November 2nd, 2005
Moving From Average Value to Personal Value in Search/News
by Joshua Porter | 
7 Comments
A great discussion came out of The Testosterone Meme by Shelley Powers, who wrote about her lack of confidence in the new aggregation service Memeorandum. Shelley (and several commentors) noted that the posts that become popular on Memeorandum tend to be A-List bloggers. The A-List bloggers tend to be men, and they tend to have been around the Web for a while.
Shelley’s argument is that there isn’t enough diversity in the aggregation service. She feels that there aren’t enough women represented, and there aren’t enough voices of “little guy/gal” bloggers who write interesting and relevant content but aren’t so well connected to be part of the A-List.
The Echo Chamber Dilutes Value
Shelley, of course, is right. We’ve long known that there is an echo chamber in the blogosphere, that certain ideas are amplified depending on who said them and who linked to them. As a result, though, several commentors were very critical of Memeorandum, but Gabe Rivera (creator of Memeorandum) stood his ground and pointed out that every news service is biased in some way.
As the conversation got more interesting, I sided with Gabe, arguing that Memeorandum accurately reflects what I see going on in the blogosphere, for better or worse. I think Gabe has done an excellent job keeping out the noise, as I’ve written before, but I do feel that his strategy of starting with a few known bloggers and letting the service grow outwards will eventually be limited. But you’ve got to start somewhere, right?
At this point the conversation turned into a sort of anti-A-List conversation, with folks like Dave Rogers arguing that this is all just a popularity contest.
There may indeed be many elements of popularity here, but what human systems don’t include them? Isn’t it human nature to listen to people you’ve listened to before, and have gotten value from before? And isn’t it OK that they are your friends?
The Attention Economy & Unequal Value
Ok, so this stance gets me in hot water all the time. My view is that in the attention economy, you can’t just get attention for nothing. Just like money for nothing. You have to earn it, and if that means that you need to know the right people, or have certain people to link to your blog, then that’s the challenge, isn’t it?
The major argument I get is that I’m too willing to accept this as simply the rules of the game. That I’m saying it’s a market economy and it’s OK. People don’t like that: they think that it’s unfair, and that the people who enjoy the privilege of being A-List bloggers enjoy more than their fair share of attention.
So let’s assume that to be true: the long tail of blogging is sloped too steeply. What can we do about it? Can we build a system that accurately reflects value? Does one already exist?
Google Approximates Average Value
What about Google? Google does approximate value by giving weight to links. However, the increasingly obvious problem with Google is that it is approximating an average value of content, not personal value of content. This is the same problem that we have with the A-Listers. They’re not approximating value for us, but for themselves.
In other words, if I want to know what the world on average thinks is valuable, I can go do a search on Google. But if I want to know what things would be most valuable to me, I can’t go to Google. Google isn’t just approximating the weight of my links, it’s approximating the weight of everyone’s links.
Years ago, Search was so bad that this was more than enough. Now, people are getting restless with services like Google and Memeorandum because they’re not personalized enough. This is a good thing: it means that our software is improving!
Huge Opportunity to Approximate Personal Value
So there’s an opportunity for an amazing product in here. If someone can figure out how to approximate the personal value that we place on things (by looking at what we link to, what we pay attention to, what our closest friends pay attention to), they’ll create the next generation search engine. This will succeed because it is user-centric: in the end what is important to us is personal value, not average value.
Anybody know of a product that is attempting to do this?
ABOUT
Bokardo is a blog about interface design for social web sites and applications. I write about recommendation systems, identity, ratings, privacy, comments, profiles, tags, reputation, sharing, as well as the social psychology underlying our motivation to use (or not use) these things. If this sounds interesting to you, grab my RSS Feed. If you want to know more about me, check out my about page.
Designing for the Social Web
Building a social web site or application? I wrote a book just for you!
Find out more or order from Peachpit or Amazon
Greatest Hits
- Five Principles to Design By
- The Del.icio.us Lesson
- Designing for the Social Web: The Usage Lifecycle
- Facebook's Brilliant but Evil Design
- Digg's Design Dilemma (Part II)
- 7 Reasons Why Web Apps Fail
- 7 More Reasons Why Web Apps Fail
- Which Movie to Watch? An Overview of Recommendation Systems
- Ugliness, Social Design, and the MySpace Lesson
Upcoming Speaking Events
- Webstock '09
Wellington, NZ / Feb 16-20, 2009
LATEST POSTS
- Are Designers also Marketers?
- Book Recommendation: Letting Go of the Words
- The Live Web
- Slides from Leveraging Cognitive Bias Talk
- Activity-Centered Design
- Free chapter of Designing for the Social Web available
- My Proposals for 2009 SXSW Talks
- Passionates
- Tripit Employee Pretending to be Disgruntled Dopplr User?
- Can Interfaces be Evil?
Comments ( 7 Responses so far )
1. JS on November 2nd, 2005 (Comment) #
I think people are working on parts of this problem. Rollyo, for instance, lets me search a collection of URLs that I’ve put together and labelled. You can see my collection, though I don’t know that you have any particular incentive to do so — maybe if I’ve named my collection descriptively and you are interested in the topic. This might be a better indication of who one thinks is knowledgeable, as you get no real juice from including a site in a searchroll, the way someone would be linking in a blogroll. Playing to vanity isn’t appealing when it comes to search; it just adds noise to the results.
Del.icio.us can be used for part of this, too — I can look to see who else has bookmarked the things I have, or used the tags I have, and see what else they’ve done. Del.icio.us now recommends other users or other URLs to me, and near as I can tell, these recommendations are not based on who is on the blogger A-list.
I often squirm when I read the “where are the women” posts. Sometimes think they make valid points, and sometimes I think they are just noise. One of the best I’ve seen recently was by David Weinberger — on his giving up a spot so organizers could invite someone else, a woman. I liked his post because it was about taking action to make things different. That is what women who blog are doing. And either we’ll get picked up by sites like Memeorandum, or we won’t. I’m busy, and I have better things to worry about.
2. Andrew on November 2nd, 2005 (Comment) #
You write: “You have to earn it, and if that means that you need to know the right people, or have certain people to link to your blog, then that’s the challenge, isn’t it?”
That exactly right. If people want their writing to appear on sites like Memeorandum, they need to write more often, more clearly, and more insightfully. And they need to publicize themselves in some way, and get people to link to them. We all know how this game works, and it ain’t that un-linked, isolated sites somehow get magically noticed by spiders.
Also, it’s wrong to equate what you say at the beginning of this post–that there’s not enough diversity represented in the aggregation space–with what you say at the end–that people complain Google and Memeorandum are not “personalized” enough. Personalization != automatic diversity. Maybe I mis-read your argument here?
“…personal value, not average value.” Not necessarily. Both are important; it’s not an either/or situation. I *do* want to know what news is relevant, on average, to most people. I *do* want to know about things like top ten lists, or most-linked to, or bestselling. Knowing those things, even if I don’t like them, or agree with them or seek them out, means I am a member of society. If anything, hyper-personalization can lead to *less* diversity.
3. Josh on November 2nd, 2005 (Comment) #
Andrew, you’re right to call out that I equated diversity and personalization. I wasn’t clear at all.
From what she said, it is clear that Memeorandum doesn’t suit Shelley’s personal desire for more diversity. I didn’t mean to imply that she said that it wasn’t personalized enough. I made that judgment myself…
And so I agree: personalization doesn’t equate with diversity. In fact, many people I know go so far as to shun diversity in the information sources they seek.
And yes, of course, things that we find personally valuable will very often include many things that others find valuable.
Thanks for keeping me straight.
4. Josh on November 2nd, 2005 (Comment) #
JS, I’m putting together a post on the personalization engines out there…thanks for your additions, and for the pointer to David W.
5. Brian on November 4th, 2005 (Comment) #
Josh … great post. This is a problem I experience and am currently researching. Would be interested in sharing any findings. I’ll likely blog on the topic soon with some research results. To-date, I haven’t seen anything out there that “gets” the referral and personalized blog parsing problem, or at least that does it well.
6. paolo on November 27th, 2005 (Comment) #
You might want to read “Trust Management for the Semantic Web” by Matthew Richardson, Rakesh Agrawal, and Pedro Domingos.
See http://moloko.itc.it/paoloblog/archives/2003/11/14/trust_management_for_the_semantic_web.html
Also there are people working on personalizing PageRank.
In this case, an interesting paper is
“An Analytical Comparison of Approaches to Personalizing PageRank”.
The problem with running a local trust metric is that you have to run it once from the point of view of every node. This means that Google shold run a different pagerank over the complete Web graph for every (!) user!!
My take is that in the (short?) future we will run local trust metrics for ourselves on our devices, a mobile or a laptop. Instead of a big corporation running a locat trust metric for everyone, we will have everyone running a lot trust metric for herself. I guess we will see how it ends …
7. Frank Urro on November 30th, 2005 (Comment) #
Very insightful,
I would like to discuss your thoughts more. My name is Frank Urro. I am a principal and cofounder of Vanquish Inc, and the inventor of Personal Value Control™.
Personal Value Control is applicable to every form of electronic interpersonal contact and is designed as an Open Architecture enhancement. It brings value to the consumer and adds value at the network provider level. It will eliminate all forms of unsolicited nuisance interruptions and will invite well targeted legitimate contact.
You can learn of my thought process here: Respect101 and of the company here Vanquish
I look forward to a detailed and engaged discussion.
Best Regards
/fju