The Implicit Web: Last.fm, Amazon, Google, Attention Trust
As we rush through life and a myriad of web pages, we leave traces. We don't have time to think deeply about any of this. The blogs that we visit, the music we listen to, the movies we watch; we take all of them for granted. Yet, all of the different kinds of information that we interact with defines us. In turn, we change the world by issuing a verdict of what we like and what we do not like. Sometimes explicitly via bookmarking, rating and digging. But most often, implicitly.
The concept of the Implicit Web has been around for some time. In 2005 Seth Goldstein coined the term MyWare. Seth explained that sites that we visit reveal what we like and do not like. He formed the Attention Trust, an organization that advocates consumer rights and ownership of your attention information. But perhaps the most elegant and succinct formulation of the value of myware came from Fred Wilson. He said: "If someone is going to spy on you, it's probably best if its you."
What is the Implicit Web?
The basic concept of the Implicit Web is simple. As we touch information, we vote. When we come across an article we like, we spend time reading. When we like a movie, we recommend it to our friends and family. And if a piece of music resonates with us, we listen to it over and over again. We do this automatically, subconsciously or implicitly. But the consequences of our behavior are important. The things that we are paying attention to have great value to us, because we like them.
Things we pay attention to
What if there was a way to capture our choices automatically? What would that mean? Firstly, as Fred Wilson put it, we would be able to spy on ourselves. That is actually not as strange as it sounds. Just as we often refer to our respective browsing history, it would be great to automatically refer to the things we liked. Secondly, if our web-wide preferences were accessible to different web sites and web services, then they would be able to do a better job of personalizing our online experience. From recommendations to look n' feel, a Web where our implicit preferences are automatically considered, would feel a lot more personal.
The Mechanics of the Implicit Web
The Implicit Web is powered by clicks. When we click on things, we vote. When we spend time on a page, we vote. And when we copy and paste, we vote some more. Our gestures and actions reveal our intent and reactions. While it is impossible for an algorithm to always capture our intent with 100% certainty, a lot of software has gotten pretty good at doing it.
Typically, software that powers recommendation engines or search engines takes clicks, time and actions as inputs and feeds them into a sophisticated optimization algorithm. Usually, the algorithm assigns weights to different input parameters and then outputs a verdict - e.g. how much the user liked or disliked something.
Last.fm - the Implicit Web at its best
Perhaps the most successful example of the implicit web up to date is Last.fm, which was recently acquired by CBS for $280M. Part of the reason that Last.fm was so successful is that it was completely automated. The service offered a plugin into iTunes and other major music players. The songs that you played were automatically captured and added up. Based on this information, the service then automatically inferred your favorite artists.
And then Last.fm took the next logical step - introduced you to people and new music based on the music that you already liked. What is important is that Last.fm delivered a service where users don't have to do any work. As a user, this is an attractive proposition. If all you need to do is listen to your music, and the rest is done for you automatically, then you are likely to be a happy user. CBS validated the model by writing a large check. Let's hope they can preserve and enhance the service; and keep the users happy.
Amazon and Google - The Implicit Web around us
We didn't notice this (perhaps this is ironic!), but the Implicit Web has already arrived and is working for us every day. Amazon must be the grand daddy of the Implicit Web, although limited to its domain. Every time we give Amazon a click, it uses its sophisticated algorithms to suggest to us what we should buy next:
In our post on recommendation engines, we described different ways that Amazon leverages its system for each individual customer. This complete system, based on users implicit choices, is at the heart of Amazon's selling strategy:
Another company that has been leveraging our implicit behavior is Google. Each click has been used as feedback into the complex, ever-changing algorithm that Google uses to show us search results. Google does this completely automatically, without asking its users; and because of that it has recently been criticized for not respecting user privacy. That issue with Google gives us a good chance to clarify an important point; and also to lead into our next segment. In the Implicit Web, the user's privacy is of paramount importance. Nothing should be done without informing the users and giving them a way to opt-out.
AttentionTrust - ensuring that users are in control
The problem with our fast-paced tech world these days is the lack of standards. We are rushing to create the Next Big Thing, often without establishing the proper infrastructure and standards to lean on. For the Implicit Web to work, users must be in control. No program and no company can claim the ownership of your information. Myware is not another kind of Spyware.
Inspired by the vision of users owning their information, Seth Goldstein and Steve Gillmor co-founded AttentionTrust, a non-profit organization that we covered extensively in our Attention Economy post. The mission of the organization is to establish the basic framework for discussing ownership of people's attention information. While it has not been a smooth ride so far, the founders put forth a set of governing principles that capture the essence of attention data - i.e. it belongs to the users. The four basic principles behind AttentionTrust instantly make any use of our information subject to our approval. According to AttentionTrust, in the Implicit Web the users call the shots.
Conclusion
So there is a paradox of sort. The Implicit Web is already here and is being actively used by many companies. Yet, consumers are not aware of it - at lest not explicitly. And that is not a good thing. The Implicit Web and its close cousin Attention Economy are big and important concepts; and they are still in their infancies. What we hope is not going to happen is that people get turned off because of mishandling by big companies. So having an organization and standards around this very sensitive subject is really important.
Regardless, the companies are not going to stop leveraging implicit behavior because of business and innovation. The competition is tough these days, so whatever companies can do to get ahead, they will. Given that consumers are pressed for time, the algorithms that discern their intent correctly are likely to win. So not only we already experiencing the Implicit Web already, we are likely to see more of it. Indeed, perhaps the whole web experience of the future will be like the auto-complete feature of word processors. How does that idea sound to you?
Editor's note: if the topic of Implicit Web interests you, it will be covered at the upcoming Defrag conference (where Read/WriteWeb is a Media Sponsor and Alex's company AdaptiveBlue is a silver sponsor).
Share8 TrackBacks
TrackBack URL for this entry: http://www.readwriteweb.com/cgi-bin/mt/mt-tb.cgi/2267
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
Great post!
Of course the ascendancy of implicit ranking systems begs the question: where would explicit moderation bring something big to the table? Explicit carries it's own problems (gaming, abuse), but I think many are solvable. I've said so about search before, e.g.: http://valleywag.com/tech/informed-speculation/jason-calacanis-next-venture-256549.php#c1372120
I have a fundamental issue that a users click actions on a site ‚Äúbelong‚Ä? to them. If Staples has video cameras in their store to watch the behavior of shoppers does that footage belong to Staples or the shoppers? Also, click data is great but it never unlocks the latent intentions of users. Often this ‚Äúlatent web‚Ä? is where the best opportunities exist.
Really all your talking about here is analytics and optimization, namely through BT. Much of this data is collected and acted upon without PII. There’s a lot worse going on with PII in the offline world than online. Also there are studies that users frankly don’t care about data collection as much as they say (or you say) they do. However, the sensitivity here is surely appreciated and users are in control of their actions.
At the end of the day this is all about creating and delivering relevance. The businesses that do it the best will be the winners because user will approve of it by those very action they control, not their words, or the words of others.
"The four basic principles behind AttentionTrust instantly make any use of our information subject to our approval"
I hear this type of statement all of the time and I always wonder what exactly is meant by this?
if by "our information" you mean "our real name's", "our real address", "our real phone #", etc... I would definitely agree with this statement and only in the case where they were giving it to another entity without your knowledge or approval.
If by "our information" (and every scenario here assumes that in most cases you will be using an anonymous user name for sites that you have an account with)you mean "what links you clicked on on the site", "how long you spent there", "what browser you used to view it with", etc... basically, tracking your use of the website.
I would have to say that I have a problem with the notion that people think that they would "own" that information & should have the the right to tell the website owner that they can not record their anonymous use on the site and but still provide them access to the site.
In all reality, you already have the option to "opt-out", do not use the site in the first place, nobody is holding a gun to your head and forcing you to use the site.
I would agree that it is probably wise for all sites to inform users of what they do track, what services they can offer to the user as a result of that, and what, if any, info they can provide to other sites and tools outside their site that a user also uses that may be of benefit to them.
In the last scenario mentioned above, they will have to get your approval as you would have to provide them with the necessary info (like your user name and the service that you want to integrate).
The simplest way I can put it is, they own what you do on their property, you have the choice of whether or not you want to interact with them on their property.
I work in the property management industry and if you want to rent an apartment from us you have to give us you SSI, employer, income, we check your credit and criminal history, and you would be laughed right out the door if you told us we had to rent you the apartment without that info, and believe me, we use that info for many purposes when it comes to many different factors that effect others that rent from us, such as, how much of a deposit to charge for someone who has negative credit history which is based off of the history of those with similar data and how likely they were to skip or have to be evicted and how much more that costs compared to those with better credit scores, again, you would be laughed right out the door if you tried to tell us that we could not use your use of our properties because you somehow owned it and had to give us permission to use it.
In the end, I know there are a lot of gray areas here that I might not have thought of yet, so I am curious as to what you actually mean by your info?
There is a cool service that keeps track of the use of your software. I use it to see how I spend my time behind the computer and to discover new software.
it's called wakoopa. ( http://wakoopa.com )
The FAROO peer-to-peer web search is using the concept you described for ranking the search results.
@BrianR Its the click stream and you are hitting right into the center of the problem. Please read this post for more:
http://www.readwriteweb.com/archives/open_data_workshop_2007.php
Alex
I would have to say that I have a problem with the notion that people think that they would "own" that information & should have the the right to tell the website owner that they can not record their anonymous use on the site and but still provide them access to the site.
anyway,thank u for your information
Very good post. Thanks that you include this in your page. I would like to ask if I could translate it and put it on my site in section about interesting topics. If that would be possible email me. Really great text. Greetings!
Great post, Alex. I agree that we will see more of the Implicit Web, and in my opinion the following is essential to the success of Implicit Web initiatives:
Profiling -- it's all about *appropriate* and accurate profiling of behavior and intent, which can be scary to people.
User Value -- you have to directly and transparently couple a user value proposition to the act of profiling, such that the deeper and richer your profile the more value you receive as a user. This value prop can't be fear-based, it has to be interesting, motivating and rewarding.
Privacy Controls -- users have to be in control. If your profiling quality and the coupled user value propositions are outstanding, then why not let users edit their own profile, or even allow others to use it (with permission)?
No work -- this is why social search (in its current form) won't work. It requires explicit user action (bookmark or a manual vote of some sort). I believe the Implicit Web has to leverage the activities of the user already, their publishing and browsing. No incremental work is required to achieve incremental benefit.
The success of AttentionTrust is dependent on a user revolution against all the Web sites out there, because there's nothing in it for the Web sites. That ain't gonna happen, as evidenced by some of the previous comments. Nonetheless, I believe there IS an opportunity for a business to be a central repository and broker of people's behaviors/intentions/context, provided the above requirements are met.