Towards the Infrastructure for Attention Ecosystem

This article was originally published in Web 2.0 magazine.

With this article I hope to reach out to the companies and thought leaders working in the attention space and start the dialog on the infrastructure for the Attention Economy. Note that the views presented in this article are my own and do not represent the views of the AttentionTrust organization.

I have previously discussed here the exciting developments happening in the attention space. The Attention Economy and Attention Architecture are in the early days, but there are clear indications of growth. Ever since Steve Gillmor and Seth Goldstein co-founded AttentionTrust, the topic of attention is getting, well, a lot of attention. And very rightly so – AttentionTrust created principles that put the user in control of the data and open the door for new, exciting set of personalization applications.

The AttentionTrust was created based on the four core principles that guarantee that the users are in control of their data and that the privacy is always respected:

• Property: You own your attention and can store it wherever you wish. You have CONTROL.
• Mobility: You can securely move your attention wherever you want whenever you want to.
  You have the ability to TRANSFER your attention.
• Economy: You can pay attention to whomever you wish and receive value in return.
  Your attention has WORTH.
• Transparency: You can see exactly how your attention is being used.

These foundational principles embody the spirit of the Attention Economy and provide the rules and philosophy for building out the Attention Marketplace. In this article we will focus on exploring additional technical infrastructure that might be beneficial for enabling the users and companies to participate in an open Attention Ecosystem. In particular, we will discuss the format for storing extended attention information as well as the generalized interface for the Individual Information Storage Service. (Note that in previous articles I referred to this as AttentionVault and changed the name to remove any possible associations with AttentionTrust and RootVault)

The first blocks of the attention ecosystem infrastructure have been mapped out since AttentionTrust created an add-on to the Firefox browser called AttentionRecorder.

The recorder in its current form, provides basic, yet encompassing and essential function of capturing timestamped click stream, otherwise known as implicit attention.

The Diagram 2 shows the XML output of the AttentionRecorder, which is essentially a set of HTTP transactions.

In accordance with the Property principle the recorder offers the user a choice to either store these attention records in a local file or to direct them to one of the approved attention services. If the user chooses a remote storage, the records are sent via HTTP.

The current infrastructure facilitates capturing and storage of the implicit attention. In a way, it defines the interface for storing this type of attention, but it, intentionally, does not focus on the interface for accessing the stored attention records. Also, the format that AttentionRecorder uses today is not generic, but rather specific to the click stream. In the following sections I will discuss how the current infrastructure can be extended to accommodate additional types of attention and will discuss the interface for reading/writing the data from/to the Individual Information Storage Service.

The types of Attention
As we discussed, currently the AttentionRecorder captures implicit attention. There are also other types of attention. For example, bookmarks are a variant of explicit attention. The difference between implicit and explicit attention is that the user makes an explicit effort to store the information.

We can gain further insight into explicit attention by looking at the famous social bookmarking service called del.icio.us This services has popularized the notion of tags – the labels or attributes attached by the user to a piece of information. Seth Goldstein told me that del.icio.us’s founder, Joshua Schachter, thought of tags as a crystallized attention. The tags illuminate the personal angle on the information – what exactly the user is paying attention to here?

Like tags, the rating captures another aspect of attention. For example, the services like BlinkList and Ma.gnolia include the rating information in addition to the URL information. By rating an article the user specifies the attitude or the outcome of the attention. Recognizing the importance and the value of the rating, the sites like Netflix and Amazon have always allowed the users to rate movies, books , music, etc.

And there is yet another type of attention – explicit semantical attention, that is the attention related to objects like books, movies, music, wine and restaurants. The blueorganizer that we are developing at adaptiveblue focuses on capturing this kind of attention. Here is the XML illustrating how the blueorganizer represents books:

Abstracting Attention format
As shown in Diagram1 the Attention Economy brings together the users and the services creating an ecosystem where the users choose what attention services they want to receive. From the examples above it is clear that there are different types of attention: implicit, explicit, tagged, rated, semantical are just a few that we looked at. So to enable a flexible ecosystem, where the services can harness a variety of attention, we need to extend the current attention format to encompass the different types of attention.

Looking at the current format of attention recorder and the format used by the blueorganizer, we note that they are quite different except for the URL, title and the timestamp information. But really, the blueorganizer’s format just extends the attention recorder format, by adding semantical information about the object, in this case, a book, contained in the page. So what should the generic attention format look like? To answer this question we observe that it is not necessary to enumerate all possible types of attention in advance. Instead, we need to capture common aspects of different attention types and then provide a room for extensibility.

Since this an informal article, I will just give you a flavor for what this format will be like. At the same time, more formal specification will be emerging via discussions and feedback from the companies working in the attention space.

The format is very simple. The common attributes are grouped under the recordHead, while specific attributes are placed into the recordBody. Note that this format immediately facilitates decoupling and flexible paring of attention types and attention services as shown in the Diagram 5.

For example, a personalized recommendation service might take advantage of bluemarks, while personalized news service could leverage a combination of the user’s click stream, OPML and bookmarks. The generic attention format allows the user to be in control of different types of attention, and at the same time facilitates flexible attention services.

Abstracting the Individual Information Storage Service
Another piece that needs to be in place in order to make the Diagram 1 a reality is the interface for accessing the Individual Information Storage Service. Having this service is central to the Attention Economy, because it puts the user in control of the information. Without this storage, each personalization service would have to capture the user data separately and create, effectively, an information silo.

For example, the sites like Amazon have a rich history of the user interests, which can not be utilized anywhere but on Amazon. Clearly, Netflix recommendation engine could give better suggestions if it had knowledge of the data in the Amazon Wishlists.

The API for the information storage puts the user in control and solves the attention silo problem. The question is what should this interface look like? It needs to be simple (think del.icio.us) and include just the essentials to enable reading and writing the attention records. Here are the key operations:

AddRecord         – adds the attention record to the store.
DeleteRecord       – deletes attention record from the store
UpdateRecord       – replaces an attention record in the store
ListRecords        – lists the user records updated or added since given time
LastUpdated        – returns the timestamp when last add/update/delete occurred

Conclusion
The current attention infrastructure laid out by AttentionTrust is a major step towards enabling the Attention Economy. However, its primary focus is the capture and storage of the user’s click stream. To facilitate truly open ecosystem, this infrastructure needs to be augmented with generic attention format, allowing the other types of attention, and Individual Information Storage API for both reading and writing the records. With these two pieces in place, not only the users will be in control of their information, but they will also be able to receive wide range of attention and personalization services. The chosen service providers will be able to utilize a variety of attention data and make their services more agile, but more importantly better and more relevant to the end user.