the site he intended to go to.

A popular allegation is that typosquatter profit is “unfair” because it relies on aggregating, through buying many domain names, the profits earned by accidental clicks on ads. Yet I don’t think anyone would disagree that content websites also pay-per-click, maybe more, based on users who click accidentally. Profiting from user error is perfectly legal so long as a reasonable effort is made to inform the user against this error. And it’s this “reasonable effort” that I wonder couldn’t lead to a few “rules and regulations”

For example, typosquatters and domain parking services that knowingly generate misspellings of a trademark, could be required to redirect to the real site in a clearly visible manner. They could also be required to declare that they are not this website. I think a mere “did you mean to go to www.BestBuy.com?” could prevent a great deal of litigation.