August 16, 2005

More comments on...

a community based algorithm and the attendant issues...

Michael Frasse on Information authority and ranking:

    Hodder says, rightly, that the metric for assessing weight in the blogosphere should be open, not closed. “Bloggers should have input about the importance of one social gesture over another,� she writes. “One metric over another, and know what it is that is included because it will be used to describe them.�
    Kevin Burton has published an excellent response to Hodder that really advances the discussion. Burton has been working in this area since its inception (what, five years ago?) and has clearly been thinking about these issues. His main point seems to be that transparency in a blogosphere ranking and reputation system is problematic because of technological complexities and intellectual property barriers.

And Kevin may be right, but I think we should at least consciously talk about why something ought to be secret, and choose it, rather than just falling unconsciously into having a secret algorithm. I'd like to hear more about this, and discuss the reasons in the community, explicitly.

Barb Dybwad/The Social Software Weblog on Small is good, too: on quantifying connections:

    There’s all this fixation on getting links, getting traffic, getting on the lists. Why? That’s about monetization, that’s about fame/celebritism, that’s about gaming a structure that uses heuristics to measure what is supposedly value. And distantly, it’s about having your voice heard — I get that. But it gets blown way out of proportion. We forget that getting on the Technorati 100, in the end, is supposed to be about getting your voice heard by the people you want to hear it — instead it just becomes yet another pissing contest.
    But that’s an old model — it’s a vestige of the economics of scarcity, the analog world. The fact is, not everything I write needs to have a mass appeal. That’s precisely the beauty of the long tail.

It's a great post.. you should just read the whole thing. I only quibble with one statement:

    If you want someone to pay attention to you, if you want to make a new connection, just ask.

Well.. if it means asking for a link from another blogger when they wouldn't have otherwise given that attention, it to me sounds like gaming the system, to get a higher link count to rise up the power curve, rather than putting our efforts to measuring conversation within topic communities. If it means really conversing with someone, and getting their attention that way, then I think it's great.

She finishes with this, which I very much agree with:

    We don’t need to form new companies just so we can make new lists, in order to take action. We need to make new connections, and foster the connections that are honest, authentic and strong, as opposed to merely proliferating a bunch of opportunistic weak links.

I think making a representation of what we want to see measured and giving it to existing companies with existing databases in a community-company partnership is a great way to go.

Kevin Burton on Indexing and Blog Post Popularity:

    My thinking is that we should write up how an ideal metric would recommend posts. The algorithm implementation problem is another discussion as this can be implemented in many different ways. The devil is in the details I'm afraid.

Kevin actually did a comprehensive response to my original post, much of which I'd thought of, but it was great to see his take on the same issues. I think that what is important right now it that the community discuss what it sees as important to weight in an algorithm, or system, to participate in the process. Oh, and Kevin, sorry I didn't get any of your response into my first post about what people were saying, as you were a really early responder. My mistake!

PC4Media on Blog Social Network Analysis:

    Q: But, what happens when the scale gets bigger?
    A: (Yes, I am interviewing myself.) Of the 250 odd people that read my blog and the 300 or so that have linked to it, how many of them are actually people that have a meaningful relationship with me? Or even just a two way relationship with me? How many of them know someone I know? Or 5 people that I know? Do those people have meaningful relationships with each other?
    I have no freaking clue. But, damnit! I would love to know.

Jay Rosen, in the comments on Burningbird (where there is a fantastic discussion going on about this stuff):

    what a blogroll “says� in my understanding of how to blog well, although I must add that people can and do use their blogroll for any damn thing they like. My point is it makes perfect sense to link to the post that is already in your conversational field (blogroll.) The idea that there is something basically corrupt about it is a fantastic suggestion, yet perfectly normal for this discussion.
    But it is also worthwhile, I think, to reflect on what would seem to be “natural� linking practices, because they’re one way a blogger can get stuck in ruts, or become narrow minded without realizing it.

An excellent point.. because I think it's not just that we are not aware sometimes of our habitual behaviors in ourselves, but also, that any system has to take into account in some way the many 'kluges' many of us have developed, as early adopters as well as newbies, linking styles and 'technological workarounds' because blogging software just isn't that old. It's very early and we are still experimenting with lots of things. What appears from a blogger to be a social choice may be due to some funky piece of technology. It's yet another reason why I think having a public discussion where people talk about their habits and reasons for doing things is so useful, and will add to the solution.

Sarah Allen on The Value of Conversation

    There was a time when the web was a small community. When I started working on web software in 1995, a search on Yahoo! would list a small number of results. I don't know whether the whole web was indexed, but I think they kept up with the majority of it. I could view most of the web pages on a topic that interested me and make my own choices about what was most interesting and credible. When the Shockwave player was released, we could view every Shockwave movie that was created as it came on-line.
    As the web grew larger, it became more challenging to keep up. Thanks to the innovations of the search engines, we can do a quick search and find something relevant to our area of interest most of the time. However, with the creation of these search and ranking algorithms certain voices are omitted. I rarely look at more than a few pages of search results. Is it possible that the pages that would be most relevant to me might be later in the list? Or even worse, because of some artifact of gender use patterns could it be that pages more interesting to me will often be toward the end of the list?
    I like the idea of tracking conversations instead of purely inbound links. If we had an alternate index that tracked only links from posts (rather than blogrolls and other collections), then we might see where conversations are happening. Adina Levin describes this as a cloud presentation.
Posted by Mary Hodder at August 16, 2005 08:01 AM | TrackBack
Comments

First of all, thanks for being a great filter for all of this ongoing discussion. Secondly, I just wanted to clarify that the part of my post about asking for a new connection was meant to be taken precisely in your second interpretation -- spark a real conversation with a blogger who *would* have given you the attention (due to shared topic of interest, e.g.) had you been more visible in a Technorati/IceRocket/blah blah search. Certainly there are folks who do gain benefit from the "ask for an opportunistic link" strategy (and at the extreme end of the continuum we call them "spammers" ;)), but that's exactly what I'm arguing against. There's a tendency many of us seem to have to gravitate toward the short-term gain of getting a quick link and a temporary traffic spike and hoping to retain some of that traffic. Taking a different approach that builds relationships first is the less travelled path because it can come without the observable near-term gains in terms of direct linkage -- but over time, as those relationships are strengthened, the linkage will be a natural outcome of that connection as well as all the other new connections that stem from it as communities form. Plus, expending the effort to cultivate strong ties is, for many, a more enjoyable and authentic process than continually grasping at quick hits from weak connections on the list(s) -- though none of that emotional subtlety is going to be captured or rewarded by these algorithms (can we hope to capture any emotional subtlety in an algorithm? Topic for a separate post, methinks :)).

It seems to me that the tools (Technorati et al) are reinforcing the old "flood and retain" model of finding one's audience, and it would behoove us to build an alternative set of tools that would help us foster the "dig the well" method. As such, the idea of measuring conversation within topic communities is exciting, because it rewards effort and involvement over status and prestige. I worry, though, about churning through any algorithm and producing yet another list, and I'm not convinced that a list of any kind will satisfy what I'm most hungry to know about -- e.g. Peter Caputa's words above: "how many of them are actually people that have a meaningful relationship with me? Or even just a two way relationship with me? How many of them know someone I know? Or 5 people that I know? Do those people have meaningful relationships with each other?" I echo the sentiment that dammit, I'd love to know that kind of information. How do we build a framework deep enough to let that kind of information emerge? And how do we visualize the results of a community algorithm? I can only imagine it has to be something other than a list. If we're going to be modeling social networks, our view will always be impoverished by collapsing them into lists, and I'm really craving a way to visually present those interrelationships in a more intuitive way that I imagine will start to take on qualities of clouds (thanks, Adina!), or navigable node maps (a la Mitch Radcliffe, http://www.ratcliffeblog.com/archives/2005/08/cloudmakers_r_u.html). Can we generate these kinds of visualizations from such a community algorithm?

Posted by: barb dybwad at August 18, 2005 11:13 PM