The idea of the “Implicit Web” is based on the fact that users are defined by the websites they visit, indicating what they like, and by the websites they avoid, indicating what they do not like. In turn, users change the world by those sites they visit and those they avoid.[1]

This happens when users stop to read an online article, recommend a favorite movie to their family and friends, and play a favorite song over and over again. Yet, virtually all users never stop to think about how valuable that information is, to both us and to advertisers and companies trying to gain their attention.

Companies do not always use implicit means in order to obtain information; often users themselves will give out that information when they rate their favorite news stories, music, movies, and more. Sites like Digg.com, reddit.com, Helium.com, and others like them utilize their users’ votes to determine in what order stories should appear on their websites.

One of the main elements of Web 2.0 is the fact that users help to determine what information is most important, what movies and music are the most popular, and what products, services, and websites are considered to be the best. So attention economy and Web 2.0 are closely related to the success of popular and successful businesses and websites, as well as related to the expansion of the Internet itself, since Web 2.0 requires user feedback and continues to take the Internet by storm.

Advertisers and companies are focusing on software that uses complex algorithms to find out what exactly users like and then present options that the software believes those people would also like. Some websites already use this strategy in order to make their websites more personal to users in an effort to attain and keep their attention. Four examples are Last.fm, Google, Amazon and Netflix.

Last.fm

Last.fm is a service that provides users a plugin into iTunes and other major music players. It captured the songs and artists users chose to play, then used that information to infer what artists were their favorites.

What was revolutionary about this service though was that this site took the next logical step to keeping your attention: it introduced you to new artists and new songs that were comparable to the artists and songs you already listened to. As a user, this would make you very happy because this was all done automatically; the software would figure out the type of music and artists you liked listening to and then calculated what new artists and music best matched up with those parameters you gave via the songs you chose to listen to.

The idea is certainly a valuable one, as the CBS Network bought Last.fm for $280 million in May 2007.[2]

Amazon.com

Amazon.com has been utilizing the Implicit Web for a while now, as every time we click on a product link, it shows us recommendations on what we might be interested in based on what we click. It does this via algorithms that calculate what we would likely be interested in based on what we click.

Not only does this occur in the “New For You” section, but it also occurs in the “Recommended Based On Your Browsing History” section, as the algorithms also calculate any related items that match up or can be utilized by the items you clicked onto, and in the “Others Purchased” section, where items that are comparable to what you are looking at and that other people have purchased are displayed.[3]

It took Amazon over ten years to build and perfect a system that is able to tap into your browsing history, past purchases, and purchases of other shoppers in order to encourage you to buy products that correlate with that information. It’s even more remarkable that the system is able to “remember” what you clicked on a few minutes ago or a few years ago.[4]

Google.com

Google is another company that utilizes our implicit behavior in order to show us search results we would likely be interested in. Clicks are used as feedback for its complex, ever-changing algorithm.

The big problem with this is that Google does this automatically, without the user even knowing her information is being gathered and used. That is why Google was recently criticized for not respecting user privacy.[5]

Privacy International published a report in September 2007 that identified Google as having the lowest rankings when it came to consumer privacy, due in part to the broad spectrum of Google’s product services and their ability to share extracted data between these services, Google’s market dominance and the size of its user base, and their “aggressive use of invasive or potentially invasive technologies and techniques.”[6]

Netflix.com

Netflix rents movies over the web. As amazon, it is crucial for Netflix to anticipate what users will like and suggest it. Personalization algorithms are so important for Netflix that they offered a large dataset and their reference algorithm to beat in a public contest. Netflix provided 100M ratings (from 1 to 5) of 17K movies by 500K users. These essentially arrive in the form of a triplet of numbers: (User,Movie,Rating). They hoped to attract the most talented machine learning people that way. An increase in a few percentage units in predicting the client tastes makes a big difference in the average number of movies rented.

The fact is that Netflix, like any major corporation cited here, they kept as much of their user’s attention traces as possible, and have a research department worried with exactly one thing: how to use that information (the cloud that you left behind) to make more money.

One criticism is that Google, Amazon etc are using your attention as if it was their property. In fact, Google determines what you will read (ads included) every time you click on their search button! And Google gets cash as an advertising company more than anything. Of course, if their algorithm brings up only content that has paid to be there, people wouldn’t use it… So Google walks a fine line between offering relevant content and profiting from that content. Some content may be less relevant and more profitable. Every single page online plays that game: some sell links to less-relevant content, but attract users.

Wikipedia mentions that “The paid inclusion model, as well as more pervasive advertising networks like Yahoo! Publisher Network and Google’s Adsense, work by treating consumer attention as the property of the search engine (in the case of paid inclusion) or the publisher (in the case of advertising networks).”

The fact is that there is a lot of information about us already out there ready to be mined. And this brings up privacy concerns. Most social web 2.0 sites have users displaying very personal information. Bloggers give you details of their lives you don’t want to know. You can have full access to demographics and bio of mostly anyone with a facebook profile. And there are services that specialize in aggregating those bits and pieces about you (and maybe sell them to the highest bidder).

On August 4, 2006, AOL released a compressed text file on one of its websites containing twenty million search keywords for over 650,000 users over a 3-month period, intended for research purposes. AOL pulled the file from public access by the 7th, but not before it had been mirrored. AOL apologized for releasing the data into the public domain. Even though AOL used numeric codes for each user, discovering the identity of some users wouldn’t be that hard–if someone wanted to try. That’s because people search on their own names (ego surfing) and also search telephone numbers, social security numbers… and names of (ex)loved ones. The New York Times (Barbaro & Zeller, 2006) [8]published an article that proved that one could reconstruct quite a lot of information from the internet searches AOL published, and that this was an implicit risk. The key factor here is informed consent: the NYT-interviewed user was right in that “[She] had no idea somebody was looking over [her] shoulder.” Worse, users felt they had no control over those data.

Note that AOL’s users did not sign their consent for these data to be collected, and they did not have direct control over which data was stored. They often did not know that such data were being collected.

AOL uses google as their search engine backend, so google has at least as much information about you as AOL had about their users when this database was released.

The basic idea is giving users the power to administer their attention profile instead of leaving it behind for any company to profit from it. Plus, the information that is ‘already there’ or ‘already being collected automatically’ does not have user’s consent (or this consent is implicit at best). If users could encapsulate their attention profile, then anyone wanting to use it would need to ask for explicit consent.

Of course, the danger of consolidated information in one place: if an attacker breaks into your attention profile, he will have more data available than any individual organization has now (including most governments’ intelligence sections). However, it could be that we are actually safer this way. From Clevercogs[7]: “The privacy aspect of attention profiling is brought up quite often when I talk to people. They consider their [attention profile] as their private property and are usually afraid their browsing behavior will be exposed to prying eyes. I look at the privacy aspect of attention profiling from a different angle: right now sites like Facebook and Google collect usage data from and about me. They know about my interests, they know what sites I open and they know who my friends are. At the moment all this is a one-way operation: they collect the data that I give to them and I get no insight as to how they filter the content they or their advertisers offer to me. I prefer to have that information distilled into an attention profile so that I can at least have control over whom I share this information with.”

The important thing here is who has the control: the companies (as it is right now), or the user.

[1] http://www.readwriteweb.com/archives/the_implicit_web_lastfm_amazon_google.php

[2] http://www.readwriteweb.com/archives/the_implicit_web_lastfm_amazon_google.php

[3] http://www.readwriteweb.com/archives/the_implicit_web_lastfm_amazon_google.php

[4] http://www.readwriteweb.com/archives/recommendation_engines.php

[5] http://www.readwriteweb.com/archives/the_implicit_web_lastfm_amazon_google.php

[6] http://www.privacyinternational.org/article.shtml?cmd[347]=x-347-553961

[7] http://www.cleverclogs.org/2007/10/basics-of-atten.html

[8] Barbaro, M., & Zeller, T. J. (2006). A Face Is Exposed for AOL Searcher No. 4417749. The New York Times.