Attention on the brain

Attention was the subject of the SDForum Search SIG tonight at AOL's Mountain View office. In fact, it's been the subject of attention and lively debate for the last few weeks, especially at the recent ETech to PC Forum conferences. 

In my post from ETech, "The amorphous attention economy," I outlined various views on attention espoused by the presenters. What's clear from that post and the discussion at the SDForum SIG is that there isn't a common language that makes it easy to discuss the meaning, implications, rationale or basis of "attention" or "inattention." From what I've heard over the last few months, attention is about time optimization, improving quality of life, mental property ownership, click stream dating, economics, scarcity, smarter search engines, participation, intentions, intensity, gestures, focus, myware, brokering, anonymized data banks and more. Are you paying attention?

Attention has a practical and specific online context. It has to do with managing the flood of data and advertisements vying for your time, which is finite. The metadata 'exhaust' (feeds, clicks, links, search terms, etc.) you create can become a useful tool to help filter and funnel information so that your time isn't wasted on irrelevant or marginal information you encounter online. It filters what enters your field of attention, based on implicit and explicit data. Attention is also about people and participation (tip to Mary Hodder)--who you influence, who influences you, who you choose to pay attention to or to avoid.

There is also the notion of 'intention economy,' which Doc Searls says is about people engaging online with a clear notion of what they want. "I want my intention to buy or find something to be serviced by the system, whatever it becomes. Hearing 50 companies tell me how they are competing for my attention doesn’t cut it; they are still looking at me as an eyeball,"  Doc said. It inverts the traditional advertising model.

Technorati CEO Dave Sifry extended that idea succinctly during the Search SIG discussion: "In the world of eyeballs, a good consumer is someone tied to chair consuming content and crapping cash. Participating in the growth of a site or community is about contributing to somethng larger than myself and receiving something from others."

Enough said as a preamble to the demos and sometimes testy discussion that took place at the SDForum event, whch started with brief product demos. Seth Goldstein showed Root Markets' Vault, a 'web trends' for people that allows them to record and data mine their Web tracks and a trading platform for the exchange and pricing of user data. Root intends to give users the option of brokering their attention data with those who want to access it, such as marketers. (I blogged about Root here and here.)

From left: Gabe Rivera, Mike Arrington, Seth Goldstein, Steve Gillmor, Dave Sifry

Gabe Rivera show his popular memetracker, Memeorandum."I designed this to be easy to understand so if it requires a demo I haven't done a good job," Gabe began. Memeorandum, along with its brethren such as TailRank and Megite, are basically content filters, using algorithms and some hand tooling to surface what is getting the most attention among users and therefore will get even more attention when displayed on the memetracker site.

"The most most linked articles are the highest ranked," Gabe said. He uses a source list of popular sites as a starting point, which auto-discovers a broader source list to define a section, such technology, politics or gossip--Memeorandum's three flavors today. Gabe is also planning to look at words and phrases to improve the results. "Lots of of people are talking about things, and maybe in aggregate it's not the case of one blog post getting a lot of links, but some concept spread acorss many blog posts," he said.

Dave Sifry demoed his blog search engine, which also has features that help take

advantage of aggregated data. Technorati calculates blogger "authority," based on link data (how many people, rather than pages, link to a blog ) and some natural language processing as a prism for collaborative filtering. A slider lets you refine search results based the level of authority.

The demos were followed by a discussion led by attention guru Steve Gillmor.  All of the panelist see the potential of mining clickstream data. "We are still throwing away an enormous amount of data about the click stream. Imaging a tool that collects that information for you and allows you to divvy that up in ways userful to you," Dave Sifry said. He envisions have a tool at Technorati that watches your behavior online and transparently customizing the experience for users.

Gabe Rivera said that he isn't taking advantage of counting clicks or data such as a how long people stay on the site, and doesn't believe it will help improve his memetracker. "This could be part of big formula, but it's not to open entire new worlds of information," he said.

Gabe noted that despite request from users for personalization features, he doesn't think it would improve the results significantly. "The goal is a great goal, and the ideal people have in mind is the Holy Grail, but I don't expect the results that I could produce would be worth the trouble. I can produce something better if it weren't personalized. The key thing to evaluate in any personalized filter is does it save me time, and not only because you are no longer reading things that you were reading before." He proposed a simple test On any give day if if the aggregator presents 200 things, supposed that 20 are really valuable. The test of a personalized filter would be that those 20 things show up. "I'm not confident that today's technology can produce a filter that is good enough," Gabe said.

Dave disagreed. "The algorithm Gabe is describing is almost unattainable, but bringing useful discovery for people is not impossible," Dave said. RSS aggregator interfaces aren't ideal attention amplifiers, but the community you associate with can be a filter. "That means sometimes I don't see everything on Memeorandum, TailRank or Megite, but that's ok. If it's really important it will pop up or I will learn about it through the community. With Favorites [on Technorati] users can define their favorites, such as bloggers and  RSS feeds, and then allow the larger blogosphere or some sub-segment to be the collective filter." 

Dave concluded that linking behavior is an overall proxy for authority. "In the world of 31 million bloggers and 70,000 posts an hour, you have to find some proxy or ways to make sense out it. Authority is a pretty good first derivative indicator over time." So, attention in this instance, is about saving time and increasing the utility of the information that enters your brain, and is measured as authority in Technorati's schema. What's of value to the community is discovered and circulated based on that concept.

Seth countered that "influence" rather than link-based authority is a better measure of attention. "Influence is a measure of the attention you get and what you give," Seth said. "By paying attention to people I trust, who have influence over me, I can pay less attention to people and items I don't trust.  Therefore influence in some strange way subsidizes (or maybe the better word is simply prioritizes) my attention," Seth further explained in an email to me.

Attention is therefore influenced by who you trust, which is related to who is viewed or algorithmically calculated as authoritative. Dave replied that influence and authority aren't mutually exclusive. Mike said that the linking "gesture" is "extremenly important and relatively free for blogopshere to harvest, facilitating traffic movement around the Internet."

Seth brought up Del.ic.ious founder Josh Schacter's concept of tagging as "crystallized attention." Dave noted that a news reader with an OPML file is an "enormous benefit and can be shared with services to go good things for you." Seth described a deep anxiety among the incumbents, such as Amazon  and Google, about sharing the attention data. "It's not just there's to do with what they want," Seth said. Amazon developed an API so that customers could scrape their shopping data, but applying it to other services isn't a reality. Mike Arrington of TechCrunch said he uses del.ic.ious precisely because he can take his bookmarks with him. "That kind of openness I applaud. That's why I use del.ic.ious and not Google--there is no hope they will let my data out." When asked at PC Forum whether users have rights

to their data, Google head of sales and business development Omid Kordestani was non-commital:  "If it makes sense, we should probably allow it," he said. Seth said that allowing people to use their Google data outside of Google would have "deeply profound implications." Yahoo's Jeff Weiner has stated that would be fine with sharing data with users, scenario, which certainly make it easier to move from site to site and to apply your data fully across the Web.

AttentionTrust.org, which was co-founded by Steve Gillmor and Seth, was established to exert greater control over their attention data, to educate people about attention and to build a "community of individuals and organizations that will guarantee users' rights to own, move, and exchange their attention data, in a transparent environment that gives users the freedom to decide how their data will be used." At the Search SIG, Goldstein announced that the non-profit received funding from the Omidyar Network. How much, he wouldn't say.

AttentionTrust.org has an Attention Recorder (Firefox only for now) that allows users to capture parts of their clickstream and history. "The Attention Recorder was the teeth that allowed to us to begin the process of recording the data before it was scraped and cannibalized by Google and eveyrbody else," Steve said. According to Seth, one of the outcome of users taking control of their attention is that the individual become the 'unit,' not the corporate database.

Steve also announced that he resigned from his presidential post at AttentionTrust.org to focus on a new venture,GestureBank, an anonymized pool of aggregated metadata that could be used to compare and make useful inferences against individual data. "At the end of the day, we won't solve the problem about user rights unless the user maintains their fundamental rights," Steve said. "Users want to take services and be able to contour them to their interests, represented by their own perpsective and by the people, institutions and groups that influence them. Harnessing that power is what gestures are about." Steve said that a pool of 1,000 people pooling their data could give statistically relevant results for making inferences.

During the taping of the Gillmor Gang today, Steve expounded on the role of GestureBank. AttentionTrust.org's Attention Recorder doesn't have a mechanism for aggregating the data in contract with users that would allow for an open pool of metatdata, in contrast to Google, Wal-Mart and every other multi-national on the planet with proprietary inference engines, Steve said.

"Improving this last mile or foot from the screen to your brain is where all the action is," Steve said. "GestureBank will be one of the primary vehicles for improving that beyond strategies heard about today. An open pool of metadata will validate and create an inference engine for the gestures of people, who have attributes that turn out to be proper for improving the system....Users have to able to control and own their own gestures [clicks, votes, bookmarks, ratings, links, RSS feeds etc.] and  be free to license them, based on their ability to get compensation in terms of money, other services and reputation. You can't partake of the ecosystem unless you contribute to the pool."

As I said at the beginning of this post, the language of attention is all over the map, but there is coherence in the idea that your attention has value, and there is value in bringing to your attention what matters to you. Just as siloed information fails to provide a fully mature or relevant version of the truth for a corporation, it limits how the Internet can enable users to seek, find and resolve. Recording your Web metadata and pooling it represent mechanisms for fine tuning the system to do your bidding. You can imagine this Web collective or coop in which the attention, intentions or goals of individual users or groups are brokered in a near frictionless environment, but getting from here to there isn't at all clear...