SDForum's Search SIG: The Search for Attention

Tonight I'm at a SDForum Special Interest Group discussion on "The Search for Attention" at AOL's offices in Mountain View. Steve Gillmor, founder of GestureBank and co-founder of AttentionTrust, is hosting, and the presenters are Seth Goldstein of Root Markets, Gabe Rivera of Memeorandum, Dave Sifry of Technorati and Mike Arrington of TechCrunch. (Dick Costolo of FeedBurner was supposed to be here as well but is snowed in in Chicago.)

Seth leads off with a quick demo of the latest version of Root Vaults, which can take attention data captured by users of our Attention Recorder an display it in a variety of ways. Longer-term, Root plans to serve as an "attention broker" between users who elect to share their attention data with entities interested in accessing that data. (Disclosure: Seth was the other co-founder of AttentionTrust and continues to serve on our Board of Directors, but Root and AttentionTrust are entirely separate organizations.).

Gabe's up next with an introduction of Memeorandum, which "auto-generates a news summary every 5 minutes, drawing on experts and pundits, insiders and outsiders, media professionals and amateur bloggers."

Dave's next with an explanation of recent changes to Technorati that are related to attention. Their "Top Searches" and "Hot Tags" lists on the front page are updated frequently. (But by further promoting the top searches and tags, features like that can become self-fulfilling prophecies, yes?) Each search includes a chart showing how frequently that topic has been discussed over the last 30 days. They've added an "authority" filter to limit results for each search. Last fall they added "Blog Finder" to allow people tag their blog, and over 1 million people have done so. They're also using a PageRank-type algorithm to vet those tags.

Mike's last up to briefly mention TechCrunch, his blog where he writes about new companies.

New Steve Gillmor steps up to host the evening's second section. He asks Seth to describe what, if anything, was accomplished at O'Reilly's Emerging Technology conference, billed as "The Attention Economy" and Esther Dyson's PC Forum, both held within the last two weeks. Seth says:

There's a lot of anxiety among companies about what to do with users' attention because it's clear that it's no longer theirs to do what they please. At PC Forum, I asked reps from Yahoo, Google and other companies if I could obtain and use a copy of my search history. Their answer was essentially noncommittal.

Steve asks Mike to weigh in:

I use delicious and not Google because they allow me to take my data back.

Steve asks Gabe what's the most request feature for Memeorandum:

"My Memeorandum," i.e. personalization. (Although Steve notes that Gabe said on his podcast that it wasn't worth the trouble.) I think I can deliver a better service if it's not personalized. The key is whether a personalized service actually saves you time, and it's not clear whether "My Memeorandum" would do that.

Dave Sifry disagrees with that. He feels that RSS reader interfaces are fundamentally broken, and what needs to happen is the development of community-driven collaborative filtering that will give us all a personalized lens on the content we're interested in.

Question from the audience about how numbers can substitute for personal influence. A political blog read only by world leaders wouldn't figure heavily in Technorati, but its influence would be huge. Dave responds that the linking behavior of the Internet is a proxy for influence, and good writing will be discovered, linked to, shared and promoted.

Seth notes that the link model is fundamentally dead. "Influence," which is more subtle and multi-dimensional is replacing the link model's one-dimensional "authority." Dave notes that it's not "either/or" but "and"--multiple sources of "influence" and "authority" will continue to play a role in the distribution and dissemination of content.

Steve asks Gabe to comment on the idea that "links are dead." Gabe is skeptical, noting that our aggregated attention is still reflected in the number of links. Mike also notes that he feels that the number of links is still important--he may read hundreds of things, but he only links to one, and that gesture should carry some weight.

Seth notes that Josh Schacter has described tags as "crystallized attention," and perhaps there's a hierarchical pyramid reflecting the relative importance of different gestures, from explicit links to implicit attention.

Steve notes that links are being replaced, because they're increasingly undermined by spam blogs and other attempts to game the system. We need to focus on a tree of reputation or authority and understand how it will be mined by aggregators. That's what we really mean by attention--getting through the mountain of data that's available to us and extracting the most value in the least amount of time.

Mary Hodder (another AttentionTrust Board member) asks from the audience about the difference between "eyeballs" in 1996 and "attention" in 2006. The panel ducks, but Mary explains that the difference is participation. Today we can demonstrate our interest, our attention, via a whole range of participatory gestures--not simply by clicking on a link. And E-Tech fundamentally missed that perspective.

Dave agrees and notes that it's the same as the difference between consumers and participants. In the world of "eyeballs," we're "consuming content and crapping cash." In the world of attention, we're actively participating in the growth of the site, the growth of my community.

Someone from the audience notes that "attention" isn't just hits on a weblog, but vast troves of deep demographic profile data, and the companies that currently have access to that data have to be threatened by the "attention movement's" emphasis on user control. Seth agrees and notes that millions of people are essentially giving away huge amounts of valuable data by filling out various forms, which comprises a major arbitrage opportunity, and one role Root hopes to play is helping people capture some of that value for themselves.

Moving on, Dave notes that he sees two big trends: mobile usage and tool creation. What's interesting about this is that we're throwing away enormous amounts of data generated by our attention--what if you had a tool that allowed you to collect that data and share it in ways that were useful to you? (I'd say that's exactly what the Attention Recorder hopes to do.) He raises the point that Doc Searls, among others, is talking about the shift from attention to intention.

Seth notes that AttentionTrust, which he and Steve co-founded, exists to educate people about the value of that attention data. Steve add that the fundamental principle of AttentionTrust is that we, the users, own our metadata, and the Attention Recorder, puts teeth in that principle.

But the problem, Steve adds, is that there's no open, anonymized pool of aggregated metadata. Dan Farber asks just what we'd do with that pool--Steve responds that we "swim in it"--we'd compare non-anonymized, individual data against the pool of aggregated data in order to make inferences about the meaning of that individual's data. This a complex but really big point, and it leads directly into what Steve is planning to do with GestureBank--nothing to link to yet--which will serve as that aggregated pool of anonymized metadata.

There's an extended discussion about whether we can get our data back from company silos, but Steve notes that it fundamentally doesn't matter if we empower users to capture their own attention data for themselves (and our Attention Recorder) is a first step in that direction.

tags: attention sdforum